NOVEMBER 19, 20, 21, 1979





Town and Country Hotel

San Diego, California





Russell Seely, General Chairman

John Horton, Technical Program Chairman

George Tremain, Exhibits and Publicity Chairman

Bob Klessig, Registration Chairman

Shelby Bass, Promotion Chairman

Hugh Pruss, General Arrangements & Equipment Chairman

Dick Lytle, Financial Chairman

Bob Klessig, ISA Program Representative


V. W. Hammond, Vice-President

R. D. Bently, Secretary

D. R. Andelin, Treasurer

L. W. Gardenhire, Director

H. F. Pruss, President

C. B. Weaver, Director

J. N. Birch, Director

T. J. Hoban, Director

S. Bass, Director

1979 ITC / USA Program Chairman

J. B. Horton

Program Chairman

John B. Horton received his BSEE from the George Washington University in 1956 and his MSEE from the University of Pennsylvania in 1964.

Mr. Horton is presently a Member of the Senior Technical Staff, Communication and

Antenna Laboratory, at TRW, Redondo Beach, California. He is on special assignment for new business in satellite communications. His past experience includes work on TDRSS, several military satellite systems, shuttle payload studies, missile guidance, and airborne and large ground based radars. He is the author/coauthor of 10 papers, seventeen lectures and was guest lecturer at University of Michigan and UCLA.

Mr. Horton is a Senior Member of IEEE and a member of AIAA. He has served on the

IEEE Microwave Group Administrative Committee since 1969 and was President of MTT in 1973.

Foreword to the 1979 Proceedings


General Chairman

The International Foundation for Telemetering together with the Instrument Society of

America are pleased to bring forth this Fifteenth volume of the PROCEEDINGS of the

International Telemetering Conference, a compilation of the papers presented at

ITC/USA/’79 in San Deigo, November 19, 20, 21, 1979. These papers, dealing with both the theoretical and practical-application aspects of the ever-expanding interests of the telecommunications community, will prove a valuable reference for future work and, together with the previous fourteen volumes, will document the tremendous strides made in telemetering and telecommunications during the 1960’s and 1970’s.

This year my objective as your General Chairman was to provide you a new, more informative and more entertaining conference — one you would remember and would motivate you to future support of ITC’s growth. My excellent and devoted board of chairmen and I accomplished this by first moving to beautiful San Diego with all of the social, recreational, entertaining and romantic activities and environments. Our next task was to promote attendance. We first had the colorful Gregg “Pappy” Boyington, the World

War II Marine fighter pilot ace and leader of the south pacific Blacksheep Squadron and popular TV series with the same name, as our luncheon speaker. Next we invited the presidents or chief executive officers of the major telemetry companies to attend, Major

General Conely, Director of Flight Test at Edwards Air Force Base as our guest banquet speaker, and range commanders of the flight test centers and ranges across the country as members of a blue ribbon panel on capabilities. Finally, we added tutorials, workshops, and a blue ribbon panel to our technical sessions. You all made it a successful and memorable event. I thank you.

As General Chairman of ITC/USA/’79, I express my gratitude to Mr. John Horton, the

ITC/USA/’79 Technical Program Chairman, and to all session chairmen and authors

whose collective work and dedication have made this volume possible. Gratitude is also extended to the members of the working committees and all of you who attended

ITC/USA/’79. Without your continued interest and support, these conferences would not be possible. It has been a rare opportunity and a distinct pleasure to serve you.

Russell Seely, General Chairman ITC/USA/’79

Russell Seely is Manager of Flight Test Data Systems for Flight Systems, Incorporated,

Newport Beach, California and directs the design, development, operation, maintenance and promotion of all data acquisition, ground data processing, computer and software systems and related laboratories, as a business area. He received his B.S. degree in

Electrical Engineering in 1950, and has since received his Masters in Systems Engineering and Doctorate in Business Administration.

Mr. Seely, in the mid 1950’s, designed and developed one of this country’s first flying hybrid analog and PCM data acquisition system and ground data processing system. He directed a task force team which established a development and test plan in 1962 for the

Apollo program. Later, as program manager, he developed hardware for an unmanned spacecraft to repair malfunctioned spacecraft in earth orbit. At Flight Systems he directed the system development and test of the QF-86E drone aircraft. He holds a number of patents on flight test electronic equipment and digital and analog computers. He has appeared on television and presented numerous technical and management papers on telemetry systems, computer/software systems, and major programs including Remote

Maneuvering Units and Apollo. He has taught telemetry, advanced communications, computers and management systems at various colleges over the past 25 years. Mr. Seely directed the system development and test of the QF-86E drones at Flight Systems.





Blue Ribbon Panel on Test Facility and Range Capabilities in the United


Organizer, Dr. R. Seely, Flight Systems, Inc.

II. Panel on High Density Digital Recording in the 1980’s

Moderator, Glen H. Schulze, Consultant, Tracor, Inc.


Computers, Software and Managers

Chairman, Dr. R. Seely, Flight Systems, Inc.



Spread Spectrum ’79, the Big Picture

Lecturer, R. Dixon, Spectrack Systems, Inc.

II. Antenna Design for Satellite Communications Systems

Lecturer, Dr. J. W. Duncan, TRW Systems

III. Application of Fiber Optics to Military Systems

Lecturer, Dr. M. R. Barnoski, Hughes Research Laboratories



Advanced Communication Satellite Control Concepts, Part 1

Session Chairman

— Paul E. Brandinger, COMSAT Corporation


TDMA Satellite On-Board Switching Center Electronics

J. Kasser, COMSAT Laboratories


Spacecraft Minicomputer for Control of a Communications Payload

Russell R. Rhodes and Marilyn D. Semprucci, MIT, Lincoln Laboratory


Intersatellite Link Tracking Antenna Pointing Requirements

D. N. Srinivas, COMSAT Laboratories


On-Board Satellite Communication Configuration Control via

Communication Channel

J. MacPhee and D. Coomber, MIT, Lincoln Latoratory (Abstract Only)


Novel S and C Bant TT and C Antennas for Satellite Applications

J. Torres, Instituto Nacional de Tecnica Aerospacial, Torrejon de Ardoz,

Madrid, Spain and D. C. Patel, European Space Agency, Noordwijk, The



A Compatible STS/PAM D/RCA STATCOM Telemetry and Command


R. Hoedemaker and C. Staloff, RCA Astro-Electronics


Industrial Applications

Session Chairman

— R. Fuller, COMTEL Corporation


Reconstructing Pulse Code Modulation Telemetry Data With Dropouts

M. L. Hull, University of California, Davis; C. D. Mote, Jr. and L. W.

Lamoreau, University of California, Berkeley


Hardware Compressor Reduces Computer Loading

O. J. Strock, EMR-Telemetry


Time Response Simulation of the Guidance and Control System of an

Automatically Steered Wire-Following Vehicle

A. K. Mukhopadhyay and B. M. Dobrotin, Jet Propulsion Laboratory


Development of a Seven Channel Telemetry Transmitter

R.L. Seeley, Naval Ocean Systems Center; F. M. Long and R. W. Weeks,

University of Wyoming; J. D. Pauley, University of Colorado Medical Center


High Density Digital and Magnetic Recording

Session Chairman

— Avner Levy, Advanced Recording Technology, Inc.


Correlation of Tape Dropouts with Data Quality

K. O. Schoeck, Space and Missile Test Center, Vandenberg AFB, and G. M.

Kobylecky, Federal Electric Corporation, Vandenberg AFB


Error Control Strategies for High Rate High Density Recording

R. C. Montgomery, Sangamo-Weston, Inc. (Abstract Only)


Super High Bit Rate Recording

Tracy G. Wood, Ampex Corporation (Abstract Only)


Landsat D High Density Tape Recorders

J. H. Montgomery, Martin Marietta Corporation (Abstract Only)


Advanced Communication Satellite Control Concepts, Part 2

Session Chairman

— Paul E. Brandinger, COMSAT Corporation


An Overview of TDRSS Ground Station

P. C. Morran and J. E. Bebb, TRW Systems (Abstract Only)


MMS Command and Data Handling

R. L. Kelley and H. A. Raymond, Fairchild Space and Electronics Company

(Abstract Only)


Decentralized Control for Large Communication Satellites by Model

Error Sensitivity Suppression

J. R. Sesak and R. M. Bowman, General Dynamics Convair Division


DMSP Block 5D-1 Computer Controlled Spacecraft

Lt. Col. S. M. McElroy, Space and Missile Systems Organization (SAMSO),

L. Gomberg and R. TeBeest, RCA Astro-Electronics (Abstract Only)


Control Aspects of Multibearn or Multielement Spacecraft Antennas

P. Foldes, General Electric Company


Control of Large Communication Satellites

R. Gran, M. Proise and Alex Zislin, Grumman Aerospace Corporation


Future Modular Data Handling Concepts for Large Space Platforms

G. P. Thompson, ESA/ESTEC, The Netherlands (Abstract Only)


Laser Communications

Session Chairman

— J. D. Barry, Hughes Aircraft Company


Laser Sources and Space Optical Communications

J. D. Barry, Hughes Aircraft Company (Abstract Only)


Principles of Direct Detection Optical Communication

L. F. Eastwood, Jr., J. A. Maynard, and S. I. Green, McDonnell Douglas

Corporation (Abstract Only)


System Design for Nd:YAG Laser Communications

J. D. Wolf, J. A. Pautler, R. Z. Olshaw, and L. F. Eastwood, Jr., McDonnell

Douglas Corporation (Abstract Only)


Fundamentals of Heterodyne Detection in Laser Communications

F. E. Goodwin, Hughes Aircraft Company (Abstract Only)


The Design Of CO


Laser Communication Systems

A. J. Einhorn, Hughes Aircraft Company (Abstract Only)


Satellite Power System

Session Chairman

— R. Dickinson, Jet Propulsion Laboratory


Solar Power Satellites — The Present and the Future

G. D. Arndt, NSAS-Johnson Space Center


Managing the SPS Antenna Power Beam Performance

R. M. Dickinson, Jet Propulsion Laboratory (Abstract Only)


Active Retrodirective Array for Microwave Power Transmission

R. C. Chernoff, Jet Propulsion Laboratory (Abstract Only)


Time and Frequency Transfer by the Master-Slave Returnable Timing

System Technique — Application to Solar Power Transmission

W. C. Lindsey and A. V. Kantak, LinCom Corporation


Computers, Software and Managers

Session Chairman

— L. Wilson, Flight Systems, Inc.


Estimating Time to Develop Software

Speaker to be announced (Abstract Only)


Management of Software Development

H. Robert Downs, Science Applications, Inc.


The Woe’s of the Computer System Manager

Speaker to be announced (Abstract Only)


The Problem --- Is It Hardware or Is It Software

Roland E. Olson, EMR Data Systems (Abstract Only)


Satellite Communications and Telemetry Systems

Session Chairman

— G. M. Hatch, TRW Systems


Telecommunications for the International Solar Polar Mission

R. J. Chan, TRW Systems


Multigigibit Satellite On-Board Signal Processing

W. M. Holmes, Jr., TRW Systems


The LEASAT Communications Satellite

G. L. Dutcher and J. G. Lankford, Hughes Aircraft Company


Satellite Control System

Lt. Col. J. Baker and Lt. S. Vest, Air Force Systems Command and A. S.

Gilcrest and T. M. Rodriguez, The Aerospace Corporation


Space Shuttle Communication and Telemetry Systems, Part 1

Session Chairman

— L. M. Carrier, Rockwell International


Space Shuttle Technology Flight Instrumentation

John Dunstan, Rockwell International


Space Shuttle Communications and Telemetry - An Update

J. C. Hoagland, Rockwell International


Space Shuttle Payloads and Data—Handling Accommodations

W. E. Teasdale, NASA, Johnson Space Center and Kwei Tu, Lockheed

Electronics Company, Inc.


The Command and Data Management System of Spacelab

G. R. Bolton, European Space Agency (ESA) ESTEC, Noordwijk, The



Synchronization Techniques

Session Chairman

— J. Holmes, Holmes & Associates


Data Pattern Senstivity in Tracking Performance of an AC Coupled

Costas Loop with Hard-Limited In-Phase Channel

Y. H. Park, Jet Propulsion Laboratory


Performance Analysis of Noncoherent AGC for Signal Presence

Detection and Autotrack Signal Extraction

D. D. Carpenter, TRW Systems (Abstract Only)


Landsat D Wideband Communication Subsystem Design

C. C. Chen, TRW Systems (Paper not received in time for publication)


An Optimum Asymmetric PN Code Search Strategy

J. K. Holmes, Holmes & Associates and K. T. Woo, Aerospace Corporation


Panel Discussion on Future Trends in Spread Spectrum Systems

Moderator - J. Holmes, Holmes & Associates


Communications Technology and Techniques

Session Chairman

— F. Gerardi, Aerospace Corporation


What the System Link Budget Tells the System Engineer or How I

Learned to Count in Decibels

B. Sklar, Aerospace Corporation


Microprocessor-Based Analog Voice Scrambling Techniques

S. Udalov, Axiomatix


Convolutional Error Detection on an Additive White Gaussian Noise


M. A. King, Aerospace Corporation


U. S. Domestic Communication Satellite Systems

D. H. Martin, Aerospace Corporation


Random Coding Bounds for Noncoherent mFSK Multiple-Access


J. K. Omura, UCLA


Generalized Feedback Decoding of Convolutional Codes

Wai-Hung Ng, Aerospace Corporation


Space Shuttle Communication and Telemetry Systems, Part 2

Session Chairman

— L. M. Carrier, Rockwell International


Space Shuttle Payloads Support Capability

F. Torres, Rockwell International


Inertial Upper Stage/Shuttle Orbiter Communications

G. K. Huth and S. Udalov, Axiomatix


NASA Standard Experiment Command and Data System for

Shuttle/Spacelab Payloads

L. H. Kasulka and D. D. Wilkinson, McDonnell Douglas Astronautics



Standards and Applications

Session Chairman

— F. Hartzler, Pacific Missile Test Center


Optimum Digital Data Storage on Magnetic Tape

W. R. Hedeman, Jr., Annapolis, Md. and E. L. Law, Pacific Missile Test

Center (Abstract Only)


Serial PCM Recording Standard

Telemetry Group, Range Commanders Council (Presenter: E. L. Law)

(Abstract Only)


A Proposed Time Code Standard for Telemetry and Space Applications

A. R. Chi, NASA, Goddard Space Flight Center (Abstract Only)


Military Telemetry Systems

Session Chairman

— G. W. Davis, Naval Weapons Center


A Microcomputer Interface for Transfer of Data Between Multiple

Computer Systems

B. L. Smith, Vandenberg Air Force Base, Ca. (Abstract Only)


Navy Shipboard Weapon Information Telemetry System

E. A. Dahl and L. Bates, Naval Ship Weapon Systems Engineering Station


Dual Beam Single Axis Tracking Antenna for Tracking Telemetry

Instrumented Airborne Vehicles

A. Sullivan, Electro Magnetic Processes, Inc.


Millimeter-Wave Systems Technology

Session Chairman

— H. J. Kuno, Hughes Aircraft Company


Millimeter-Wave Technology Overview

J. C. Wiltse, Engineering Experiment Station, Georgia Tech


Tunable Millimeter-Wave Communications

S. Becker, AIL Division of Eaton Corporation


Millimeter-Wave Solid State Transmitter Components

C. Sun, Hughes Aircraft Company


Millimeter-Wave Receiver Components

F. Bernues and P. Pusateri, Hughes Aircraft Company


Future Performance Limitations for Ground and Spaceborne Millimeter

Wave Receiver Systems

A. G. Cardiasmenos, TRF Division, Alpha Industries


Satellite and LOS Communications Ground Systems

Session Chairman

— J. F. Niebla, Infotec Development, Inc.


The Application of Frequency Offset Advantage (FOA) in Frequency


S. Raghavan and J. Armes, Spectrum Analysis & Frequency Engineering

(SAFE) (Abstract Only)


A Review of the Public Broadcast Service TV Distribution System and

Plans for the National Public Radio System

R. S. Kellow, Rockwell International (Abstract Only)


A High Performance 8 GHZ, 8 PSK Digital Radio

G. G. Russo and P. R. Hartmann, Rockwell International

Appendix “A”

Eighteen Annual Report of the Telemetering Standards Coordination Committee

Michael Pizzuti, TSCC Chairman 1978-1980





Dr. Russell Seely

Flight Systems, Inc.

Newport Beach, California

The National Range Commanders Council is made up of range commanders from each of the major military ranges and test facilities across the United States. It includes Edwards

Air Force Base, Patuxent River Naval Test Center, Eglin Air Force Base, White Sands

Missile Range, Navy Point Mugu, Navy China Lake to mention only a few. These ranges and test facilities have the most modern and sophisticated telemetry equipment that can be imagined to support real time and off line analysis operations of our airborne, space, and missile systems programs. Additionally and because of this complex capability, the council also establishes the well known IRIG standards to be used in discussing, procuring, and operating these telemetry systems.

The purpose of this Blue Ribbon Panel is to inform the telemetry professional community of the ranges capabilities and the on going efforts and status of the IRIG committee.




Moderator, Glen H. Schulze

Consultant, Tracor Inc.

Austin, Texas

Although High Density Digital Recording (HDDR) arrived on the instrumentation scene in the early 1970’s, it is as yet far from a mature and widely accepted tool in the telemetry community. Most housekeeping telemetry is still recorded via analog recorders. HDDR techniques are currently reserved almost exclusively for high data rate imagery sensor recording applications and the like. The data from these sensors are ultimately processed digitally via computers. The maturing of HDDR is expected to occur in the 1980’s and the panel will discuss the main issues pertaining to this maturing.

Major topics to be discussed by the panel are the following:

• HDDR codes

• In-track packing densities

• Areal packing densities

• Head/tape formats

• Bit error rates

• Error detection, concealment, and correction

• HDDR standardization for crossplay compatibility

Status reports of HDDR development projections by the leading manufacturers will be solicited. Audience participation will be invited to promote the maximum exchange of information between the potential future users of HDDR systems and the suppliers.

The goal of the panel is to identify and develop the expected trends in HDDR technology during the 1980’s. To achieve this goal the panel will be comprised of invited technical spokesmen from the leading suppliers of HDDR systems, including Ampex, Bell &

Howell, Odetics, Martin Marietta, RCA, and Sangamo.




Chairman, Dr. Russell B. Seely

Flight Systems, Inc.

4000 Westerly Place

Newport Beach, California


The computer system manager is faced with two large problems today — how to obtain reliable manpower and time estimates for developing software, and how to manage computer hardware and software people who point their fingers at each other and say “my system is working fine, the problem is yours.” As a result of these two problems, the cost of software development is growing at alarming rates and the computer manager remains relatively ill equipped with evaluation and decision making criteria similar to that which has been developed over the years for hardware developers.

This workshop discusses four key factors which are designed to improve the computer system managers decision and software development estimating capabilities. The four key factors are:

1. “Estimating Time to Develop Software”

2. “Management of Software Development”

3. “The Woe’s of the Computer System Manager”

4. “The Problem — Is It Hardware or Is It Software”


R. C. Dixon


Spectrack Systems Inc.

5300 Orange Ave. #113

P. O. Box 1164

Cypress, Calif. 90630


Spread spectrum techniques are finding their way into many types of communication, navigation, and identification systems today. In fact, these techniques are often critical to the successful operation of these systems in their intended application. Systems such as

Global Positioning System (GPS), Joint Tactical Information Distribution System (JTIDS),

Single Channel Ground-Airborne Radio System (SINCGARS), SEEK TALK, Position

Location Reporting System (PLRS), and others are examples of such systems. Even commercial communications users are beginning to consider the possible advantages of the spread spectrum methods for their use.

This tutorial session discusses the capabilities of spread spectrum techniques in the light of today’s state of the art, and considers possibilities and limitations to their future use, in commercial as well as military or space applications. It is intended for both technical and non-technical personnel.



Dr. J. W. Duncan

TRW Defense and Space Systems Group

One Space Park


The past decade has seen an enormous increase in the capability and complexity of satellite antenna systems. This lecture will present a tutorial overview of technology trends in satellite antennas. A brief introduction to the basic properties of antennas such as gain, beamwidth, polarization, and axial ratio will be followed with examples of typical satellite antennas. The antenna complexity brought on by requirements for multiple beams, frequency reuse, regional coverage beams, and jamming protection will be illustrated with examples from recent satellite programs.




M. K. Barnoski

Hughes Research Laboratories

3011 Malibu Canyon Road

Malibu, California 90265


Interest in optical data transmission, both in the telecommunications industry and the DoD is rapidly increasing. While the potential advantages of optical communication have been recognized for a long time, recent technological breakthroughs in the fabrication of lowloss optical fiber waveguides have made a variety of commercial and military applications both feasible and attractive. The development of low-loss fibers in turn provided the incentive to accelerate development of other elements of optical data links, e.g., light sources, modulators, couplers, detectors, and receivers. At the same time, increased concern about the effects of EMP and EMI on critical data links, coupled with requirements for higher bandwidth, a high degree of intercept security, dielectric isolation, and a solution to crosstalk and ground loop problems, has spurred interest in fiber optics for aircraft, naval ship, and ground base communications systems along with application to undersea systems and guided weapons systems. The problems encountered and the requirements on the fiber cables and the other system components vary, depending upon the particular system application. The demands on fiber technology created by military applications range from those encountered by the commercial telecommunications industry to those which are beyond the capability of today’s technology. It is the intent of this paper to briefly review the application of optical fiber technology to military systems.



Joe Kasser

COMSAT Laboratories

Clarksburg, Maryland 20734


The introduction of TDMA into satellite communications links increases the complexity of satellite onboard communications path routing circuitry. This paper discusses two different design approaches to the control electronics of an onboard TDMA satellite switching center. One approach uses custom large-scale integrated (LSI) circuitry; the other is microprocessor based. The two designs are compared, and an optimal hardware configuration is developed.


The main constraints affecting the efficiency of a communications satellite system are power, bandwidth, and utilization of the available bandwidth. In future generations of

INTELSAT satellites, power and bandwidth efficiencies are expected to be increased by applying a technique known as frequency reuse by multiple narrow-beam, high-gain antennas termed spot beams. The number of spot beams can be increased by antenna pattern isolation, dual polarization, and the use of higher frequency bands. To utilize the bandwidth efficiently, the interconnection of various transmit and receive beams must be arranged to reflect the traffic requirements.

When TDMA is introduced, the opportunity arises to further increase the utilization efficiency by cyclically interconnecting, in a rapid sequence, the TDMA signals present in the various beams according to their destination. Such a technique is known as satelliteswitched/time-division multiple access (SS-TDMA) (1), (2), and (3). Hardware must be produced for the space segment to interconnect the signals present in the different frequency bands, polarizations, and coverage zones according to the traffic topology.

The space segment of SS-TDMA consists of a reliable fully integrated satellite switching center (SSC), which contains a microwave switch matrix (MSM) and a programmable distribution control unit (DCU), as shown in Figure 1.

Flexibility of interconnection must be provided to optimize the overall performance of future multiple-beam configurations. Therefore, the SSC must have the ability to connect any number of inputs to any number of outputs at any time. Another essential requirement for the SSC is functional redundancy to allow the maintenance of full interconnection between the different inputs and outputs over the operational lifetime of the satellite (7-10 years), despite a number of active and passive failures anywhere within the SSC.

Development of the space segment of SS-TDMA to date has considered the DCU and

MSM as separate entities. DCUs and MSMs were developed separately under different contracts and were then tested at COMSAT Laboratories.


In MSM technology, 8 x 8 matrices have been built. PIN diodes constituted the main active element in the switches. One MSM consisted of four 8 x 8 submatrices, thirty-two

3-dB interdigitated or Lange hybrids, and sixteen single-pole double-throw (SPDT) switches. This produced an equivalent 16 x 16 MSM assembly with 256 interconnect possibilities. Each 8 x 8 MSM subassembly consisted of eight input and eight output distribution networks connected in a cross-bar configuration via push-pull connectors.

Coaxial PIN diode switches, which were energized or driven to provide the desired, fast interconnection patterns, are hermetically sealed and integrated within the push-pull connectors. In general, redundancy in the MSM can be achieved by using redundancy networks in the inputs and/or outputs of the matrix.


The DCU can be considered a complex digital circuit driving an MSM. Parts of the DCU operate at slow switching speeds (TT&C interface), while others operate at fast switching speeds (MSM cross-point switching drivers). Implementation techniques for the DCU identified at this time include the following: a. custom LSI circuitry; b. a single chip microprocessor, LSI memories, and custom LSI circuitry; c. a single chip microprocessor, shift register, and standard LSI circuits (including LSI memories); and d. a bipolar microprocessor based design.

The DCU can be subdivided into two parts: a telemetry and control interface (TCI), which operates at slow speeds; and MSM switching control electronics (SCE), which operate at high speeds (Figure 2). This functional division can also be a physical division, which has been incorporated into all DCUs to date. An LSI version built by British Aerospace utilized two custom LSI circuits, one performing the TCI function and one the SCE function. In fact, the DCU contains one TCI and a number of SCEs. A DCU currently under construction at COMSAT Laboratories uses a microprocessor based module to perform the TCI function and custom LSI circuits for the SCEs. Table I gives performance characteristics for a typical DCU.

Table I. DCU Performance Characteristics


Frame Time Period

Frame Unit

Synchronization Bursts/Frame Period

Signal Bursts/Frame Period

MSM Interface

Logic 1

Logic 0

Rise and Fall Time

Delay Variation

Load Impedance

TT&C Interface

Logic Levels

Clock Rate

750 µs

6 µs



+4 ± 1 V at 5 mA max

0 V ± 0.2 V at 100 µA max

50 ns max

75 ns max

1 kS in parallel with 50 pF


1 kHz


Onboard equipment of the complexity of the DCU is possible only because of recent advances in the state of the art of LSI circuitry. Historically, as the state of the art advanced, discrete circuitry became integrated. Clusters of integrated circuits (ICs) on printed circuit boards were later replaced by LSI circuits. Clusters of LSI circuits are currently being replaced by custom complex LSI circuits. In effect, today’s printed circuit board is tomorrow’s custom LSI circuit.

LSI circuits are produced in two forms: the custom and the standard. Each may be analog, digital, or a mixture of both; they are differentiated only by the numbers produced. A

standard LSI circuit (e.g., a microprocessor) may be produced in quantities of millions; a custom LSI circuit (such as a TCI or SCE circuit) may be produced in quantities of tens.


The design, fabrication, and testing of circuits are time consuming and expensive (4). The reliability, mass, and power aspects of an LSI based design make it very attractive for spacecraft onboard electronics; however, the long lead time and cost aspects tend to lengthen the construction time of the device and inflate its price (4).

The British Aerospace DCU performed its task almost as specified. Any anomalies in the performance did not affect the MSM control functions and could easily be corrected in a remake of the ICs which would, however, be expensive and time consuming. The project exceeded its initial time allowance by more than a year because of the long lead times involved in the manufacture of the ICs. It should be emphasized that a chip size of 0.3 in. x

0.285 in. approaches the limits of current manufacturing technology. In addition, the circuit design proved to be more difficult than expected because of the extreme values of circuit parameters caused by the large dimensions involved (such as stray capacitance effects).

The COMSAT Laboratories DCU uses an LSI circuit to perform the SCE function and a microprocessor based module to perform the TCI task. The fabrication of SCE circuits has taken much longer than expected because of the problems associated with designing a TTL breadboard of a metal oxide semiconductor (MOS) integrated circuit. The complexity of the circuitry and the highspeed requirements have resulted in an IC that also challenges the state of the art. The microprocessor module uses complementary metal oxide semiconductor (CMOS) technology and incorporates an 1802 microprocessor. Thus, the

COMSAT Laboratories DCU differs from that of British Aerospace by virtue of the microprocessor based TCI unit.

At the time of writing, the testing of the breadboard SCE is not complete. Several anomalies remain in its operation, such as the uncertainty of whether the SCE is shifting out the contents of its memory (up to 4000 bits) with a timing error so that the data are skewed by one bit with respect to the clock. The TCI unit is thus designed to be flexible enough so that any further undiscovered anomalies built into the LSI version of the SCE can be compensated for in the TCI software.


Any technology employed in the manufacture of LSI circuits intended for use onboard communications satellites must be radiation hard. This is because ambient radiation levels at geosynchronous altitudes tend to destroy most MOS LSI circuits well within the 7-year

lifetime design goal (5). As the state of the art advances, the capability for increasing IC complexity will improve, and designers will be urged to integrate as much circuitry as possible onto the slice of silicon. If custom LSI is used, the cost of low-volume spacecraft onboard hardware will also increase. Therefore, a compromise between standard and custom LSI may be used to implement a function. For example, a DCU contains slowspeed circuits which perform tasks easily accomplished by a microprocessor. It contains a large redundant memory and circuits which control the flow of data between the memory and the outside world.

If a standard LSI circuit is used to implement the memory and microprocessor circuitry, a small custom section could be designed to perform the control function. LSI design techniques for nonrandom circuitry (e.g., memory circuits) are well established. Thus, instead of building the unit with a number of packages, a fully integrated design could be implemented on a single silicon substrate using a mixture of standard and custom circuitry layout techniques. Today, a thick film technology would be required; in the future thin film could be used.


The advance in the state of the art has made low-cost LSI circuitry possible for volume applications. Traditional spacecraft designs are using low-volume, high-cost custom LSI or microprocessor based devices. Future spacecraft hardware could be built on large silicon wafers incorporating both high-volume (microprocessor) circuits and low-volume (special purpose) circuits using thin film techniques that are equivalent to today’s thick film techniques.


This paper is based upon work performed at COMSAT Laboratories under the sponsorship of the Communications Satellite Corporation (COMSAT) and the

International Telecommunications Satellite Organization (INTELSAT).


1. Muratani, T., “Satellite-Switched TDMA,” EASCON ’74 Record, pp. 189-196.

2. Dill, G. et al., “Application of SS-TDMA to INTELSAT Networks,” 3rd Digital

Satellite Communications Conference, Kyoto, Japan, 1975, pp. 29-37.

3. Dill, G. et al., “Simulated SS-TDMA Test Results,” 4th Digital Satellite

Communications Conference, Montreal, Canada, 1978, pp. 178-179.

4. Semiconductor Test Symposium Digest, Cherry Hill, New Jersey 1978, IEEE

Computer Society Publications.

5. Revesz, A. G., “Integrated Circuits in Communications Satellites,” COMSAT

Technical Review

, Vol. 9, No. 1, Spring 1979, pp. 233-242.

Figure 1. SS-TDMA Space Segment Block Diagram

Figure 2. DCU Internal Positioning




Russell R. Rhodes and Marilyn D. Semprucci

Massachusetts Institute of Technology, Lincoln Laboratory

Lexington, Massachusetts 02173


The paper describes the design, use and testing of a spacecraft minicomputer. The minicomputer was designed as a controller for the communications payload of a

Communications Satellite with about 100 communications channels. Each of the channels has very flexible control including variable data rate, Doppler offset, and flexible routing.


Conventional communications satellites weigh on the order of a ton and have on the order of 1 killowatt of prime power available. Most communications satellites are transponders.

They receive and filter uplink signals, translate the frequency, power amplify and retransmit the signal on the downlink. This method of communication has potentially serious interference problems. Any strong signal in the communications band can saturate the downlink transmitter and interferes with communications. The interference can be from a too strong user or jamming intentional or unintentional.

1 The performance of a communications satellite can be greatly improved by using approximately 10% of the power and weight available (about 100 lbs and 100 watts) for signal processing. The 10% reduction in power available for downlink transmission will reduce transmitter power by less than 1 dB. With about 100 lbs and 100 watts several of the following signal processing techniques can be used: (1) frequency hopping (2) demodulation to digital bits

(3) decoding and deinterleaving (4) encoding and interleaving (5) data buffering and formatting (6) remodulation (7) antenna nulling and beam steering.

1 With the above techniques the uplink can be frequency division multiplexed and the downlink can be time division multiplexed. The bit rate can be adjusted to accommodate small uplink

* This work was sponsored by the Department of the Air Force.

The views and conclusions contained in this document are those of the contractor and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the

United States Government.

transmitters and the time per bit can be adjusted on the downlink to accommodate disadvantaged downlink receivers. Noise and interference are stripped off in demodulation and are not retransmitted on the downlink. Data formatting and buffering will permit otherwise incompatible terminals to communicate.

A satellite payload with this much complexity and flexibility requires formidable control.

The MARC (Microprogrammed Adaptive-Routing Controller) a compact, highperformance 16-bit minicomputer was designed to control the payload of a complex signalprocessing communications satellite in synchronous orbit.

The environmental requirements for a component in a synchronous satellite are actually rather mild, with two exceptions: it must survive the shock and vibration of launch, plus the solar radiation. Shielding is of a very limited effectiveness. Most bipolar devices such as TTL can easily meet this requirement; however, it eliminates most one-chip microprocessors. The 8080, for example, would survive about a week at synchronous orbit.

Work is proceeding to radiation-harden CMOS and I 2 L microprocessors. One of these technologies may be a good choice in the future. The MARC is realized with low-power

Schottky TTL technology, of known and proven radiation hardness.

Weight and power are, of course, critical parameters in a synchronous satellite. For digital hardware, power is more important. The weight of digital IC’s is very small, but the weight required for generation and conversion of prime power and for extra metal to remove the heat add up to 1 to 1.5 pounds per watt. The MARC CPU (without I/O memory) requires approximate y 10 watts. Potentially flyable one-chip microprocessors require anywhere from a few hundred milliwatts (CMOS) to about 2 watts (I 2 L). The

CMOS chips are far less capable than the MARC; the I


L chips are somewhat less capable. Presently available radiation-hard memory (LS-TTL) requires about 3.5 watts per

1K words of 16 bits. Clearly, the power required for memory could easily amount to tens of watts even in a moderate sized system, swamping the power disadvantage of the

MARC. Even when lower-power memory becomes available, the MARC might still be the better choice over potentially flyble one-chip CPU’s where its superior speed and flexibility were advantageous.

The MARC packaged for flight in a typical configuration including the CPU I/O and 6K

16 bit words of memory would require about 35 watts and weigh about 20 lbs.

The MARC has a 305 nanosecond microcycle time. This rather slow time allows for derating of component speeds due to radiation. The minimum programmed instruction, like

Register to Register adds or logical functions, requires 4 microcycles (1.22 microseconds).

Memory reference instruction requires an extra microcycle to load the memory address register and extra microcycles for computing the memory address if the selected addressing mode requires computation.

Typical MARC Instruction Execution Times

Types Microcycles *

Register to Register Arithmetic and Logical

Load Immediate Word 4-6


Memory ø Register Arithmetic and Logical 10-14


Multiply 26-42

Divide 59-75

* Most testing at 305 nsec per microcycle. Maximum speed ~ 200 nsec. With new premium chips, probably could be 100-150 nsec.

The MARC Function

The MARC was designed to control the advanced technology signal-processing communications satellite payload, shown conceptually in Fig. 1. This system is designed to serve approximately 100 simultaneous communications users, in several uplink frequency bands, with a maximum of flexibility and efficiency. There are several receiver front ends, some with frequency-hopped local oscillators, several digital group demodulators capable of handling up to 16 simultaneous FDH signals with any of several digital modulation forms 2 , and several Communications Output Processors (COP’s). The latter device is a combination switch, data buffer, and TDM signal formatter which generates a downlink data stream. Selected channels can use error-correcting decoders and encoders. Any uplink channel can be connected to any downlink channel or time slot, with any of several modulation types. Completely imcompatible ground terminals can be interconnected. A communications satellite with this much flexibility clearly presents unprecendented control problems; these the MARC was designed to solve.


The MARC interprets commands from the ground controller, which sets policy/or makes specific connections. The MARC must interpret these command and change the necessary switches and data tables. The MARC also receives requests from users of the system, and assigned communications resources consistent with the priorities set by the ground controller. The MARC must set up the center frequencies, timing, data rates, and modulation formats for up to 16 uplink channels in each Group Demodulator. It must control the data rates, time division multiplex format, and data connections in the

Communications Output Processor. It must also set the bus switches to interconnect modules.

MARC System Architecture

The major sections of the MARC are shown in Fig. 2. The CPU is four AMD 2901’s. The

MARC is controlled by the microprogrammed control section. MARC program instructions reside in main memory. They are fetched by a microcode routine. The instruction op code specifies a jump to the appropriate microcode routine to implement the instruction. Microcode resides in ROM inside the “Microprogrammed control” section.

The development interface connects the MARC to a NOVA 3/12 host minicomputer in which resides all real-time support and development system software. The NOVA minicomputer can be connected to an IBM 370 via a time share line. A special microcode assembler and MARC code assembler were written for the IBM 370. The assembled files are transferred via the time share line to a disk file on the NOVA development system.

The development system can load both main and microcode MARC memory. It can also examine and modify any of the CPU registers.


Microcode memory is loaded by controlling the microcode address bus from the development system and passing data via the 16 bit Data Bus. Registers and main memory are loaded or modified by loading and starting a short microcode program which transfers data. The development system also has built in traps which can halt the MARC on any access of any specific memory location. In addition, it has the ability to trace microcode steps and keep track of microcode address and bus contents. It was discovered, however, during the devleopment that the use of a logic analyzer eliminated the need for both traps and trace. It was relatively easy to troubleshoot the MARC hardware by simply setting the logic analyzer to trigger on specific microcode addresses and Data Bus contents and then starting the program execution at specific location with registers and memory in the desired state. The minicomputer development system was also used to simulate external hardware during testing of the MARC and MARC software.

Detailed Hardware Description

The MARC is microcode controlled. A 40 bit microcode word out of the microcode ROM

(RAM during development) is divided into fields as shown in Fig. 3. The bits in each field, with a minimal amount of gating, control the various functions of the MARC, as indicated.

The MARC contains four AMD 2901’s in its central processor unit (see Fig. 4). The AMD

2901 has 16 general purpose registers and one Q register.

The microcode control section of the MARC contains two AMD 2909 program control chips. They provide 8 bits of microcode address. Two additional bits are also provided for page control. A multiplexor in the AMD 2909 allows for four selections for the next microcode address. Microprogram control is accomplished by specifying a Jump, Call,

Return or Fetch while selecting one of the FLAGS. If the FLAG is true, the Jump, Call,

Return or Fetch is executed. If not, the program continues to the next ROM address.



Vectored interrupts are accomplished using a multiplexor on the least significant four bits of the D input to the AMD 2909. The low order 4 bits are multiplexed between the normal

D inputs and the input from a priority encoder. If the upper order 6 bits from the microcode ROM are all one, the multiplexor is shifted to the priority encoder, and 1 of 16 addresses is selected depending upon which interrupt is present.

The lines labeled MDS on the left hand side of the diagram stand for Minicomputer

Development System. These lines allow the minicomputer development system to control the ROM address and pass data in and out of the Data Bus. During development, the microcode memory is a RAM rather than a ROM. The MDS lines are used for loading this microcode RAM, and also for loading the CPU registers and the main memory.

The program counter (as distinct from the microprogram) is kept in general purpose register 2 in the AMD 2901’s. During a normal fetch cycle, the program counter contents are passed to the memory address register; at the same time the program counter contents are incremented and placed back in register 2. The contents of the specified memory location are then transferred to register 1 in the AMD 2901 and to the instruction register.

The upper order 8 bits of the instruction register go to the AMD 2909’s, and the lower order 8 bits provide the B and A addresses. The microcode then performs a fetch microinstruction by taking the next ROM address from the upper order 8 bits of the instruction. It then begins executing the microcode to implement the specified instruction.

The instruction is placed in register 1 to allow part of the Instruction Field to be used as a modifier.

Interrupt Programming

The MARC has 15 vectored interrupts. Interrupts are only allowed when the microprogram fetches a new instruction. In fetching an instruction, the microprogram goes thru location 3F0. This location is hardwired to branch automatically to one of 15 locations

(3F1 ÷ 3FF) if an interrupt is pending. These locations contain jumps to the appropriate microcoded interrupt handlers. The individual handlers can either process the interrupt entirely or jump to a handler in main memory. A 16 bit hardware interrupt mask allows the user to selectively enable and disable any or all of the 15 interrupt lines. Interrupts are disabled by setting the hardware mask to all one’s. They are enabled by returning the mask to the original condition using a software copy of the mask word.

Input/Output Programming

The MARC has both parallel and serial interfaces. The parallel interfaces are 16 bits wide.

Two instructions (WORDI, WORDO) are available to work with them. These instructions allow for full word transfers to or from a GPR. The serial interface is a multiplexor with

eight I/O ports, The CHARI, CHARO instructions provide 8 bit transfers to and from a

GPR to this interface. Since it is a multiplexor, the 1/0 port number has to be specified as well as the device code. In addition to 1 word or 1 byte programmed I/O transfers, there are block transfers that perform like a software DMA. These block transfers are microprogrammed and allow a maximum data rate of 1 I/O word per instruction fetch. This is because transfers are interrupt driven and interrupts are acknowledged only on an instruction fetch. The DMAIN and DMOUT instructions are 2 word instructions which specify the device code, the number of words to transfer, and the starting address of the data block to be transferred. This address can be indirect.

Microcode Bit Assignment Details

As noted above, the microcode ROM has a 40 bit output. The 40 bits are assigned as follows: (see Fig. 3) and Appendix A. 10 bits are assigned for microcode branch addresses or microcode immediate data. 11 more bits control the AMD 2901 operations: three of these bits select the function (add, subtract, etc), three bits select the source of the data,

(A, B, O, or D) three bits select the destination of the data, (B, Q, B shift, etc). and two control the clocks to allow byte operations. Two bits are used for sequence control. These select the choice of Jump, Call, Return and Fetch in the microcode. Four bits are used for

FLAG selection allowing for 16 different branch criteria, including an unconditional branch and a don’t branch. Four bits are used to control strobing of data from the Data Bus for up to 15 different destinations plus one no-op. Three bits are used for specifying the B register address. Of the remaining six bits, four are used for the A register and two for the carry control. The A bits serve several functions. They are used for “effective-flag” selection which allows the main program easy access to the branch control. Whenever the

A register is not being used in an ALU instruction, the four bits normally used for A can be used for Bus control. The six bits which are normally Carry and A are also shared for shift control, since Carry and A are not used during shifts. The six bits then select the source for both the Q and the B register. The cleverness in reusing microcode bits only slightly affected the efficiency of the MARC, and saved a considerable amount in power required for microcode memory. However, a processor in which power is not a prime issue, it would be a good idea to have a wider microcode field and avoid a lot of worry about potential conflicts between various instructions.

Microcode Instructions

Microcode instructions could be defined in several different ways. We have chosen to define four categories of instructions: ALU instructions, Data Moving instructions, Branch instructions, and Clock control. There are really only eight ALU instructions when defined according to function. They are Add, Subtract (S-R), Negative Subtract (R-S), OR, AND

(R • S), Mask (R¯ • S), Exclusive OR, and Exclusive NOR. For the arithmetic Instructions

the following modifiers apply: Source, Destination, Carry, A Address, B Address and

Strobe. For Logical instructions, the Carry modifier is omitted. There are several variations of these eight instructions for ease of programming. For example, Add has two other definitions, INC which increments one General Purpose Register and puts the result on the

Bus and INCR which allows a Register to Register transfer of one incremented value. The

Data move instructions Load and Store are special cases of the OR ALU instruction.

There are eight Branch microinstructions: Call, Fetch, Return, JMP and the same four with the effective flag specified. Call and JMP require a specified address, Fetch and Return do not. All branch instructions require the specification of a flag, and execute if the specified flag is true. The clock instructions are used for Byte manipulation in the ALU. The microcode instructions are defined by the OPDEF’s in Appendix A.

Main Code Instructions

In addition to a full complement of register-to-register arithmetic and logical instructions

(including a microcoded multiply and divide), the MARC has double precision instructions to add, subtract, load, store and shift. Also available are instructions to add or subtract the contents of memory locations to registers and store the result in either register or in memory. There are one-word immediate instructions which add or subtract 0-15 to a general purpose register (GPR).

There is a move block (MOVBF) instruction that moves data from one part of memory to another. The MARC instruction set was not modelled after that of any other particular machine. It combines some of the best features of the PDP-11 and NOVA instructions sets. Generally, one MARC instruction can do anything either of these popular machines can do in one instruction, plus some things neither can do, such as double-word operations, single-bit operations, and block move.

MARC Instruction Format

The MARC’s instruction set is composed of single word (16 bit) and double word instructions, each containing an 8 bit opcode. The 8 bit opcode allows for a maximum of

256 distinct instructions. Because the instruction set was designed for a flight system in which memory space is limited and power consumption is a prime concern, there was an effort to make as many single word instructions as possible. The single word instructions referring to memory either have an address entirely in an index register (that may automatically be incremented or decremented) or they have an 8 bit address that is either absolute (0-255) or relative to the program counter (PC ± 128). In addition, there are single word immediate data instructions for add, subtract, and load byte (left or right, clear other half or not).

Because the instruction set is microcoded, it is flexible and not cast in concrete. There are about 40 unassigned instructions and others that could be given up depending on the needs of the project. There is a tradeoff between microcode memory (40 bits wide, must be high speed) and main memory (16 bits wide, can be a variety of speeds). However, if a group of instructions is used often, it may be worth microcoding them as one instruction.

Since our project required the MARC to maintain large status tables, bit manipulation instructions were very important. We have instructions to set, clear and test any single bit in memory tables and in registers. The memory tables can contain up to 32768 bits. The

MARC Instruction set is included in Appendix B.


The MARC has 16 (16 bit) registers, 9 of which are general purpose registers (CPR) that can be used in any of the following ways:

1. Accumulators for keeping the results of arithmetic and logical operations.

2. Address Registers that can be automatically incremented or decremented.

3. Index Registers where the contents are added to the word following the instruction to determine the effective address.

The remaining 7 registers are used for the program counter,, the stack pointer, and scratch registers for the microprogram.

Addressing Modes

Because of the need to keep the program size small there are a variety of addressing modes, many needing only one word to describe the address. Where these are inadequate, two-word instructions are available. The addressing modes are as follow:

Single Word Instructions

Page Zero - contains an 8 bit absolute address (0-255).

Relative to Program Counter - contains a two’s complement 8 bit displacement that is added to the PC to determine the effective address

Register indirect - full 16 bit address contained in a designated CPR.

Double Word Instructions

Absolute Direct

Absolute Indirect

Relative to Index Register

Relative to Index Register


16 bit address (0-32767)

16 bit address is added to the contents of a CPR to determine the effective address

Up to 4 levels of indirection are allowed. All instructions are either register-to-register or register-to-memory, Except for the move block instruction, which is memory-to-memory.

Except for a full word load-immediate, all double-word instructions are memory-to-register operations where the second word is a 16 bit address. These instructions allow for direct addressing of 32768 words. By adding an X to the instruction mnemonic, an index register can be added to the 16 bit address to get an effective address. By adding an I to the mneumonic, the effective address becomes an indirect address.

Branch Control

The MARC has a hardware condition code register containing the results of arithmetic, logical, shift and “load and test” instructions. The MARC can branch on the condition of the last result as follows: zero, non-zero, negative, positive, greater than zero, less than or equal to zero, carry = 1, carry = 0, overflow, no overflow, bit shifted out = 1, bit shifted out = 0, and unconditional.

There are three instructions that test these conditions, jump (JMP), jump and store return

(JSR), and return (RET). JMP(X)(I) and JSR(X)(I) are two word instructions containing a

16 bit address plus an index register if the X suffix is used. The effective address can be indirect if the I suffix is used. If the specified condition is true, a jump to the effective address is executed. In addition, the JSR instruction will push the old value of the program counter onto the stack before jumping. The RET on condition is a one word instruction that, if the specified condition is true, will pop the return address off the stack and add a specified displacement to it (0-15), storing the result in the program counter.

In addition to the jumps on condition, there are two instructions JIZ(X)(I), JDZ(X)(I) that will increment or decrement a CPR and will jump to the effective address if the result is zero.

Besides jump instructions, there are test and skip instructions that allow the testing of bits in registers or in tables in memory and skipping the following two locations if the bit is in the selected state (0 or 1).


Register 3, the Stack Pointer (SP), contains the address of a stack in main memory. The

Stack expands and contracts as data is written onto it and read out of it. There are 8 instructions that affect the Stack: JSR, RET, SAVRG, RSTRG, SAVCC, RSTCC,

SVALL, RSALL. The JSR and RET are described under branch control. The Save registers (SAVRG) and restore registers (RSTRG) instructions will push onto the stack and pop off the stack a specified number of registers starting with a specified GPR. The

SAVCC and RSTCC saves and restores the condition code on the stack. The SVALL,

RSALL, will save and restore the contents of all 9 GPR’s and the condition code at an address contained in a register. That register can be the stack pointer (R3) or any other

GPR. The Stack register must be set up by the main program before using any of the above instructions.


In order to microprogram the MARC we needed a general purpose assembler that would allow us to define an instruction set and then use multiple instructions from this set to define a single microprogram word. We wrote this assembler to run on our large IBM 370 time-sharing computer. It allows word sizes up to 64 bits and numbers can be entered in decimal, octal, hex, or binary.

To define in the instruction set, the OPDEF command is used. It defines the bit fields associated with an instruction. There are five types of field specifiers.




#T defines a fixed field of length # bits defines a variable field of length # bits.

defines don’t care bits causes the next field to start at bit position








Using the above definitions, the microinstruction

ADD 1, 2 and JUMP FETCH would be encoded as follows. (Assumes FETCH is at location 2).

Note that the 2 opeodes in the example have no conflicting fields. If they did, an error message would result when they were used together. If none of the opcodes in an instruction fill the don’t care bits, they are padded with zeros or with the master default word if one has been defined.

The assembler outputs an absolute core-image file that can be transferred over the IBM

370 time share terminal lines to a DATA I/O PROM programmer or to the development system for RAM loading. Also output is a listing file with a concordance map.


A breadboard version of the MARC and critcal portions of the signal processing system including the group demodulator and COP were tested as a system to prove feasibility. The

MARC was tested but not extensively operated in the programmed modes.

The microcode ability of the MARC was tested by using it to control the time and frequency tracking for the demodulator. The demodulator provided accumulator contents for up to 16 channels to the MARC. The MARC would convert in-phase and quadrature amplitude to phase, subtract phase due to modulation, and accumulate the phase error to determine frequency offset. It also measured differences in phase on specific data combinations to measure timing offsets. The demodulator was capable of providing an output to the MARC a maximum once every 25 microseconds. The MARC software and hardware to provide this function were tested by providing a simulated input from the minicomputer development system. Inputs were taken from a simulation 3 of the demodulator which was run on the IBM 370 and placed in a file. This file was transferred by the time share line to the disk in the minicomputer development system. Then the system provided approximately real time outputs to the MARC for testing the time and frequency tracking algorithms. We were able to determine the performance of these algorithms before they were actually hooked to the demodulator hardware. The MARC was tested with the demodulator and performed the time and frequency tracking task as expected. In a flight system the time and frequency tracking task would be performed on a special purpose microcoded device under the control of the MARC, using the algorithms developed on the MARC. Special purpose hardware would be more power efficient for this computational intensive task and this would avoid using up large quantities of MARC time which would be required for the control task.


The design of the MARC has demonstrated the feasibility of flying a processor with minicomputer capability in a Communication Satellite. The MARC has a full, versatile complement of all of the popular instructions plus some less common instructions which are tailored to the Control Job. These special instructions include Single bit instructions and Block move instructions. The MARC has 9 General Purpose Registers available to the programmer and is easily microcoded to provide special instructions and I/O handlers for speed intensive jobs.


We would like to acknowledge the assistance of B. H. Hutchinson, Jr., G. P. Gagnon and

D. C. Rogers in the design of the MARC and preparation of this paper.


1. L. S. Metzger, “On-Board Satellite Signal Processing”, IEEE National

Telecommunications Conference, Birmingham, Ala., Dec. 3-6,1978. pp. 8.1.1-8.1.5

2. I. Kalet, and L. N. Weiner, “Close Packing of PCSFS Signals-Model and Simulation

Results”, ICC 1978 Conference Record; June 4-7, 1978

3. D. B. Coomber, “A Group Demodulator for Close Packed Signals”, IEEE 1979

International Conference on Communications, June 11-14, 1979. pp 58.5-58.5.4

Appendix A

Field Assignments

Appendix B



D. N. Srinivas

COMSAT Laboratories

Clarksburg, Maryland 20734


Intersatellite links (ISLs) appear advantageous for future communications satellite systems. This paper considers satellites spaced 10E apart on the geostationary orbit.

Based on proposed communications performance goals, it discusses applicable ISL antenna acquisition and tracking system requirements and cites limitations on the initial

ISL antenna pointing accuracy and the required size of the antenna scan. Also examined are particular antenna configurations (single or 2-antenna), implementation concepts, and influences on spacecraft subsystems such as structure, power, and thermal considerations.


As the name suggests, an ISL connects two satellites which would have different locations in orbit and serve several ground stations. Typically, an ISL system would consist of a complete communications package and a tracking antenna assembly as a key subsystem. The communications package would receive and transmit at two different frequencies, which are presently assumed to be 25 and 32 GHz. These frequencies, which were proposed by INTELSAT to be shared with the radio navigation service, must be approved by the World Administrative Radio Conference (WARC) during its September

1979 meeting. The link would operate with 10-W transmitters and employ FM remodulation to save power. Global coverage satellite antenna beams would be used for the earth-to-space and space-to-earth links, and very narrow beams would be used for the

ISL. A major design constraint on the ISL transmission system requires that acceptable quality be attained during solar conjunction with both satellites.

This paper, which is based upon Reference 1, describes the ISL orbital concept and the

ISL antenna pointing errors due to spacecraft position and attitude errors. It also establishes the need for an antenna tracking system, outlines two typical antenna configurations, and stresses possible impact on spacecraft design.


Figure 1 shows the orbital deployment of two satellites as currently practiced; each satellite connects different ground stations (a, b, c, or d) within its respective coverage area. Under this arrangement for example, station a cannot reach station d directly. For such a connection, first the transmission from station a must reach station c and then travel via a ground link between stations c and d or by means of another satellite link which would require additional antennas at earth stations c and d.

An ISL would simplify this problem as shown in Figure 2. As indicated, station a reaches station d via the ISL and maintains contact with station c at the same time. Hence, the advantages of ISL implementation are that it reduces the number of up- and down-link transmissions. The longitudinal separation angle (a) between any two satellites can be varied as desired. However, for this ISL study, a separation angle of 10E was considered.

(The maximum value is about 160E when earth blockage occurs.) The 10E separation was based upon the concept of three satellites (two operational and one spare) with 5E minimum separation. Two satellite antenna configurations were considered.


For satisfactory communications performance, it was determined that a 52-dB ISL antenna gain would be satisfactory at both 25 and 32 GHz. This gain was based upon an antenna aperture diameter of 2 m and an assumed efficiency of 55 percent. The ISL antenna “loss of gain” was limited to 1 dB, a boundary that establishes the required accuracy of the antenna system.

At 32 GHz and for a Gaussian antenna pattern, the required angular accuracy for a 1-dB gain loss is or ±0.12E on either side of the peak, where c = speed of light (3 x 10


m/s) f = radio frequency (32 x 10



D = antenna diameter (2 m), and n = aperture illumination efficiency (0.55).

The above accuracy of ±0.12E must be supplied by the antenna control system.

After examination of the antenna pointing capability, it can be observed that the pointing errors result from the positional and attitude errors of the two satellites. Hence, restricting these satellite errors will reduce the pointing errors. Another factor which contributes to the pointing error is the longitudinal separation angle, ". Table I shows pointing error budgets for three different " angles with identical satellite orbital parameters. The orbital parameter limitations assumed for these calculations are typical values that could be achieved in the 1980’s.

Table I. ISL Pointing Errors







Position Attitude







Error (deg)






Position Attitude
















As indicated, and are the total ISL pointing errors in the normal (vertical) and in plane (horizontal), respectively. The total combined error of both vertical and horizontal, ,


, is equal to . It can be observed from the table that the separation angle directly influences the total combined error and that the position error dominates the error budget, which becomes smaller as angle " increases. However, with a larger separation angle, the antenna size would increase for the same frequency and antenna efficiency.

A comparison of the values of 2


(required accuracy) with ,


(satellite provided pointing accuracy) clearly indicates the need for an ISL antenna tracking operation mainly comprised of two distinct phases: mutual beam acquisition and mutual tracking during operation.

Acquisition Phase

In this phase, the antenna will be pointed by ground command toward the estimated location of the partner satellite. The following limitations affect the accuracy of the initial pointing: a. positional error of the partner satellite, b. attitude error of the host satellite, and c. resetting accuracy of the ISL antenna.

The resetting accuracy of the tracking antenna is assumed to be ±0.05E; however, different accuracies will be determined for different configurations. The scan box limits for the two satellites 10E apart in the mutual scan plane for acquisition would be

1.251 + 2 x 0.05 = 1.351

0.24 + 2 x 0.05 = 0.340

vertical (yaw) horizontal (pitch).

Tracking Phase

During this phase, the two partner satellites mutually track each other to satisfy the required pointing accuracy. The beam steering can be performed electronically or mechanically. Electronic scanning is constrained by the requirement to steer the transmitting beam using either multiple transmitter tubes feeding multiple horns or a variable power divider feeding multiple horns. Neither of these approaches is presently attractive. However, tracking error derivations can be accomplished electronically and used with conventional azimuth and elevation motor drives to steer the receiving and transmitting beams together.

Other promising methods would be monopulse tracking or step tracking. Monopulse tracking can be fast; if an amplitude monopulse sensor is used to close the ISL antenna tracking loop, two extra receivers (one for the azimuth and one for the elevation error signals) are needed. Proper comparison of the received signals determines the horizontal and vertical angles of arrival. Closed-loop servosystems operate by driving their output until the error signal is very small. A satisfactory error signal-to-noise ratio can be obtained by providing a beacon signal at the transmitter, thereby enabling a narrow bandwidth to be used in the error receivers. The error signal in the vicinity of boresight becomes weaker than the residual noise, thus defining a dead band for the pointing system.

The step tracking system commonly employed in earth stations to track synchronous satellites can also be used. Its sampling nature allows the step commands and error

signals to be looped back to a terrestrial control station. The step size established in either open-loop or closed-loop tracking must be large enough to prevent unnecessary stepping forward and backward, and yet small enough to permit several steps within the 1-dB contour of the antenna. Since the 1-dB contour is estimated at ±0.12E for a 2-m aperture, a step size of about 0.04E could be used.

Another method limits the ISL antenna pointing error and controls pointing through an onboard ephemeris computer model, which is an open-loop system receiving commands periodically from the ground. The ground command feeds the satellite orbital position errors into this model, and the computer then calculates and directs the ISL antenna to point in the desired direction. However, for this method the satellite attitude must be considerably small.

It should be noted that the greatest demand on the acquisition and tracking system occurs during the periodic satellite stationkeeping maneuvers. The maximum acceleration caused by the unwanted torque induced by firing the stationkeeping thrusters is estimated to be



. At this rate, the antenna would move away from the 1-dB beamwidth (±0.12E

from boresight) in approximately 11 s. This movement must be carefully compensated by selecting a very small bandwidth for the bandpass filter in the error detection circuit.


Concept 1

Figure 3 shows two independent Cassegrain antennas located on the east and west sides of a satellite. These sides would face the partner satellites in orbit. The reflector assembly would be mounted on a gimbaled joint to permit rotation at least around two axes. This rotational freedom facilitates the achievement of the pointing accuracy and the required tracking capability. A Cassegrain antenna has certain advantages over a comparable pointfed paraboloid: smaller f/D ratio, compactness, lower waveguide losses, and moment of inertia tracking movements.

Concept 2

Instead of using two separate antennas as in Concept 1, a single antenna system could be employed to perform the ISL function as shown in Figure 4. An offset-fed antenna is shown on the north face of the satellite. In this position, the antenna assembly is mounted again on a gimbaled bearing support assembly. During operation, the antenna assembly not only rotates about 180E but also tilts to accomplish the tracking function. The advantage of this concept is that it requires only one antenna assembly. The disadvantages are that it creates satellite unsymmetry and that the antenna rotational mechanism may become cumbersome.


Physical - The implementation of an ISL system would certainly increase the total satellite weight and, more importantly, require additional power. Table II lists the ISL system weight and power estimates based on separate discrete components. The antenna location should not block any field of view; it should clear the path of solar array rotation. It is desirable to maintain spacecraft symmetry, since otherwise unbalanced disturbance torques will arise.

Attitude Control System (ACS) - Symmetrical ISL antenna deployment about the spacecraft axes should not have any major effects on the ACS. Single antenna versions which create satellite asymmetry must be examined more carefully.

Thermal - The impact on the thermal control system will be observed in the event blockage of heat dissipating radiators, especially on the north and south facing sides. It is highly desirable to locate the ISL antenna as far as possible away from any dissipating area, not only to clear the blockage but also to avoid any local heating of the antenna.

Table II. Weight and Power Estimates

Item (No.)

Transmitter (2)






Power Supply (6)

Antenna Assembly (2)






















*Only one receiver andone transmitter are energized at any one time.


Basically, the design of an ISL system for satellite communications appears feasible.

Improvements in spacecraft control systems would simplify the ISL antenna tracking system design. Although it should not be difficult to build the antenna tracking system with existing technology, further investigation is desirable. Another demanding task would be the packaging of an ISL system on a spacecraft with minimal disturbances to other major subsystems.


This paper is based upon work performed at COMSAT Laboratories under the sponsorship of the International Telecommunications Satellite Organization (INTELSAT).

The author sincerely acknowledges the contributions made by G. Welti, M. Butcher, J.

Hsing, and C. Pentlicki.


1. Welti, G. R., “Intersatellite Link for Multiple-Access Telephony,” EASCON ’78

Conference Record, pp. 432-440.

Figure 1. Current Communications Link

Figure 2. Communications with ISL

Figure 3. Concept 1

Figure 4. Concept 2




Joseph V. MacPhee and David B. Coomber

Massachusetts Institute of Technology

Lincoln Laboratory


Features of the design of the telemetry and command system for the LES-8 and LES-9 experimental communications satellites are described. Particular emphasis is placed on access to command and telemetry functions for the communications user community.

Features described are:

1. Telemetry and command access through a communications channel in addition to dedicated telemetry and command channels.

2. The sets of telemetry and command functions for the communications user and for the satellite ground control center. Command set structure to allow for separable command sets.

3. Methods of display and control to provide for user control of the telemetry and command functions.

In addition, extrapolation of user control of satellites by an experimental test community to a more general user community is projected.

* This work was sponsored by the Department of the Air Force.

The views and conclusions contained in this document are those of the contractor and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the

United States Government.



J. Torres

Instituto Nacional De

Tecnica Aerospacial

Torrejon de Ardoz

Madrid, Spain

D. C. Patel

European Space Agency

Noordwijk, The Netherlands


Full spherical coverage from a single antenna cannot be provided at microwave frequencies on a normal satellite because of blockage by the satellite itself. However, cardioid coverage is sufficient. Achieving such coverage from a single antenna minimises the complexity of the on-board TT and C subsystem and so optimises both cost and reliability. Accordingly, investigations have been made into the feasibility of such an antenna in both S and C band versions.

The design is a cylindrical waveguide, propagating a rotating TE mode, terminated in a circumferential array of axial slots and a short-circuiting plate into which an aperture is cut. A short metallic skirt can be added to the outside of the antenna if it is desired to reduce the coverage from the maximum achievable. The design has the advantage of being

‘dual mode’, i.e. it has two isolated Rf connectors providing identical pattern coverage.

Thus two separate transponders can be ‘hard-wired’ on for reliability.

The paper deals with the design, main development stages, an performance - including environmental testing results - of the antenna, and includes the results of investigations into a ‘fill in’ antenna intended to complete the spherical coverage if desired.


Analysis of telemetry and telecommand antenna requirements indicates that for most missions, the coverage requirements may be met by a single antenna with cardoid shaped pattern. With this type of pattern, the influence of the satellite body is inherently minimal, a feature which enhances the application of such a standard antenna to a variety of spacecraft bodies. Typically the antenna would be mounted on a boom projecting from the satellite structure.

Two versions of such antennas - one for S band and one for C band - have been developed and tested, including full environment testing for the S band version. The antenna designs, development problems, the final resultant configurations, and their performances are discussed in detail.


The essential performance parameters for the S band TT and C antenna have been established as follows:

Operating Frequency band:

Total Power Gain (when boom mounted):

2.0 - 2.3 GHz

>-5 dBi for 2 <0 <"E, (0E <0/,<360E) where " = 140E for TM band and " = 120E for TC band

Maximum Azimuthal Ripple:


Channel Isolation:

Back Radiation:

1 dB


10 dB minimum

-10 dBi maximum

Rf Power: 10 watts under all conditions

The spherical coordinate system adopted is shown in figure 1.

To meet these requirements the antenna design shown in Figure 2 was initially developed.

The cylindrical waveguide carries a rotating TE


mode which is set up by the two probes excited in phase quadrature by means of a hybrid. The isolated port of the hybrid gives the

‘dual mode’ capability. The radiation takes place through the equispaced axial slots placed at the end of the waveguide. For this position of the slots and a rotating TE


mode in the waveguide, the axial slots radiate a circularly polarized field towards boresight. In the opposite direction towards the satellite, the radiation will be circularly polarized but of the opposite sense and of low intensity. In other directions, the antenna will be essentially horizontally polarized which is desirable. A conducting ring is placed inside the waveguide near the centre of the slots to match the slots to the field in the waveguide.

Such a configuration was analysed theoretically and shown to be promising. However, the mutual coupling phenomena, between the probes resulted in the requirement of an asymmetric (-2.5 dB) excitation. Furthermore, the VSWR, the bandwidth and the azimuthal ripple performances were found to be inadequate.

The essential undesirable feature, namely the requirement of an asymmetric excitation was investigated in detail. Of the various alternative feeder configurations like two crossed loops, planar crossed dipoles, crossed dipoles in log periodic arrangement, the last-named showed the most promise. Tests performed with a breadboard incorporating such a feed exhibited good VSWR characteristics but the ellipticity of the rotating TE


mode in the waveguide and hence the azimuth ripple proved to be difficult to maintain over the bandwidth. Further examination identified the apex of the feeder to be the problem area. A special printed circuit was manufactured and assembled, with satisfactory results.

In order to obtain a high radiation efficiency, the slotted waveguide section should be matched to the incident TE


mode. Numerical analysis indicates that the slots act as a shunt conductance to the incident TE


mode and may be matched by a shunt capacitance which can be realised by means of a conducting ring. The diameter and the position of such a ring was empirically established. This matching ring has a significant adverse effect on the radiation patterns because of reduced coupling through the end plate. An increase in the size of the hole, however, in the end plate restored the radiation performance.

The capacitive nature of the resonant ring makes it susceptible to charge to high potential on account of the secondary emissions from the surface of the spacecraft. To prevent breakdown and possible transmission of noise associated with such discharge to the command receiver, a D.C. bonding from the ring to the waveguide wall, having a high inductance per unit length at 2 GHz was incorporated.

A refined breadboard antenna incorporating the promising modifications was manufactured and tested. The coverage performance achieved is summarised in Table I.

Minimum Gain (0E < 2 <40E), dBi

Minimum Gain (40E < 2 <12E), dBi

Minimum Gain (120E <2 <140E), dBi

Equatorial Ripple (2 = 90E), dB

Antenna with skirt







<1 <1

Antenna without skirt








<1 <1

Table I . Coverage performance of the S band antenna.

It may be noted that the basic pattern is nearly toroidal. The addition of the skirt provides a means of further control over the radiation pattern. Alteration in the skirt dimensions

(angle, slant length) affect the total pattern while sliding the skirt along the waveguide provides the desirable control of coverage in the region from 2E = 90E to 2 = 180E.

Figure 3 shows E0/ polarisation patterns for the basic antenna as well as the antenna fitted with skirt.

The performance characteristics discussed above are applicable to an isolated antenna, while in practice the antenna will be mounted on a boom attached to the spacecraft. It is essential therefore to verify the ‘in-situ’ performance. The influence of the satellite body on the radiation patterns was computed using GTD and verified experimentally for variable boom-lengths (up to 1 meter) when mounted on a representative satellite body. From these measurements it was concluded that the effect of the spacecraft body is to introduce a ripple, as shown in Figure 4, in the region 0E <2 <40E and a somewhat steeper fall near the limit angle. No significant changes were observed when the metallic boom was replaced by a dielectric one.

For those missions for which full omnidirectional coverage is required, a fill-in antenna is necessary. The fill-in antenna must have conical coverage and adjustable polarisation in order to minimise the interference over the crossover region. One suitable antenna configuration identified and verified experimentally is the open-ended waveguide with a ring termination, to provide d symmetric pattern.

Subsequent to the developmental effort in which the feasibility of a cardoid antenna was established, effort was concentrated on the space application aspects. A refined lightweight model (0.8 kg), incorporating space qualified materials and processes was therefore constructed and space qualified through the normal environmental testing.


TT and C systems for a range of application satellites in particular, operate at C band

(4/6 GHz). In principle, to meet the coverage requirements, a C band antenna could be scaled from the S band antenna developed; however the frequency gap between the uplink and downlink bands is too large, and warrants an alternative approach.

The principle performance requirements for the C band antenna have been established as follows:

Operating Frequency Bands: 3.943 - 3.957 (TM)

6.168 - 6.182 (TC)

Total Power Gain

TM: (on-station):

TM (transfer orbit):

TC (on-station):

TC (transfer orbit):

Maximum Azimuth Ripple:


Channel Isolation:

Back Radiation:

Rf Power:

>-4 dBi over -20E <2 <20E (RHCP)

>-2.5 dBi over 70E <2 110E linear (E2)

>-4 dBi over -30E <2 <30E (LHCP)

>-2.5 dBi over 55E <2 <125E (E0/)

2 dB


15 dB

<10 dBi maximum

10 watts under all conditions

The performance requirements indicated that at least two radiating elements will have to be provided to comply with the polarisation specifications. The slotted waveguide antenna developed for the S band would radiate orthognal circular polarisations towards the axial directions as required for the on station coverage but only horizontal polarisation for the transfer orbit coverage. Hence, for this coverage the slotted waveguide antenna can only be used for either telemetry or telecommand, and the other operation must be carried out from another antenna with vertical polarisation.

In view of the frequency gap between the telemetry and telecommand bands the initial concept developed was to use the axially slotted waveguide for TM transmissions, and an arrangement of ring slots on the top plate for telecommand radiation. In the latter case, the ring slots would be excited in TM


, mode. Experiments however confirmed that while the axially slotted antenna would adequately meet the requirements for on station and transfer orbit telemetry case, the ring slots proved to be poorly coupled for effective radiation. The

ring slot concept was therefore abandoned, and an alternative approach was introduced for the telecommand case.

For telecommand case, the configuration adopted consisted of separate radiators for the transfer orbit and on station requirements.

For the transfer orbit phase, the coverage is obtained by means of a monopole placed on top of the waveguide antenna which is provided with a choke to decouple the monopole from the waveguide. The on station coverage can be obtained with two crossed dipoles in a self-phasing turnstile arrangement placed on the waveguide antenna top plate or with a helical antenna. The crossed dipole arrangement, being mechanically more attractive, was investigated experimentally. The tests indicated that the cross polarisation component is rather high and that the performance is very sensitive on account of dimensional criticality of the self-phasing circuitry. Though the crossed dipole arrangement can be improved with further development work, an alternative in the form of a five turn helix was designed and tested with satisfactory results.

The final version of the overall antenna assembly is shown in Figure 5. The helical antenna is excited via a coaxial coupler with the same cable exciting the monopole. Both the telecommand antennas are fed via the waveguide using a TM


, mode excited by means of an axial probe. The waveguide telemetry antenna is excited by means of a self-phasing crossed dipole with sheath balun arrangement (instead of a log periodic arrangement for the S band version). The axial slots are coupled to the waveguide by means of a resonant ring placed at 10 mm from the top plate. Since the top plate now supports the telecommand antennas, the straight slots have taken ‘I’ form to compensate for the lack of radiation through the hole in the top.

Tests performed on this antenna assembly have confirmed predictions and met the requirements. Like its S band counterpart, this antenna will be space qualified in the near future.


Two separate TT and C antennas (one each for S band and C frequency bands) have been developed to comply with the coverage requirements of most missions. The antennas are structurally simple, and can be boom-deployed from a satellite body or mounted on top of tower supporting feed elements of reflector antennas. While the requirements at S band have been met from a single slotted waveguide antenna, the wide frequency separation at

C band necessitated addition of separate radiators for the telecommand function. The radiators however are structurally integral.

In view of the performance requirements achieved it is expected that these antennas will be standard components for most European satellites.


1. Albertsen, N. C., Balling, P., and Laursen, F., “New Low Gain S-Band Satellite

Antenna with Suppressed Back Radiation”, 6th European Microwave Conference,

Sept. 1976, pp. 359-362.

2. Brain, D. J., “S-Band Telemetry Antenna Systems for Scientific Satellites”,

International Conference on Antennas for Aircraft and Spacecraft, London 1975, IEE

Conf. Publ. No. 118, pp. 64-67.


The work was carried out under contract for the European Space Agency. The theoretical and computational support provided by TICRA ApS , Denmark, is gratefully acknowledged.

Figure 1. Reference coordinate system.

Figure 2. The S band slotted waveguide antenna (with skirt).

Figure 3. Influence of the skirt on antenna radiation pattern.

Figure 4. Antenna pattern comparison for ‘isolated’ and

‘satellite mounted’ configuration.

Figure 5. C band TT and C antenna assembly.



R. Hoedemaker and C. Staloff

RCA Astro-Electronics, Princeton, N.J. 08540


This paper presents a solution to the problem of providing the electrical interface with the

NASA Shuttle and PAM D for an RCA Satcom Communications Satellite. The paper addresses the transition to the NASA Shuttle from the present era of expendable launch vehicles as it applies to the command and telemetry electrical interfaces. Final preparation of the spacecraft for launch into the transfer orbit rests with the Orbiter crew rather than the spacecraft contractor launch team. The added dimension of crew safety and the importance of simplifying the preparation task led the requirement for Electrical Airborne

Support Equipment (EASE). The conceptual design of this microcomputer-based equipment is described. The EASE monitors the health of the spacecraft and takes adaptive action where appropriate. It also sends prestored command lists to the spacecraft in response to crew-initiated keyboard command functions.


RCA Satcom is a 24-channel transponder, domestic communications satellite, the first of which was launched on a Delta 3914 expendable launch vehicle in December 1975. The fourth satellite of the series, called F-4, and other future spacecraft are required to be compatible with the NASA Space Transportation System (STS) using the MDAC PAM D.

If the shuttle is not ready, the F-4 will be launched on the interim shuttle backup and expendable launch vehicle, the Delta 3910.

The F-4 is fundamentally an existing spacecraft design. The spacecraft command and telemetry system with its concomitant ground processing software is a composite operating system which is difficult and costly to change. The construction of F-4 started in July 1979 and the modifications that are being made do not include changes in the telemetry frame or command system. Therefore, the electrical interfacing between the STS and the spacecraft must be accomplished via electrical airborne support equipment (EASE) that has the complexity and versatility to interface an existing telemetry and command system with the

STS system.


For an expendable launch vehicle, the job of human intervention in the preparation of the spacecraft for launch into the transfer orbit is completed with lift-off from the pad. For an

STS launch, the orbiter crew takes charge at lift-off and their participation in the launch continues until the spacecraft, attached to the PAM, is deployed from the shuttle bay.

Crew safety imposes a new set of requirements during this interval. Additional safety interlocks are required and the real-time telemetering of the status of these interlocks to the crew is essential.

Final preparation of the spacecraft for launch on the expendable launch vehicle is done by the contractor launch team. For the shuttle launch, this task reverts to the orbiter crew, which is not intimately knowledgeable about the spacecraft design. Therefore, the spacecraft preparation must be simplified and the go/no-go criteria condensed.

The spacecraft must be configured for several operating modes, launch, parking orbit, deployment from the shuttle bay, and a return landing if the mission is aborted.

Configuration for launch will be done by the contractor spacecraft launch team via the T-0 umbilical attached to the shuttle. The launch operating mode is identical to that used for an expendable launch vehicle except that the beacon telemetry transmitters are turned off, the propulsion system latch valves are closed, and the spacecraft batteries are in the trickle charge mode. When in the shuttle, power to the spacecraft is supplied by the STS via the

EASE. The spacecraft telemetry is monitored and processed by the EASE for further dissemination. These electrical connections are made through an umbilical attached to the spacecraft and passing through the PAM D slip rings provided for spacecraft use.

The PAM D is a solid spinning rocket motor designed to take the spacecraft from the 160 nautical mile altitude circular parking orbit to the 19,324 by 160 nautical mile altitude elliptical transfer orbit. The PAM D spacecraft combination is spun-up to 60 RPM by an electric motor driven spin table prior to separation from the orbiter.

The spacecraft parking orbit operating mode is configured by turning off the spacecraft momentum wheels. These units are run at the low speed of 200 RPM to prevent brinelling of the bearings by the launch vibrations. During this mode, the EASE cycles the heaters on the spacecraft solid propellant apogee kick motor (AKM) to maintain its temperature at

25EC. The AKM is housed in the central structural cylinder of the spacecraft and is used to take the spacecraft from the elliptical transfer orbit to the equatorial geostatignary orbit.

The deployment operational mode commences approximately one-half hour before ejection from the shuttle. Implementation of this mode on the spacecraft starts by turning on the

hydrazine thruster catalyst bed heaters. The thrusters are used to provide nutation damping for the combined PAM D/spacecraft after separation from the shuttle. Once out of the shuttle, there is a 45-minute coast period before the PAM D solid rocket motor is fired.

The orbiter uses this time to take evasive action. The spacecraft provides the thruster damping during this period to prevent coning build-up.

Nine minutes before deployment from the shuttle, the spacecraft beacon is turned on in high power. The momentum wheels are commanded to the 200 RPM low speed, and the battery chargers are placed in high charge at five minutes before deployment. The EASE power to the spacecraft is turned off at this time, which places the spacecraft on internal battery power. The AKM heater is also turned off. Then at one minute before deployment, the AKM is armed simultaneously with the arming of the PAM D by rotating the safe and arm motor. This is initiated manually by the orbiter crew via the standard switch panel

(SSP) located in the crew cabin.

Placing the spacecraft into each one of these operating modes is accomplished by a single general-purpose computer (GPC) initiated keyboard command. The keyboard command is transmitted to the EASE by the astronaut via the sequence control assembly (SCA) of the

PAM D airborne support equipment. Upon receipt of the discrete command from the SCA, the EASE issues the appropriate command sequence to the spacecraft at the correct time.

Hence, the crew has one command to issue to put the spacecraft into the parking orbit operational mode or to put the spacecraft into the deployment operational mode. The

EASE automatically does the rest of the required action. The arming of the AKM is initiated manually in parallel with the PAM D via the PAM D signal conditioning unit


If there is a mission abort, the action taken can be reversed. The AKM can be placed on safe via the SSP. The spacecraft can be placed in the parking orbit operational mode by keyboard command, and for the actual re-entry returned to the launch operational mode by keyboard command.

After deployment from the shuttle, the PAM provides the command to turn on the spacecraft parking orbit nutation damper (POND). Prior to firing the PAM solid rocket motor, the PAM provides the signal to turn off the POND. The PAM also enables the spacecraft launch omni antenna deployment system. The launch omni antenna, which radiates the beacon telemetry signal, is folded to fit inside the shuttle.

The antenna deployment is initiated by separation switches that activate when the spacecraft is separated from the PAM after the PAM achieves the transfer orbit.


Spacecraft and EASE telemetry is sent to the ground via the orbiter Payload Data

Interleaver (PDI). This data is then separated from the other payload data and sent to the

RCA Americom tracking station at Vernon Valley, N.J. for monitoring by the spacecraft engineers.

Normally, there would be no need for the spacecraft engineers to uplink commands to the orbiter for the spacecraft. The capability however is planned to be able to send any spacecraft command to the orbiter GPC and from there have it transferred to the PAM

SCA for transmission via the EASE to the spacecraft.

The downlink data contains the standard formulated spacecraft telemetry plus an additional telemetry frame which contains a record of the commands transmitted to the spacecraft by the EASE, the data sent by the EASE to the GPC via the PAM SCA, the detailed results of the telemetry limit checking, and the EASE status and temperature.

Voice communications also exist between the orbiter crew and ground personnel to resolve possible equipment problems. The ground has access to all telemetry and has the capability to send commands to the spacecraft via the orbiter GPC. The orbiter crew has access to all safety related telemetry, spacecraft health status and the spacecraft operating mode telemetry. The crew can command the spacecraft into any one of its operating modes and also switch over to any of the redundant EASE units.


The EASE provides the interface between the spacecraft, the PAM D airborne support equipment, and the orbiter. A system block diagram is shown in Figure 1. Table 1 contains a list of interface signals to the PAM SCA and the EASE, and Table 2 contains a list of umbilical connections.

The EASE equipment is divided into four parts: the Command Master (CM), the

Telemetry Master (TM), the Power Master (PM), and the spacecraft interface box. The fourth part is used only during pre-launch activities and will not be discussed further.

Each of the first three parts is redundant. The redundant CM can be selected by the crew via the standard switch panel. The redundant counterpart of the other two units, the TM and PM, may be selected via the GPC keyboard command.

The EASE receives spacecraft mode configuration discrete commands from the GPC via the PAM SCA. From these, the EASE formulates necessary commands to send to the



Electrical Power Spacecraft

Electrical Power Return

Unregulated S/C Bus Voltage TLM

S/C Battery 1 Voltage TLM

S/C Battery 2 Voltage TLM

S/C Battery 3 Voltage TLM

S/C Telemetry Return

Safety Status Data*

Safety Status Return

RTM Data

RTM Data Return

Data Frame SYNC

Data Frame SYNC Return

Data Word Clock

Data Word Clock Return







Command Tones

Command Tone Return


POND ON Return



Omni Deploy Enable

Omni Deploy Enable Return


Power Master










TLM Master


TLM Master


TLM Master







CMD Master








Terms: S/C (Spacecraft), PAM (Payload Assist Module), SSP (Standard Switch

Panel), CMD (Command), TLM (Telemetry).



Power Master

TLM Master

TLM Master

TLM Master

TLM Master

TLM Master

TLM Master

TLM Master

TLM Master


TLM Matter


TLM Master









CMD Master







*Refer to Table 3.

spacecraft. All command codes are checked against a hazardous set of commands and if hazardous, they are not transmitted to the spacecraft.

After commands are sent to the spacecraft, the telemetry is checked in a delay look-up table for the appropriate response. The EASE receives from the spacecraft the basic commutated telemetry (RTM data),additional safety telemetry (safety data), and several analog telemetry points which may be used directly for control of the power subsystem.

All of this telemetry data is processed, GMT inserted, and put in a standard format suitable for transmission via the orbiter PDI link to the ground. Critical telemetry points are limit checked and where necessary, the limits are changed as the result of commands sent to the spacecraft. The results of limit checks, the commands sent to the spacecraft, and EASE status are also processed for inclusion in the PDI transmission.

The EASE TM sends discrete and analog data to the GPC via the PAM SCA. The safety related telemetry is transmitted as discretes to satisfy the requirement for real-time monitoring by the crew. Also included in the discrete transmissions are the spacecraft operating mode, and a composite indicator as to the outcome of the spacecraft telemetry limit checks. The analog telemetry is converted into digital form by the PAM SCA.


The STS launch imposes additional safety requirements. The spacecraft has several potentially catastrophic hazardous systems that could cause personnel injury, loss of life, or prevent the safe return to Earth of the Orbiter Vehicle. These are the AKM, the hydrazine propulsion system, solar array deployment system, launch omni antenna deployment system, and the momentum wheels. In addition to catastrophic hazards, there are critical hazards which if they occur would result in damage to equipment, or the use of contingency or emergency procedures.

Critical hazards must be single failure tolerant and catastrophic hazards are required to be two failure tolerant. The orbiter crew must have the capability to monitor in real time all of the safety inhibits to the potentially catastrophic hazardous systems. It is this latter requirement that has increased the number of spacecraft telemetry points and creates the need for the special safety data. The problem of making this additional safety status telemetry available to the crew is solved by collecting it on the spacecraft and formulating it into three 8-bit digital words. Table 3 is a list of this telemetry. The squib bus is common to the AKM and solar array deployment. The AKM SAFE and ARM Telemetry signals are wired to the Signal Conditioning Unit (SCU) for a PAM/Spacecraft combined display on the Standard Switch Panel in the crew compartment.



- Squib Bus Status*

- AKM Fire Relay Status

- Squib Bus Separation Switch Status

- Safe and ARM Motor Status

Solar Array Deployment

- Solar Array Squib Fire Relay Status


- Latch Valve Status*

- Thruster Solenoid Status (twelve thrusters)

- Thruster Logic Power Separation Switch Interlock Status

- POND ON/OFF Relay Status

Launch Omni Antenna

- Omni Antenna Squib Bus Enable Relay Status

- Separation Switch A Status

- Separation Switch B Status

Momentum Wheel Assembly

- Wheel Speed*

* These telemetry points are in the existing telemetry frame. In all cases there are non-RF command safety interlocks to prevent spurious RF commands from removing all safety inhibits in a hazardous system.


The spacecraft subsystems consist of command, telemetry, propulsion, power, thermal, and transponder communications. The transponder communications subsystem is not powered in the shuttle.

A spacecraft command consists of a 32-bit digital word sometimes followed by a 16-bit data load word. The 32-bit command word contains synchronization bits, spacecraft address bits, execute bits, a bit to distinguish between direct commands and data load commands, and an eight-bit plus parity operation code word. These are transmitted to the spacecraft at 100 bits per second as tones. A mark is a 8 kHz tone. A space is a 5 kHz tone, and an execute is an 11 kHz tone. The execute tone of the following command executes the previous command. The command system is redundant and so is the EASE


The CM puts the commands into the correct tone format and sends them via the umbilical to the spacecraft. Side 1 of the CM operates exclusively with side 1 of the command subsystem and side 2 CM operates with command subsystem side 2.

The command subsystem normally receives commands via the transponder communications antenna, the launch omni antenna, or the orbit omni antenna. The antennas feed a command communications receiver which demodulates the FSK tone data.

The system is modified for shuttle operation to receive the command tones directly via the umbilical bypassing the RF link. The command, ranging and telemetry system simplified block diagram is shown in Figure 2.

The S/C telemetry system consists of a 128-sample, two-second commutated telemetry frame, safety telemetry, and analog telemetry. The commutated telemetry is normally transmitted as a pulse amplitude modulated signal.

The pulse amplitude frame is converted into a digital frame and transmitted down the umbilical to the EASE TM with a frame sync and word clock. The safety telemetry is formed into three, 8-bit digital words and transmitted serially as a separate line down the umbilical to the TM. The AKM safe and arm telemetry indicators are carried directly in the umbilical to the TM and PAM SCU. The battery voltages and unregulated bus voltage are carried in the umbilical as 0 to 5 volt conditioned analog voltages.

The pulse amplitude modulated telemetry frame also frequency modulates a 14.5 kHz subcarrier, which in turn phase modulates the beacon transmitter. The beacon transmitter operates with a low power, 0.15 watt, output into the transponder communications antenna

and with high power, 1.8 watts, into either the launch or orbit omni antennas. Refer to

Figure 2.

The power subsystem unregulated bus runs directly off the solar array or the spacecraft batteries. The upper voltage is clamped by the shunt current limiters and the lower voltage is set by the batteries. The EASE PM must provide power to the spacecraft at a voltage less than the shunt limit and greater than the battery voltage so that the batteries can be trickle charged. Figure 3 shows a simplified block diagram of the power subsystem.

Thermal control of the spacecraft is achieved passively by using optical surface radiators, and thermal multilayer blankets. Passive control is augmented with heaters on the batteries,

AKM propulsion tanks , valves, and lines. Thermal control devices turn the heaters on and off except for the AKM. The AKM temperature is monitored by the TM and heater commands issued by the CM to turn the AKM heater on or off as required to maintain its temperature at 25EC.

The EASE interfaces with the spacecraft umbilical. There are two 37-pin umbilical connectors, which are used redundantly. The umbilical connections to the EASE pass through the PAM sliprings of which 52 are available for spacecraft use. The umbilical signals in each umbilical are listed in Table 2.



The TM is a dual redundant EASE unit whose redundancy selection is under the control of the CM, and which is powered by the STS power bus as distributed by the PAM D Spin

System Distribution Box. The TM is a multi-task unit whose primary function is the signal conditioning and formatting of data for downlink transmission using the STS PDI interface port. A secondary task is the testing of collected data with reference to programmed end limits to determine S/C and EASE health and status. Another secondary function is the buffering of selected data for presentation to the STS crew via the PAM D Sequence

Control Assembly and the STS General-Purpose Computer. The TM additionally exchanges data with the CM units to enable selected adaptive commanding to correct potential malfunction and control requirements.

TM Data Frame Formatting

The TM receives data from multiple sources which it signal conditions and reformats into a PDI standard format. The resultant output data stream is encoded in a differential Bi0/-L format for ground transmission via the STS PDI port. The composite data stream consists

of a 256-word, 2-second major frame. Each frame contains a contiguous time buffered

128-word data sub-set which is the original S/C TLM data. By maintaining the format integrity of this data sub-set, the existent processing ground station software can be used without modification. The full frame content is shown in Table 4. The frame is initiated with a five-word frame sync followed by the GMT Time Code. The four-word time code represents the first complete time code transmission following the start of frame sync of the preceding telemetry frame. The internal EASE status is telemetered in the next nineteen words. SCA formatted discrete and analog points occupy the next group of thirteen words.

The Safety Status data as described earlier in this paper is included in the next group of three words. Command verification of CMD OP Codes and Command Data Words is assigned to the next group of twenty-four words. The verification words are derived from a pushdown stack. This implies that the last command received is the first verified. The stack is reread in each succeeding frame with the oldest commands eventually discarded as new commands fill the stack. This results in multiple transmission of verified command lists. The next set of three words is dedicated to “Out-of-Limit Words.” These words are internally generated within the TM. Selected data points are continuously monitored and tested against stored end limits. When received data falls outside of valid limits, an out-oflimit word is generated which gives the source address. This out-of-limit word is both telemetered and sent to the CM where appropriate commands may be generated to rectify the out-of-limit condition. Delta limit commands are placed in the next two-word slots.

These are commands which the CM sends to the TM to inform it that test limits must be changed to correspond with new commanded functions. Each command is identified by its assigned eight-bit op code. The following thirty-two words are used for memory dump. On command, the TM will sequentially unload its memory in 32-word blocks per frame until the complete data field is dumped via the telemetry link. On completion of the dump cycle, this data area will be unassigned and filled with a PN word. Twenty-four spare words follow; this allows for an approximate 10% growth in transmitted data. The last 128 contiguous words are the S/C standard telemetry data starting with word 1 and ending with word 128 (S/C frame referenced).

Data Limit Testing

A preprogrammed set of data sources is selected for limit testing. Under normal operation when measurements fall within the prescribed boundaries, no special action is taken. In the event that a test shows abnormal operation, the TM flags the CM via a dedicated interface and concurrently informs the ground via special frame words of the test results. Specific remedial action is then relegated to the CM or to the ground operations center which may choose to disable automatic responses and manually control the S/C. In several cases test limits will rely on the commanded state of the spacecraft subsystems. For example, thermal measurement limits will depend upon equipment ON/OFF states. To allow for this factor, the CM transfers to the TM via a unique interface a ) Limit CMD Word. These words are


WDS 0-4


























RTM DATA retransmitted to the ground as part of standard frame format. This enables full ground station check and control of EASE operations. In the case of AKM heater temperature control, the TM/CM delta limit word and command response is used as the normal method of loop closure to control the requisite temperature.


An interface diagram of the TM is shown in Figure 4. At the top left of the drawing are two sets of eight inputs each, which are sourced by the S/C telemetry system. These signals contain the S/C standard and safety status data in conjunction with frame sync and word sync. This enables the TM to receive, buffer, and subsequently output this data in its

PDI frame format. At the left center of the drawing is a set of six commands which control the operational modes of the unit. The TM ON/OFF commands enable the selection of either redundant half of the unit in an EXCLUSIVE OR selection. The RTM source selection is used so that either S/C RTM (redundant) can source either redundant TM section, thus enabling a full cross-strap configuration. The Data Test Override is a ground sourced command which can disable the data limit test function. The remaining input signals depicted at the lower left of the drawing are required for the input of delta limit commands from either CM and for the input of GMT data and unregulated power.

The TM outputs are illustrated at the right of the figure. The PDI Data and Return line carry the differential Bi0/-L data directly routed to the PDI Port. The middle set of four lines supply the out-of-limit measurements redundantly to both Command Master Sides.

One of the remaining outputs is status to the SCA (Table 1) which requires twenty-five discrete points (system capacity is for 40 points) and eight analog points. The other output

is data to the T-0 umbilical. During the pre-launch mode, the composite telemetry frame will be output via this port to enable a rapid checkout and evaluation of the S/C-EASE equipments.

The TM block diagram illustrated in Figure 5 depicts the detailed organization of one redundant side of the TM. The unit is structured as a microcomputer controlled data bus.

All input and output signals with the exception of the SCA analogs are processed by the eight-bit-wide semi-duplex data bus. Bus control and computer I/O is serviced by a discrete logic unit, the “Central Sequencer and Computer I/O Controller.” The discrete logic and the microprocessor are all constructed using C-MOS logic. This technology also includes the RAM and PROM chips. The microprocessor selected for the application is the

CDP 1802 manufactured by RCA and qualified and used extensively in space and STS equipments. The diagram has a Phase Locked Loop Block. The loop is used to generate a bit clock from the input word clock, enabling the proper storage of the serial RTM data streams. All required internal voltages are derived from an internal DC/DC converter powered from the unregulated STS power bus.

In addition to the digital and power processing circuitry, support analog circuitry is required to multiplex and A/D convert analog inputs. Capacity is built into the unit to process up to sixteen analog signals using C-MOS analog multiplex chips supported by a single chip qualified 8-bit A/D converter.

All components meet both Hi-Rel and mission life and radiation requirements. The total

TM is estimated to weigh 5 pounds and occupy a volume of 5"x8"x8".



The CM is a dual redundant EASE unit whose redundancy selection is under the control of the Standard Switch Panel (SSP) and which is powered by the unregulated STS Power

Busses. The CM receives command inputs from the PAM D SCA which it processes into the appropriate format and routes to the S/C or EASE units as applicable. The CM also receives commands via the T-0 umbilical which it distributes. Upon receipt of out-of-limit

S/C conditions as tested by the TM the CM takes appropriate remedial action. (This adaptive facility can be overriden by command.) The CM supplies to the TM an update of commands processed which influence the limits for TM data tests. The output commands delivered to the S/C are synthesized as tone sequences which are identical to the S/C command receiver output. This facilitates the S/C command interface and guarantees maximum command noise immunity on a systems basis.

The CM has stored in its memory command tables which can be accessed by single commands. These tables include both operation codes and time delays. A typical operation would cause the first tabular command to be executed with the table call up. After execution, a time delay associated with that command is processed and when the delay the next command in the sequence is outputted. The next command is processed in accordance with the time delay following the second command. In this manner a long sequence of commands with appropriate time interval spacing can be executed and called using only one table reference command. Current design includes three such tables:


S/C to Return Mode


S/C to Parking Orbit Mode


S/C to Deployment Mode


A system I/O drawing of the CM is shown in Figure 6. At the upper left of the drawing, the top five input lines are associated with the application of bus power and with redundant side selection. Additional detail of bus distribution to the equipment appears in Figure 7.

One bus line is wired to CM side 1 and the other to CM side 2. A switch on the Standard

Switch Panel is used to select which side is to be used. Once the CM redundancy selection is performed, all other redundancy selections (PM and TM) are achieved by commands processed via the selected CM side. Both sides are fully equivalent in their functional capabilities and are powered with one side or the other on exclusively. The next input set shown relates to the commands processed by the PAM D SCA. As shown in Table 1, twenty-one commands are required; this falls within the SCA capability of outputting 48 discrete commands. Capacity is designed into the CM to accept up to 24 commands. The

T-0 umbilical input shown in Figure 6 is the path for the pre-launch S/C commands which are properly formatted and supplied, encoded as tones. The CM does minimal processing with this input; it “ORs” these commands with the CM generated ones so that they can be supplied to the same S/C input port. The last four inputs which are shown at the lower left of Figure 6 relate to the out-of-limit measurements processed in the redundant sides of the


The CM outputs are shown at the right of the figure. The upper four lines are used to redundantly supply in a cross-strap mode those commands which alter the TM test limits.

The final outputs are redundant lines which supply the tone-encoded, formatted serial command words.

The CM Block Diagram, Figure 8, shows the internal organization of the unit. The design is based on the use of a data bus under microcomputer control tied to each of the sundry

I/O ports. The actual control of the I/O operation is controlled by a Central Sequencer and

Computer I/O (CSCI) Controller working in context with the microprocessor. The CSCI is designed using SSI and MSI COS/MOS logic. The microprocessor (CDP 1802) and its associated RAM/PROM memory are also CMOS units specially selected for highreliability, radiation hardness, high noise immunity, and low power consumption. In addition to the digital processing capability, the unit has an internal DC/DC converter and an analog tone generator. The tone generator is used to simplify the interfacing of commands to the S/C command system.

The complete redundant CM unit is estimated to weigh 5 pounds and occupy a volume of



The interface system herein described enables a spacecraft designed for launch via an expendable launch vehicle to be mated to the STS with minimum spacecraft modifications.

All STS safety and mission requirements are complied with and complete operational flexibility is maintained.

Figure 1. STS/Spacecraft Block Diagram

Figure 2. Command, Ranging, and Telemetry System Block Diagram

Figure 3. Power Subsystem

Figure 4. TM Input/Output Drawing

Figure 5. Telemetry Master Block Diagram, Side 1

Figure 6. Command Master Input/Output Drawing

Figure 7. EASE Redundant Power Distribution

Figure 8. Command Master Block Diagram, Side 1



M. L. Hull

Assistant Professor

Dept. Mechanical Engineering

University of California, Davis

C. D. Mote, Jr.


Dept. Mechanical Engineering

University of California, Berkeley

L. W. Lamoreux

Senior Development Engineer

Biomechanics Laboratory

University of California, Berkeley


A data handling system was developed which transfers serial pulse-code modulation

(PCM) telemetry data from an analog tape to a digital tape for analysis on a digital computer. The PCM data were collected from field experiments in snow skiing which measured the excitation between the ski boot and the ski. PCM data were FM transmitted and stored on an analog tape recorder. In the laboratory, data were decoded from the tape and parallel input to a Nova minicomputer which buffered the data and wrote tapes compatible with a CDC 6400 computer. A custom-built PCM decoder provided control and status commands to drive the minicomputer interrupt logic. The data transfer rate was

50 kbits/s.

Special consideration was given to the problem of information losses. Lost data frames were not stored by the minicomputer and the data time series on the digital tape is discontinuous at each loss point. Spectral analysis of data with discontinuities produces erroneous results. Fourier coefficients and power spectra were computed for both continuous and discontinuous signals. Discontinuities caused significant reductions in amplitude and increase in bandwidth of spectrum estimates. Unique software eliminated data discontinuities by reconstructing the original time-base with linearly interpolated pseudo-data. Results are presented which show the enhanced accuracy obtained in spectrum estimates with the reconstructed data.


In field applications involving remote data acquisition by RF telemetry a common and frustrating problem is that of telemetry dropouts. The word “dropout” usually refers to information lost in the telemetering process. As will be discussed, however, information losses may also stem from equipment used to handle stored data. The word “dropout” will be used herein synonymously with the words “information loss”. The word “dropout” will be qualified to denote whether data handling equipment or telemetry was the source. When the relative position between the transmit and receive antennas changes, nonoptimal telemetering situations may develop where the transmitted signal is actually not detected by the receiver. Accordingly, the information transmitted during this interval is lost. The mechanisms responsible for lost reception are difficult to isolate because of the complexity of the RF transmission process. At least three different mechanisms, however, are associated with transmission difficulty over short ranges (less than 10 km). Most RF telemetry is FM because of the inherent superior signal/noise. The quality of FM transmission is sensitive to maintaining a line-of-sight to the receiver. One telemetry dropout mechanism is objects which momentarily block the line-of-sight path between transmit and receive antennas. Signal strength drops below the receiver threshold and reception is lost. Another mechanism is the changing impedance of the transmit antenna as it receives RF signals reflected at varying signal levels from the immediate surrounding environment (1). Because the transmitter frequency depends on the load impedance, the transmitter carrier actually shifts and reception is lost when the receiver becomes detuned.

This problem can be circumvented by incorporating an RF isolator into the path between the transmitter and transmit antenna. The isolator, by absorbing the reflected energy, maintains a nearly constant load impedance. A final mechanism is the cancellation of the received signal when reflected RF waves arrive ~180E out-of-phase with the line-of-sight

RF transmission (2). This is a common problem at UHF telemetry carrier frequencies in the 215-250 MHz band. Reception improvements are possible by using a directional receiving antenna which is insensitive to reflected incident waves outside its radiation pattern.

To insure high quality telemetering, attention must be given to each of the three mechanisms. Even with line-of-sight transmission, an isolator and a directional receive antenna, telemetry dropouts may still be problematic, however. In telemetry over long distances, for example, transmission quality is affected by the ionization characteristics of the atmosphere. Also, even over distances less than a kilometer, the receive antenna may not be sufficiently directional to eliminate unwanted signals. Finally, the radiated signal strength at any point in space is dependent on the radiation pattern of the transmit antenna.

if the orientation of the transmit antenna changes, then the radiated energy at the receive antenna may be sufficiently attenuated to lose reception (3). The severity of this problem is reduced by using omnidirectional antennas such as the helical dipole but the transmitter

power requirements are increased (4). So, while in many cases the telemetry dropout problem can be minimized, dropouts cannot be completely eliminated.

The dropout problem is important because it bears not only on the quality of transmission, but also on the data reduction process and interpretation of results from data analyses. As noted, dropouts result in lost information. In certain situations it is possible to estimate or reconstruct the lost information. If the dropouts are long in duration compared to the lowest frequency of interest in the data, then it is not possible to reconstruct any meaningful information. If the dropouts are short compared to any frequency of interest, however, then it is possible to estimate lost information using some interpolation scheme.

Estimating lost data (i.e. replacing noise with meaningful information) is critical for certain basic data analysis techniques where numerous short duration dropouts would preclude analysis altogether. An example is spectrum analysis where one wishes information regarding the frequency content of the data. Here, one must have a data segment free of noise at least twice the length of the period corresponding to the lowest frequency of interest. Without such a data segment, valid analysis is not possible.

This paper discusses a new technique for partial reconstruction of pulse-code modulation

(PCM) telemetry data with dropouts. In the research application which yielded the PCM data, it was desired to compute power spectral density estimates. The enhanced accuracy of the spectrum estimates obtained with the reconstructed data are illustrated on a sample problem. The data reconstruction technique is actually one phase of a large intermediate data processing scheme which is also discussed.


The injury rate in snow skiing, approximately 3 lower extremity injuries per 1000 skier days, has been maintained over the past two decades. In 1974, a research program was undertaken at the University of California with the objective of measuring the resultant excitations between the ski boot and ski during both cruising skiing and the critical situations which call upon the ski release binding to release. This data is important to the design and adjustment of the ski release binding. These research objectives demanded that accurate data be remotely acquired. Accordingly, a complete instrumentation system was designed and developed for the skiing injury problem. A precision PCM-FM telemetry transmit system shown pictorially in Fig. I and schematically in Fig. 2 was custom built to transmit data from two six degree-of-freedom strain-gage dynamometers mounted inside the test ski to a receiving station about 3 km distant. In the telemetry package shown in

Fig. 1, dynamometer signals were conditioned by a high gain, low noise carrier amplifier system. The 13 data channels were serially pulse-coded into 12 bit words each at a 520 words/s digitizing rate. The 16-word data frame also shown in Fig. 2 was completed by a sync word (word with a constant binary value for synchronization of data frames) and two

12-bit frame counter words. The 520 words/s digitization rate yielded a real time data transmission rate of 100 kbits/s. Subsequent to PCM encoding, data were further coded into bi-phase L(Bi-N L) and FM transmitted.


A data handling system is necessary in many telemetry applications because it is often not feasible to analyze data in real time at the receiving site. A bulk storage device is required so the data can be subsequently analyzed. PCM data analysis necessitates transferring the data from the field storage device to the computer. In this case, most data handling operations include the following:

1. An analog tape recorder which can record the data in real time and reproduce the data either in real time or at a reduced data rate;

2. A PCM bit and word decoder;

3. A hardware interface to the decoder which writes computer compatible tapes in real time or at a reduced data rate, or

4. A minicomputer with a data interface and the capability to write computer compatible tapes.

When the skiing excitation data were received (Fig. 2), they were decoded in two stages by the 12 bit, 16-word PCM decoder shown in Fig. 3. The bit synchronizer stage formed the serial PCM data and the word synchronizer stage blocked the PCM data into 12-bit words and 16-word data frames. To monitor system function, data were presented visually in rows of 12 (bit) panel lights and any two channels could be selected for D/A conversion.

The data consisted of a number of approximately 30 second skiing runs sequentially recorder on a 7 track, 250 kHz Sanborn instrumentation recorder. To achieve maximum data density, runs were recorded on one track until it was full and then recorded on the next track after the tape was rewound, etc. Recording data on all seven tracks at 60 ips gave a data density of ~12000 bpi.

In the laboratory, bi-phase data, reproduced from the analog recorder, were played into the custom built PCM decoder as depicted in Fig. 4. The decoder design is discussed in detail by Hull and Mote (5,6). As the serial bi-phase data were decoded, they were output along

12 parallel lines to the data interface in the Nova minicomputer. The PCM decoder provided status and control commands along three additional lines to drive the minicomputer interrupt logic. The timing diagram for command signals is shown in Fig. 5.

On one line, a 1 µsec FRAME PULSE denoted the beginning of each new data frame and on another line a WORD PULSE signified a new word. On the third SYNC LEVEL line, a transition from +4.0 V, or logical true, to 0.V signified a synchronization loss arising from an information dropout. The sync word recognition circuit in the decoder, which counted the number of properly timed sync words, controlled the level transition. If three consecutive sync words were 16 words apart, the circuit was synchronized (true); if three consecutive sync words failed to appear or were improperly timed, a lock-loss (false) was registered. Therefore, the minimum information loss for a recognized dropout was six data frames or 6/520s.

A Nova (Data General Corp.) minicomputer was used to buffer the data, check the timing of the control commands, and write CDC 6400 compatible tapes. According to the flow diagram in Fig. 6, word pulses were counted and checked against arriving frame pulses.

Errors caused by a word count different from 16 effected a program halt. As data were accepted, they were stored in one of two 1500-word buffers which were alternately being loaded with data and written onto CDC 6400 tape. The sync word (word 0) at the beginning of each frame was not stored and each buffer contained 100 data frames. The

CDC 6400 utilizes a 60-bit word so that the fifteen 12-bit words comprising a data frame were written onto digital tape in three 60-bit groups. Using software developed for the

CDC 6400, data tapes were ultimately rewritten so that data frames consisted of fifteen

60-bit words each with 12-bits right justified. When synchronization was lost, storage of the data ceased while the machine waited for a SYNC LEVEL true. During this period, the computer backed up three frames and began storing from that point because those frames contained faulty information. The flow diagram in Fig. 6 does not show the alternating buffers. This feature in itself did not increase complexity but rather the software became more involved because synchronization (sync) losses might have occurred just prior to the buffer transition. For example, if a sync loss appeared after one frame had been stored in buffer 10 (the 10th buffer to be filled), then the last two frames in buffer 9 would have to be erased in addition to the frame in buffer 10. In spite of the additional machine time required for this programming feature, the data transfer rate from analog tape to computer compatible tape was 50 kbits/s or half the original data transmission rate.

To make the data handling process easy to operate and provide monitoring and checking capabilities, a number of program features were added. Program control was achieved through two teletype commands which initiated or terminated data storage. When the termination command was typed, storage in the current buffer was completed, the buffer was written onto tape, and the last data word was followed by an end-of-file mark. Since these commands could be utilized at any time during the data transfer, the user had great flexibility in determining what segments of data were desired for analysis. The program also incorporated a display feature such that any two data channels could be selected through the accumulator switches, D/A converted, and displayed on an oscilloscope. Data

could be displayed whether or not the initiate-data-storage command had been entered. In this way, the data were scanned so that only those sections of interest were written onto digital tape. Computer costs were reduced since the volume of data was minimized. A final software package was written to facilitate the tape handling. This package also enabled checking the data transfer by playing back the data tapes and displaying channels on the oscilloscope.


One type of data loss occasionally encountered in the current program was due to the receiver becoming detuned when RF reflection patterns changed as the skier moved down the mountain. The duration of the losses ranged from .1 to .3 seconds and a typical loss record is shown in Fig. 7. A type of synchronization loss, which also occurred, was accountable to the Sanborn instrumentation recorder. The bi-phase waveform (serial PCM data), which was recorded at the 100 kHz bit rate, was degraded by the limited frequency response (3 db down @ 250 kHz) of the direct record-reproduce amplifiers. Losses appeared during data playback when the waveform degradation became too severe for the decoder to function properly. These data losses, which appear as sharp spikes in Fig. 7, were usually short in duration (~ .02 s) so that only 6 to 10 data frames were lost.

In the minicomputer process, data storage ceased when an information loss was encountered and began when data again became valid. This resulted in a data discontinuity or time base shift at the loss point. If the data sampling interval is )t, then for data with no information loss t(i + 1) = t(i) + )t (1) where the i represents data points on the digital tape and t is the time relative to the start of the data. In the case of an information loss of say P frames between i and i + 1 t(i + 1) = t(i) + P @ )t

This discontinuity is displayed in the 2 Hz sine wave of Fig. 8 with P = 50 frames or ~.1



Spectral analyses of data segments with discontinuities do not accurately give the frequency content of the original data. Figures 8a and 8b compare the power spectral density estimates for the signal waveforms shown and illustrate the inaccuracy for this particular example. These power spectrum estimates were obtained by using the Cooley-

Tukey Fast Fourier Transform (FFT) method (7) to compute Fourier coefficients F k

(3) k is the kth frequency component, N the total number of data points and A n

the data point values referenced to zero mean. The sampling rate for this example was 1024 points/s.

Data preprocessing included cosign tapering of the boxcar filter shape to achieve minimum leakage as recommended by Bendat and Piersol (8). Raw power spectrum estimates G k were formed by

(4) where the data time interval h is .976 ms. After scaling raw estimates for tapering effects, frequency smoothing was employed to reduce the normalized standard error. The dropout spectrum in Fig. 8b exhibits a 30% reduction in power at the 2 Hz estimate and a bandwidth increase from ~ 9 Hz to ~ 19 Hz at the 20 db power level. These result from the high frequency components in the discontinuity. Clearly, interpretation of such results may lead to erroneous conclusions and it is desirable to minimize these errors.

Errors resulting from data discontinuities were minimized here by reconstructing the original time base. The simplified flow chart in Fig. 9 shows a software technique coded in

Fortran which was used to rewrite data tapes on the CDC 6400. After a data physical record had been read by the CDC 6400, dropouts were monitored by checking for sequential counts in the least significant frame counter word. A result +1 caused buffered storage of a single frame with program control returning to check the next count. Counts different from +1, except for a counter turnover, originated an information loss. When a loss was encountered, the exact number of lost frames was determined from the word count difference. Lost data frames were reconstructed by storing pseudo data word values linearly interpolated from the verified data bounding the information loss. A record of the data loss was provided by replacing the most significant frame counter word (See Fig. 2) with a dropout flag. When data histories were plotted and analyzed, this separated the actual data and interpolated pseudo data. Data intervals with either a high dropout density or a long duration dropout were not analyzed. When the fixed length buffer was full, it was written onto an intermediate file. Control returned to the dropout checking program which processed and read more data as required. Program logic became complex because a buffer could become full during a dropout, in the middle of a physical record unit (PRU), between PRU’s, etc. when processing was completed, the intermediate file was written onto a new tape which was then used for data reduction.

It should be noted that when a dropout is longer than half the period of the lowest frequency component in a signal, then interpolating the pseudo-data points cannot enhance

the accuracy of the waveform and these data sections should not be analyzed. More generally, if a dropout is longer than half the period of any frequency component, that information is not retrievable and the error of the spectral density in this component may be large.


The data preprocessing system outlined features ease of operation, versatility, and flexibility. The system can be easily adapted to accommodate any length data frame or bit rate. Regarding the bit rate, the real-time data transfer is 50 kbits/s as discussed. Much higher telemetered bit rates can be handled by utilizing an analog recorder with compatible frequency response and reducing the data playback speed. The real-time data transfer capability is dependent primarily on the minicomputer cycle time rather than the writing capabilities of digital tape machines. Minicomputers such as the PDP-11 are available with cycle times at least 6 times greater than the Nova, so that the data transfer rate could be faster. Different data frame lengths will require various degrees of software modification depending on the types of computers used. The present system with equal sampling rates on all 16 channels, 15 words/frame, and 12-bit words fitting nicely into three 60-bit words, is probably the simplest case. For a lesser number of channels, the extra 12-bit words can be loaded with zero. This maintains simplicity of the tape format at the expense of decreased data density. A multichannel system which encodes data channels at varying rates would present unique problems in the interpolation process. Not only will the number of lost frames have to be computed, but also the interpolated points must be based on knowledge of the frame structure (sampling sequence).

Another note concerns the two frame counter words (see Fig. 2). In the restructured data frames, the most significant frame counter word was replaced with a dropout flag as discussed. It is, of course, not necessary to have an unused word in the transmitted data frame to eventually use for dropout identification. This flag word might be created by the software and incorporated as part of the data frames written on the final reconstructed data tape. For the application here, one 12-bit frame counter word is adequate because it turns over every 7.88s at the 520/s sampling rate. Information losses are shorter than this time so that one can account for all data frames. If the original time base is to be known, then the counter turnover must be long enough to include any duration loss possibly encountered.

A final point concerns when the data reconstructing is actually performed. Here, all reconstructing was done in a batch process on the CDC 6400 computer. While this method was workable in the present program, it augmented the cost of data analysis and degraded data handling efficiency. These undesirables could be eliminated by reconstructing data in real-time with the minicomputer. This would, of course, complicate the minicomputer

software and probably degrade the data transfer rate. Inasmuch as the technique of realtime reconstructing is currently under development, rates cannot be estimated as yet.


We are indebted to the United States National Science Foundation for financial support.

We thank Professor Forest S. Mozer and Mr. Henry Heetderks of the Space Science

Laboratory, University of California, Berkeley, for designing the PCM encoder and decoder. Finally, we thank Ms. Lesa Havert who very ably prepared the manuscript.


1. Mozer, F., Acting Head, Space Sciences Laboratory, University of California,

Berkeley, personal communication, 1976.

2. Vreeland, R. W., Rutkin, B. B., and Yeager, C. L., “Tuned Loop Receiving Antennas for Indoor Telemetry”, Biotelemetry III, Academic Press Inc., San Francisco, 1976, pp. 333-336.

3. Terman, F. E., Electronic and Radio Engineering, McGraw-Hill, New York, 1955, pp. 86-4 -934.

4. Neukomm, P. A., “Body Mounted Transmitting Antennas: Radiation Patterns and

Design of Helical Dipole Antennas”, Biotelemetry III, Academic Press Inc., San

Francisco, 1976, pp. 345-348.

5. Hull, M. L., and Mote, C. D., Jr., “PCM Telemetry in Ski Injury Research: I.

Instrumentation”, Biotelemetry, Vol. 1, No. 4, 1974, pp. 182-191.

6. Hull, M. L., and Mote, C. D., Jr., “Skiing Injuries: Field Loading and Analysis”,

Dept. of Mech. Engr., University of California, Berkeley, Nov. 1975, pp. 116.

7. Cooley, J. W., and Tukey, J. W., “An Algorithm for the Machine Calculation of

Complex Fourier Series”, Mathematics of Computation, Vol. 19, April 1965, pp. 297-


8. Bendat, J. S., and Piersol, A. G., Random Data: Analysis and Measurement

Procedures, Wiley - Interscience, New York, 1971, pp. 286-330.

Figure 1. Test Ski With Dynamometers, Anemometer, and

Telemetry Package Containing Transmit Electonics.

Figure 2. Telemetry Transmit and Receive Systems.

Figure 3. 12-Bit, 16-Word PCM

Figure 4. Data Handling System.

Figure 5. Control and Status Command Timing Showing Lock - loss.

Figure 6. Simplified Flow Diagram of Minicomputer Software.

Figure 7. Loading Histories With and Without Information Losses.

Figure 8.

Signal Enhancement for 2 Hz Sine Wave With 0.1s Dropout and Comparison of Power Spectra a) Actual Data; b) Minicomputer Processed Data; c) Reconstructed Data.

Figure 9.

Simplified Flow Diagram of Software to Enhance Data Waveform



O.J. “Jud” Strock

Senior Applications Engineer


Sarasota, Florida


A hardware Compressor examines measurement data prior to computer entry, discards redundant or otherwise uninteresting words, and passes the appropriate information with tags to a computer. Continuous rates of 100,000 to 1,000,000 words per second are accepted. Under some conditions, 95% to 98% of the data can be discarded while passing every measurement which is of value in real-time analysis.


In a real-time telemetry-computer system, the computer and associated storage media are generally the limiting factors in data throughput. This condition is brought about by the need to examine every data measurement, even though much of the data is not of immediate importance to a system user. Data examination in the computer under software control is a slow process, and after this time-consuming operation is completed on a given measurement most data is discarded as insignificant to the real-time analyst. If a measurement does not change meaningfully for several milliseconds (even several seconds), the use of the computer to examine each occurrence of the measurement imposes an unfair burden on the computer and contributes to overload and the related consequences. Even when throughput rate is not critical, compression is useful in reducing data storage requirements in a system.

For years, the technique of data compression has been considered by telemetry users, with the hope that some form of word-by-word data examination could take place prior to computer entry in order to discard uninteresting data and allow the computer to spend full time in analysis of important measurements. This is not to imply, of course, that discarded measurements are lost forever; generally a tape recording is made of all data, such that subsequent detailed analysis in relative leisure can take place.

Past proceedings of the ITC and other conferences during the past ten to fifteen years have studied the technique of data compression in great detail, and many studies have been

made to determine the algorithms which are most desirable in real-time telemetry data processing. Many of the mnemonics used in this paper are adopted from previous publications and studies. This paper is not a study of the technique of compression, but is the description of a unique piece of hardware which implements the compression technique by high-speed hardware preprocessing.

Responding to the requirements of a telemetry user several years ago, EMR developed a hardware Compressor and integrated it into a system in order to reduce the volume of data being entered into the computer. That Compressor was updated for a second user, and a third, and so on as the years went by. New requirements were incorporated as new systems were developed, and finally, a standard Compressor was defined and put into production, reflecting the composite requirements of several users. This Compressor is described on the following pages.

System Application

A typical system application of the Compressor is shown in Figure 1. Data measurements are input from up to six asynchronous sources (such as PCM, PAM, PDM, FM, or analog data streams) and merged with time data. Each measurement is examined in accordance with a uniquely specified algorithm (or algorithms), and is output to the specified ports.

Port 1 is usually a direct-memory access (DMA) computer input; Port 2 can go to the same or another computer, or to some other device such as an “array processor”; Port 3 is a special interface to computer memory or to a data display device.

The dynamic merge capability is especially important in many systems, where independently generated but functionally related data streams are to be entered into a computer.

Multiple output ports prove to be valuable also in many systems applications. The use of a hardware device to pre-sort data allows separate buffer areas in the computer to be allocated to separate data functions. Alternatively, two separate computers can be used with the Compressor pre-sorting data measurements for the computers as it merges, compresses, and distributes data words.

In the typical EMR system, a Digital Equipment Corporation (DEC) PDP-11 or VAX-11 computer is used, and data is entered into memory automatically via DEC’s UNIBUS as shown, using a special EMR direct-memory interface channel without the need for program intervention. The Compressor is set up and controlled via UNIBUS; individual data measurements are processed in accordance with instructions stored in core memory in the Compressor.

Description of Hardware Compressor

Measurements are input to this Compressor via Ports 1 through 6, along with IRIG time

(one millisecond resolution). When two or more measurements occur simultaneously, or when the momentary input rate exceeds the throughput rate of the algorithm processor, an input FIFO acts as a buffer.

Each measurement is assigned an interim identifier, consisting of the input port number and word number in the data frame (and location in the subframe or sub-subframe, if applicable), plus status bits. The identifier becomes a pointer to the 16-word algorithm processor which is unique to that measurement, and which contains all instructions and constants necessary to examine the data content each time the measurement occurs.

Figure 2 is a simplified flow chart which depicts the handling of measurements, and

Figure 3 is a simplified block diagram.

Four major modes are possible under program control:

• Normal compression -- each measurement is examined and passed or rejected according to its algorithm test(s).

• Limited-group compression -- high-priority processing takes place; measurements which are not in the high-priority group are rejected without test.

• Force-one-look operation -- one output is made of each measurement (or of certain selected measurements), whether or not it passes the algorithm test.

• Throughput -- all data is output without test (not shown in Flow Chart).

Data Testing

Each measurement value is tested in accordance with the appropriately selected algorithm or algorithms (see Figure 4 and Table 1). If it fails the test(s), it is rejected; if it passes one

(or more), it is output.

If the “status” bit in the algorithm process is set, a status word and tag are generated and output with the measurement. This word tells which algorithms were applied to the word, and why the data passed.

One bit in the algorithm process is an “alarm.” bit. If this is set for a given measurement, an alarm pulse is generated and output whenever the measurement passes the test(s). This can be used to trigger a computer interrupt, or to cause other action in the station.

Each passed measurement is accompanied by a “tag,” preassigned to identify the data to the computer. By selection of the tag content, programmers can make optimum use of this important tool.

Part of the algorithm processor information tells the Compressor where to send each measurement and tag (Port 1, 2, and/or 3). Ports 1 and 2 are conventional 16-bit outputs, while at Port 3 the data and tag are on separate sets of lines. Output formats are shown in

Table 2.

Special Capabilities

Incidentally, the Port 3 output can be used with a special computer interface to set up a

“most recent measurement” buffer in computer memory. Each tag contains the memory address (or relative address) where the accompanying measurement word will be stored; compressed data can then be accessed at known locations in memory.

A special “all-pass” capability at Port 3 enables the user to bypass the algorithm processor on specified words and output them at Port 3 (data and special “all-pass” tag on separate lines). In this operation, time and status are not output, nor is compiled data such as CSU or MMA. This port can be used with a special interface to operate a Word Selector, whereby each tag specifies the address of a device (analog output, discrete bit output, decimal display, or binary display) to which the accompanying data will be directed.

Time is merged with incoming data every millisecond (or every frame or subframe) in the typical system. Minor time (milliseconds) is input with a unique tag; every second (or subframe, etc.) major time (seconds/minutes and hours/days) is input.


In summary, capabilities of the Compressor are:


(12 algorithm processes - 8 other logical tests)

• Up to 3800 separate input measurements examined by individually-specified algorithms.

• Up to 6 input ports, merged with time port.

• Two output ports for entry into a 16-bit computer interface.

• Third output port for special 32-bit applications.

• High-speed inputs (burst rates of 1,600,000 words/second).

• Built-in data simulation capability for self test.

• Group addition or deletion by single command.

• Force-one-look capability for confidence checks.

• Dynamic updating capability from computer.

• Status word generated for output with data if designated.

• Alarm output when designated words pass.

• Memory load-verify from computer.

Table 1. Algorithm Processes





BMA or


Reject without test.

Throughput (pass) without test.

Bit Match or No Bit Match

• Test the designated bits. Pass the word if there is: (a) Bit Match, or (b) No

Bit Match.







ILI or


• Constants to be put into memory for each measurement: (a) Bits to be tested, and (b) Pattern to be matched by these bits.

In Limits or Out of Limits 4.6



• Test the data value to see if it is within the designated limits. Pass the word if it is: (a) In Limits, or (b) Out of Limits.

• Constants to be put into memory for each measurement: (a) Upper Limit, and (b) Lower Limit.

Bit Change

• Compare the word with its most recent value on a bit by bit basis. Pass the word if there has been a Bit Change in the bits of interest.




• Constant to be put into memory for each measurement: Bits to be examined.


• Compare the data value with its last output value. Pass the word if the value differs from the last output value by more than a specified delta (change).





• Constant to be put into memory for each measurement: Allowable delta.

Delta Slope

• Compare the data value with its last output value to determine the sign

(direction) of the slope. Pass the word if the sign has changed since the measurement has last passed, and the level has changed by more than a specified delta.

• Constant to be put into memory for each measurement: Allowable delta.


• Discard “N-1” Sequential occurrences of the measurement, then pass the measurement once.

• Constant to be put into memory for each measurement: “N.”





MMA Maximum-Minimum 7.4



• For a period of “N” occurrences of this measurement, store the accumulated

Minimum and Maximum values. Output these extremes, reset, and repeat.

• Constant to be put into memory for each measurement: “N.”

Cumulative Sum

• For a period of “N” occurrences of this measurement, accumulate the sum of the data values. Output this Cumulative Sum as a double-precision (32bit) word, reset, and repeat.

• Constant to be put into memory for each measurement: “N.”



Chained algorithms require less than the sums of their individual processing times. Words in a subframe require 0.84 microseconds longer for processing; words in a sub-subframe require an additional 1.68 microseconds.

Table 2. Data Output Formats

Passed data is output as:

• Tag (16 bits)

• Data (16 bits)

Where Status has been designated as an output, it precedes the data as follows:

• Status ID (16 bits)

• Status (16 bits)

In the algorithm tests which involve more than 16 bits of data output (MMA and CSU), the format is:

• Tag (16 bits)

• Maximum value (MMA) or least-significant 16 bits of sum (CSU)

• Tag (16 bits)

• Minimum value (MMA) or most-significant 16 bits of sum (CSU)

When data relating to one of these two algorithms (MMA or CSU) is “forced” out, it does not represent the true conditions which are defined for the algorithm; therefore, a special output is generated:

• Tag

• Number of samples (instead of “N”) followed by the tags and data words (above)

When the “DSL” algorithm (Delta Slope) causes an output, the most-significant bit in the data word is the sign of the slope (O = negative, 1 = positive).

Every millisecond or frame interval (if data has been output within the past millisecond), a minor time word is output:

• Time ID (16 bits)

• Milliseconds (16 bits)

Every second or subframe interval, major time is output:

• Time ID (16 bits)

• Seconds, minutes (16 bits)

• Time ID (16 bits)

• Hours, days, or ID (16 bits)

Figure 1. Block Diagram

Figure 2. Simplified Flow Chart

Figure 3. Data Compression Subsystem

Figure 4A. Algorithms

Figure 4B. Algorithms Figure 4C. Algorithms





Asok K. Mukhopadhyay* and

Boris M. Dobrotin

Members of the Technical Staff

Information Systems Division

Jet Propulsion Laboratory

Pasadena, CA


The purpose of the multi-mode evasive vehicle, as defined by the U.S. Army Armored

Vehicle Command, is to provide an automatically steered moving platform for a target silhouette which will be used for tank gunnery evaluation. The vehicle, as well as its guidance and control systems, must be rugged enough to withstand battle conditions, rough terrain, hostile weather conditions, and the impact of dummy shells on its superstructure.

The vehicle’s steering and speed control are accomplished remotely by signals sent via a guide wire laid on the ground. The Army Wire-Following-Vehicle (WFV) represents new technology in the area of automatic vehicle guidance and has a host of potential applications, both military (such as a target drone) and civilian (such as underground mining and rescue as well as surface mass transit on electronic guide ways).

For an high-order non-linear system, such as the WFV with its guidance and speed control subsystems, the final design must be generated and verified by non-linear time-response simulation. This paper describes the time-response simulation studies undertaken by the authors in support of the development and validation of the guidance/steering subsystem of the WFV (including the vehicular motion on some sample guided courses). Such simulation is the cheapest and quickest way to determine the design tolerances and hence, the field worthiness of the WFV. The WFV guidance system hardware design based on the

The research described in this paper was carried out at the Jet Propulsion Laboratory, California

Institute of Technology, and was sponsored by the Department of Defense through an agreement with NASA.

*Presently with Rockwell International Corporation, Seal Beach, Calif.

continuous time-response simulation reported in this study has resulted in successful accomplishment of desired performance goals.


The WFV target was developed under the auspices of the U.S. Army Armor and Engineer

Board, Ft. Knox, KY, with the intended use of tank fire control system evaluation. In order to increase the realistic effects of the target, the target was not to be artificially protected

(by earth embankments, for example) and the tank under evaluation was to be allowed to fire on the target using inert shells.

The chassis and power plant were constructed by the U.S. Army Armor and Engineer

Board with the goal of combining high maneuverability and low cost. With this goal surplus components were selected: the chassis was an AMC M175 1-1/4 ton truck and the power plant was a Ford V8 engine. The primary modifications consisted of removing the body, transmission, and front wheel drive, and installing the V8 engine and automatic transmission. Wide tires were added to improve surface cohesion. A separate contractor

(Wood Ivy, Inc.) installed the electronic logic, including the speed control, and initially modified the chassis for remote control. The completed baseline vehicle undergoing test at

Edwards AFB is shown in Fig. 1.

The objective of JPL’s development program was to provide a guidance system which would automatically control the vehicle’s path and velocity over a prelaid course of

10 miles maximum. In addition, the guidance system’s performance was to allow the vehicle to reproduce typical tank attack maneuvers. Two such basic maneuvers are a high speed (>50 mph) straight line dash and a high lateral acceleration (>0.5g) sinusoidal maneuver. These two maneuvers were used to develop the following three performance specifications:

• Straight line stability at speeds up to 50 mph.

• A lateral acceleration (“cornering”) capability established by 1/2 of the cohesion limited acceleration up to a maximum of 0.5g.

• A speed control capability of ±10% of the commanded speed.

These specifications represented an order of magnitude increase in the cornering capability of automatically guided vehicles. Most previous systems (Ref. 1 and 2) have emphasized highway type operations where lateral accelerations were limited to 0.05g. Banked turns were assumed for high speeds, and maneuvering capabilities were minimized. In addition, a single forward mounted antenna was typically used as a sensor. Ito (Ref. 2) used a

continuous, relative strength antenna, while the Ohio State University (Ref. 1) developed a quantized antenna which was independent of field strength. All antennas were limted to a lateral displacement of ~1/2 ft., while a displacement capability of ±3 ft. was needed for the WFV.

The results of a time response simulation of the lateral guidance system performance is presented below. For a complete description of the WFV, including the speed control system, see Reference 3.


A block diagram of the overall lateral guidance system (hereinafter referred to as the guidance system), is shown in Fig. 2. The vehicle’s geometrical relationships are shown in

Fig. 3 and pertinent vehicle parameters are shown in Table I.

The guidance system consists of the following: a guideway determined by a manually positioned wire and energized by a 6 kHz sine wave signal, front and rear position sensing antennas, the signal processing electronics, the steering servo, and the vehicle steering components (linkages, wheels, etc.)

The antennas are based on the design developed by Ohio State Univ. Physically the front and rear antennas are identical and consist of 16 sensing elements mounted over a 6 ft.

span. The output of these antennas provide discrete position (staircase) information at

~4 1/2 in. intervals over a 6 ft. range.

The processing electronic block diagram is shown in Fig. 4. The front antenna error signal

(E) is fed through a proportional plus integral network and then divided by velocity. The difference between the front and rear antennas error signals (R) is fed through a proportional gain. These two error signals are then summed and divided by velocity again to produce the steering command (*).

The steering servo is a closed loop, electromechanical system acting directly on the vehicle’s steering linkage. A modified automotive power steering pump was used to supply hydraulic pressure to the servo valve, which in turn controlled a double acting hydraulic cylinder. In order to assure good response and to minimize inertia, the hydraulic cylinder was coupled directly to the steering linkage and the steering wheel gear linkage was disconnected. Servo feedback was via a rotary potientiometer connected to the steering linkage. Experimental results indicated a steering servo response of ~2 Hz.

Modification to the WFV’s suspension were made to increase the coefficient of friction and to separate the roll and pitch/yaw frequencies. Wide tread tires were installed and

“stiffer” shock absorbers were used. The original front wheel drive axle was retained, giving high steering mechanism inertias and mechanical hysteresis. Ballast was used to give the WFV neutral steering characteristics.

Table I. Wire Following Vehicle Parameters

Vehicle Weight


4000 lb.

10.7 ft.

Distance from c.g. to front antenna - 8.8 ft.

Distance from c.g. to rear antenna - 9.4 ft.

Tread Width - 6.5 ft.

Moment of Inertia about vertical axis - 3920 slug/ft

Front tire cornering force - 10 x 10 3 lb/rad.



The generalized block diagram for the guidance of a WFV is shown in Figure 4. A detailed discussion of the various feedback loops and functional blocks of Figure 4 as well as a set of 3 degree-of-freedom (DOF) equations of motion representing a moving vehicle dynamics are contained in Reference 4. Due to the modifications made in the vehicle chassis, shocks and tires, for all practical purposes, the WFV roll-steer coupling was absent; hence, in this analysis, degenerate 2-DOF equations of motion (based on the

3-DOF model contained in Reference 4) were used to describe the WFV lateral-directional dynamics. These equations assumed a stiff suspension with small roll motions, and linearized tire properties operating on a paved surface. Some of the basic simulated vehicle dynamic characteristics were plotted in Reference 3 as a function of the operating speed of the WFV. These show that:

• As speed increases, the steady-state yaw velocity gain of the vehicle (i.e., the amplitude ratio of yaw velocity to wheel deflection or steering angle) increased roughly in proportion to speed.

• The WFV had very nearly “neutral” steering properties up to 50 mph, implying that

WFV yaw rate is closely proportional to both steering (front wheel deflection) angle and speed. This fortuitous property permits the simplified guidance design given in Figure 4.

• As speed increased, the yaw response of the vehicle became slower (i.e., the phase lag increased).

• The phase lag due to the steering servo actuator was relatively large and represented most of the net WFV phase lag at low speeds.

The preceding response properties are fundamental in defining guidance system requirements for the WFV. The performance of the closed-loop WV was limited by the electromechanical servo/actuator characteristics as well as the vehicle dynamics. Also, it was evident from the above that some form of speed compensation was essential to achieve reasonably consistent performance as speed varied. Selection of the appropriate gain values is based on a compromise between resultant system response bandwidth (i.e., rapidity) and damping (i.e., tendency to exhibit oscillations and overshoot). The sensitivity to inadvertent changes in system gain (e.g., due to tires suddenly operating on a soft or slippery surface, or in accuracies in vehicle physical parameters) must also be considered in gain selection. In order to satisfy the guidance equation of Figure 4 and maintain a reasonable operating signal level for a wide dynamic gain range (gain varies inversely to the square of velocity), the block diagram of Figure 5 was generated in which the gains shown in Figure 4 can be redistributed at selected discrete speeds to produce the desired results.

The 2-DOF vehicle equations of motion for the WFV simulation contain the following simplifying assumptions:

• Fixed forward steady-state velocity, no longitudinal dynamics.

• Vehicle inertial symmetry about the vertical plane containing the longitudinal axis of the WFV.

• Sprung body rolling about a kinematic roll axis; unsprung and rigid wheels/tires remain in contact with road and surface.

• Small kinematic angles, i.e., sin X = X, cos X = 1, for heading and roll angle displacements and velocities.

• Negligible changes in tire (side force/slip angle) and suspension (steer/roll angle) properties.

These assumptions are consistent with the preliminary nature of this analysis of WFV dynamics, particularly at the lower lateral acceleration levels (less than 0.3g). The resultant equations of motion are linear, and these have been found to be representative even at higher lateral acceleration levels, for nominal vehicle/roadway conditions.

Simulated step input time responses for the refined WFV are shown in Figure 6 (a-d) and

Figure 7 based on estimates and measures of the vehicle properties as given in

Reference 4. The JPL-designed antennas are quantized (i.e., they yield a “staircase” voltage output for linearly increasing input, viz., lateral displacement with respect to the wire). However, for purposes of initial analyses they are considered as linear elements.

The servo actuator (valve, hydraulic cylinder and linkage) likewise, can have its own internal dynamics, nonlinearities, etc.; for these analyses the servo was represented by its net effective closed-loop properties (e.g., transfer function), experimentally determined.

The WFV motion variables plotted for two different WFV speeds (6.8 mph and 25 mph) in

Figures 6 and 7 are front and rear axle center displacement error with respect to the wire

(a), WFV relative heading error (deg) with respect to the wire (b) , input voltage to the servo valve (c) and finally the average front wheel angle (d). The response at two different

WFV speeds (Figure 6 for Vo = 6.8 mph and Figure 7 for Vo = 25 mph) are shown. The input in each case was a small (1 ft) abrupt displacement of the wire under the front antenna.

The step responses of Figures 6(a) and 7 show that for Kc =1 (low overall loop gain), the

WFV has a slow and underdamped lateral position response. Note the large settling times

(about 10 secs) in each of the quantities plotted in Figures 6 and Figure 7, although the magnitude of the deviations are small enough to keep the wire placed in between the wheels of the WFV and hence, within the range of the front and rear bumper mounted antenna. This large settling time is primarily due to the mechanical components of the vehicle and is principally responsible for the “poor” performance of the vehicle when following a “piecewise” continuous curve consisting of linear segments, rather than a curve with continuous rate of change of curvature. It is obvious from the block diagram of

Figure 4 as well as the time response simulation data, that the angular direction sensing loop of the overall servo contributes less than 10% of the overall error-correcting voltage to the steering servo. The reason behind this is strictly analytical and derives from the fact that the vehicle has a long wheel base (18.4 ft) and large lateral deviations of the wheel are required to generate sufficient angular deviation.

Due to appropriate gain switching in the block diagram shown in Figure 5, such response appears to improve at higher speeds (6.8 mph in Figure 6 vs. 25 mph in Figure 7). For both the preceding speeds, Figures 6(a) and 7, the transport lag between the front and aft antenna caused by the motion of the vehicle is quite evident (note the behavior in

Figure 6(a-d) with the response around 1.8 secs). The heading angle response

(Figures 6(b)) appears relatively small and in some circumstances may even be less than the one “quanta” of the segmented antenna. In the segmented antenna case, it appears the

WFV will limit cycle at amplitudes corresponding to one or two quanta and guidance loop stability would be adversely affected for small amplitude excursions. Reference 4 shows

response curves when the overall Kc (shown in Figure 5) is doubled. In this case, the rapidity of the step response is increased, and the damping is not appreciably reduced, indicating improved overall response quality. As is indicated in Figure 5, the trim integrator is shorted out at under 15 mph (K


= 0) to eliminate WFV “hunting” caused by the integrator pole at the origin. Thus, the time responses of Figure 6 does not include the integrator feedback while that of Figure 7 reflect the effect of the integrator. Based on the preceding comparison, one is led to speculate about the usefulness of the integrator feedback loop. Indeed there is no particular advantage of the integrator for straight-line course following, but the integrator is needed to eliminate the steady-state “hang-off” error that would otherwise ,be present when the WFV tries to follow a constant radius circle.

Even for a very gradual 0.05g lateral acceleration curve, the steady-state value of the lateral displacement error, (E ss

) is 2 ft; and for the target turn of 0.5g, the indicated “hangoff” error would be about 20 ft. which is unacceptable.


The analytical results included the effects of the transition from a straight line (radius = 4) to a curved segment (radius = constant, K) and provide a gradually changing radius (i.e., spiral, 4 > r > K). The actual wire course, laid out on a dry lake bed at Edwards AFB,

Rosamond, CA was intended to duplicate an operational course and was laid out manually.

Straight lines were sighted in visually, and curves were marked with a rope compass and then approximated with straight line segments. This radius of curvature varied between 4 and zero at the connecting points of the straight segments.

The analytical effort provided a linear approximation to a very non-linear system, with very good results. Predictions of system and loop gains were within 10% as given in

Ref. 3. While the gross performance (i.e., the maximum bounds) was predicted by the linear analysis, the actual response of the vehicle was perturbed by the non-linear elements. Instrumentation data consisted of strip chart recordings which provided only approximate performance results.

It appears that the primary reason for correlation between the analysis and the experimental results was due to the vehicle inertia and steering response. That is, the vehicle could not respond to the antenna measurement non-linearities, but instead established a 1 Hz limit cycle (the expected natural frequency) as a result of these nonlinearities. Curve tracking proved to be a problem. On a “constant radius” curve as described previously, the change of direction between line segments represents a large step input into the steering servo, and actually created peak lateral accelerations much higher than the average acceleration. During the sinusoid lateral acceleration phase, peak acceleration was limited by the effect of the integrator. While the integrator was needed to maintain a constant steering angle (*) with minimal offset, it also required a small amount

of time to null the offset integral when the direction (i.e., left to right) of the curve was reversed and a change in was needed. In other words, the vehicle would overshoot the curve reversal until the integrator “caught up”.


Satisfactory performance was obtained. That is, the vehicle remained controllable at straight line speeds of 50 mph as well as lateral accelerations which reached the limitation imposed by 1/2 the available coefficient of friction. The vehicle remained within ± 3 ft of the reference wire. This indicates that the simplified analysis described above is valid in predicting performance of the VFW.

Though the time response simulation presented accurately predicted the performance of the vehicle, further analysis and simulation are needed to provide detailed information about the WFV’s performance during actual operation; viz, on entering or leaving corners, the effect of curve discontinuities, etc. In addition, non-linearities introduced by the actual hardware should be simulated. However, this initial 2-DOF simulation has indicated that it is capable of producing useful and accurate results and can be used to evaluate changes in the vehicles configuration.


1. Fenton, R.E., et al., “On the Steering of Automated Vehicles: Theory and

Experiment,” IEEE Trans. Vol. AC-21, #3, June 1976, PP. 306-315.

2. Ito, T., et al., “An Automatic Driving System of Automobiles by Guidance Cables,”

SAE Paper No. 739127, presented at International Automative Engineering Congress,

Detroit, Michigan, January 8-12, 1973.

3. “Phase I Report, U.S. Army Remotely-Controlled Vehicle Analysis,” Jet Propulsion

Laboratory, Pasadena, Calif.; May 1, 1979.

4. “Dynamics Study of Wire Following Vehicle,” Systems Technology, Inc.,

Hawthorne, Calif., STI Tech. Rep. No. 2089-1, January 1979.

Figure 1. US Army/JPL Wire Following Vehicle

Figure 2. Guidance Loop Block Diagram

Figure 3. Wire/Vehicle Kinematics

Figure 4. Guidance Loop Simulation Block Diagram

Figured 5. Gain Allocation in Processing Electronics

Figure 6. System Response at 6.8 mph

Figure 7. System Response at 25 mph



R. L. Seeley *

F. M. Long **

J. D. Pauley ***

R. W. Weeks **

* Naval Ocean Systems Center, San Diego, California

**Department of Electrical Engineering, University of,Wyoming, Laramie, Wyoming

***Department of Psychiatry, University of Colorado Medical Center, Denver, Colorado


Study of electroencephalograms (EEG) under normal behavior conditions required the development of a small, reliable telemetry system. Here two hybrid ceramic packages were attached face to face to provide a hermetically sealed seven channel telemetry transmitter with glass to metal seals around the seven differential pairs of input leads and the power supply leads. The transmitter’s antenna is enclosed in the package by using two loops of gold substrate etched in a pattern around the other circuitry. The package measures .8 x

2.2 x 2.4 cm and weighs 8.5 gms.

Input noise level is below 1 microvolt (rms) and dynamic range is from 1 to 250 microvolts

(rms) with a frequency response (6 dB down) of 1 Hz to 150 Hz. Power requirements are

2.1 to 3.6 ma at 2.0 and 3.3 vdc., respectively, with at least 80% (3.3 and 9.5 mw, respectively) going to the radio frequency stage. Data are time multiplexed for pulse position modulation of an 88 to 108 MHz carrier. Maximum measured range of transmission with a 3 volt battery has been 10 m in air.

This transmitter is well suited for the study of any animal large enough to carry the package and a battery. Other biopotentials such as EMG and ECG can be telemetered by increasing multiplexor rates and/or attenuating input signal levels.


Since 1950 a great deal of engineering time has been devoted to the development of multichannel biotelemetry systems. After developing low power consumption stages for preamplification and transmission, engineers then tried to multiplex several signals

(representing different biopotentials) into one data stream for transmission on a single radio frequency (RF). Then only one RF receiver/decoder was needed to demodulate and

demultiplex the data stream into signals representative of the biopotentials originally sensed by the preamplifiers (preamps). in the early 60’s many modulation/demodulation schemes were used and described in the literature (references 1-8). By the late 60’s enough integrated circuits were available so that one could implement nearly any of these designs with flat packs soldered to a printed circuit and supplemented with a few discrete components, usually in the input (low noise) and output (high frequency) stages. The 70’s have been spent in finding and solving the problems which have reduced the utility of these designs (reference 9). Except for describing a better technique for mechanically connecting the leads between the sensors and the preamps, this paper deals with fabrication techniques or circuit design modifications that improve the reliability of multi-channel biotelemetry transmitters.


The criteria considered in this system’s development included (in approximate order of priority) minimum life of six months, seven differential channels with input sensitivity of

1 microvolt (rms), frequency response between 1 Hz and 150 Hz, at least 46 dB of dynamic range (1 to 200 microvolts, rms) power requirements of 3 milliamps or less at

3.0 vdc, transmission frequency in the FM band (88 to 108 MHz), electronics to be protected inside a hermetically sealed case, leads to penetrate the case with glass to metal headers, transmission antenna to be sealed inside a hermetically protected case, dimensions less than 1 x 3 x 3 cm, and weight less than 10 gm. In addition, a technique was needed to stabilize RF transmission against frequency shifts caused by changes in the

RF loading surrounding the transmitter. In short, we planned to combine several circuits found to be successful when built with discrete componenets and integrated circuits on printed circuit boards into a single hybrid circuit mounted in a hermetically sealed ceramic package which also contained an RF stabilization circuit and a transmitting antenna.


As regards the life of the unit, a hermetically sealed battery pack had already been developed using two parallel connected lithium/thionyl chloride cells, two dropping/interconnect diodes, and a bistable reed switch (Figure 1). It was desired that at any time during the “life” of the transmitter, data could be transmitted by activating the switch so long as the cells had not been expended. Although the cells are advertised to have a 1000 maH of life, our tests show that a fresh cell has 1440 maH between 3.59 and

3.58 vdc, and 360 maH between 3.58 and 3.34 vdc when discharged at a 3 ma rate

(Figure 2). Because one cell lasts 600 hours (from 3.6 down to 3.3 vdc), two cells in parallel with series diodes (Figure 1) are expected to provide 3.0 to 2.7 vdc (the battery range for which 3.0 to 2.9 ma are needed) for 1200 hours, or 150 eight hour days. Based

upon normal work hours, week, etc., this would be more than six months. Therefore, the shelf “life” implied above is about the same as the life of the battery pack used.

Note that the diodes in the battery pack serve two purposes. First, they permit paralleling two cells so that the power pack will supply the voltage produced by the highest voltage cell until it is drained down to the voltage of the other cell, at which time they will share the load. This also provides redundancy should one cell fail. Second, in the event that a direct short develops external to the hermetically sealed battery pack, silicon diodes

(.6 vdc drop) will act as either fuses or dropping resistors to prevent a direct short from occuring across the LiSOCl cells. Although the manufacturer of these cells has assured us that the AA size cell will not explode (as have the larger D size cells) when shorted, the diodes give added insurance against overheating.

The frequency response and dynamic range were established by the signal to be measured,

EEG in the range of 1 to 200 microvolts (rms). Power drain and transmission frequency were set according to what had worked well in the past to obtain our desired transmission range of 10m in air.

The innovative aspect of this transmitter was the design of an RF antenna as part of the etched circuit inside the ceramic package and a radio frequency stabilization circuit to compensate for RF loading variations in the environment external to the ceramic package.

The antenna was fashioned from two complete loops around the perimeter of the circuit etched from the gold substrate used for printing a layout inside the ceramic package.

Standard size ceramic hybrid packages, offering glass to metal seals around all penetrations, were used because both coated (with metal) and uncoated mating covers were available. The coated cover was used as a bottom lid for the package and served as a backplane for the transmitter, whereas the uncoated cover was used as a top lid, offering a transparent window to RF transmission. The frequency stabilization circuit was an outgrowth of work done by Weeks, Long and Pauley (reference 10). It uses a frequency doubler/buffer stage and electronic servo loop to maintain constant current in the collector of the output RF transistor. In addition to RF frequency stabilization, this circuit offers the advantage of being able to predict power consumption more accurately than in RF stages that permit current variations to maintain a constant voltage across a varying load.


The “chip and wire” fabrication method (references 11 and 12) was selected as the most economical way to improve upon the reliability of the printed circuit-solder techniques previously used. Pulse position modulation was used because it was compatible with the decoding equipment used for older units. With careful layout, it was possible to include seven differential preamps, pre-multiplexor amplifiers, a multiplexor (mplxr), a post-

multiplexor amplifier, a sub-carrier oscillator (SCO), and an RF stage in two packages measuring .4 x 2.2 x 2.4 cm each. The seven pairs of input leads penetrate the same package on the right side before re-entering the left side of the mplxr/RF package via pins that exactly match when packages are mounted face to face. The leads penetrating the right side of the mplxr package were not used and so were cut off close to the package. Power enters the preamp package from the right and the mplxr from the left, via wires which again mate exactly when the packages are glued together with Humiseal (1) under vacuum.


This circuit (Figure 3) requires 76 resistors, 53 capacitors, 4 integrated circuits, and

36 transistors (three emitter-base junctions of which are used as diodes). The preamp stages (enclosed by “.....” in Figure 3) consist of differential-pair transistors (T1, T2) coupled to the input leads by .47 uFd capacitors and very lightly pulled up (for low noise) to ½ of the power supply voltage by 1 megohm resistors. All inputs have 1000 pFd shunts to ground to eliminate any possible RF feedback. Signals leaving the differential stages are low-pass filtered (at 150 Hz) by the 1000 pFd capacitors, buffered by emitter followers

(T3) and capacitively coupled into other (buffering) amplifier stages which are direct coupled to the mplxr. Voltage out of the mplxr is amplified and DC restored to approximately 1/2 the power supply voltage (1st op amp of IC2) before driving current through a 220 K resistor to charge a 1000 pFd capacitor in an integrator (2nd op amp of

IC3). Voltage occuring at pin 2 which is above 1/2 the power supply voltage will cause voltages at pin 11 and (via a 220K-470K voltage divider) pin 5 to approach ground until the voltage at pin 5 drops below 1/2 the power supply voltage and trips a comparator (3rd op amp of IC2). As the comparator’s output (pin 9) is driven from the positive power supply rail toward ground, positive feedback through the voltage divider accelerates the tripping action and diode D1 conducts to discharge the 1000 pFd capacitor. The capacitor is discharged until the voltage at pin 5 (divided between pin 11 and pin 9 via the 220K-

470K resistors) returns to 1/2 the power supply voltage, at which time the comparator trips in the opposite direction.

The 35 microsecond pulse resulting at pin 9 drives transistor T5 into cutoff and increments the ring counter (IC3). Because the counter is stepped on the back side of this pulse, a similar transistor (T6) will generate a pulse for every 8 incrementing pulses sent to IC3.

This serves to produce equal width pulses every time the mplxr selects a successive channel and an extra pulse every time the mplxr starts charging the capacitor from channel 1. Because the time between pulses is proportional to the voltage sensed at each input channel, this SCO appears to be dependent upon signal in order to measure signal. In fact current supplied through the 220K resistor from the power supply to the 1000 pFd capacitor is sufficient ot maintain oscillation independent of input voltage. This is

necessary to prevent the erratic SCO operation obtained from sensing over-voltages such as occur when sensor leads break.

The normal and extra pulses are combined via 1.5 m resistors and used to modulate an RF carrier in the RF stage (enclosed by “-----------” in Figure 3). The RF stage consists of a

Colpitts oscillator (T7) tuned to 50 MHz and transistor (T8) buffered to drive an output tank circuit tuned to twice the oscillating frequency, causing 100 MHz to be radiated.

Current through T8 is converted to a voltage (by the 180 ohm resistor) and fed back via T9 to control the drive from the oscillator. The number of turns and turns ratio of the output transformer were adjusted for peak output (at 100 MHz) into the impedance offered by the two loops of gold substrate used as an antenna inside the package. The miniaturization of this portion of the transmitter has solved several problems at once. First, the near proximity of parts permits the design of transmission characteristics which used to be dominated by random interlead parasitic capacitances; instead of trimming capacitors or bending the antenna for best signal amplitude, frequency, distortion, etc., one can now design tuned tank circuits with sufficient accuracy to match fixed size (impedance) antennas. Second, the small investment in chip real estate required by a few extra parts offers important possibilities such as buffering the oscillator from the transmitting tank and employing frequency doubling techniques. Although these methods have been used in larger transmitter systems, they were usually not used in biotelemetry in an effort to conserve space.


The primary problems encountered in thise development appear to have been those of fabrication technique. The conductive epoxy first used (2) was either too old or of a poor composition to the extent that intermittent opens occurred around capacitors after several thermal cyclings of the prototype transmitters. The selection of a different epoxy (3) and the use of an automatic dispenser seem to have solved this problem.

Major circuit redesign was required because large capacitors (18 uFd) require too much room to be included in this package. As a result, a d.c. biasing network had to be developed for the post-mplxr amplifier which previously had been a.c. coupled. The voltage divider (470K-120K-D2) provides the correct d.c. bias to pin 12 of IC2, but we do not consider the inclusion of a trim resistor (R trim) as good practice in any hybrid implementation .... only an expedient solution to the problem.


Prototype transmitters have operated after soaking six months in a saline solution and more time will be required to determine minimum life of the transmitters. but all other design

criteria were met or exceeded. Although all circuits are functional over a wide power supply voltage range (2.0 to 3.3 vdc), high signal gain (2000) and dynamic range (1 to 250 microvolts, rms) dictate that the post-mplxr amplifier output have sufficient power supply voltage (2-6 vdc) to provide output signals 1.4 volts in amplitude (2000 x .00025 = .5 vrms

= 1.4 vpp).

(1) Type 1B66 sealant is made by Humiseal division of Columbia Tech. Corp., New


(2) Uniset C409 Silver filled, one component epoxy by AMICON.

(3) Ablebond 20-1 Silver filled, microelectronics grade. one component epoxy.

The input noise is below our lower limit of measurement at 1 microvolt, rms. The package is smaller (64%) and lighter (15%) than originally required. It contains an internal fixed shape antenna (etched from the gold substrate) and the RF oscillator buffering and current control stabilizes the transmission frequency against variation in loading on the RF output circuit. Those improvements permit the continuous transmission of data on a given frequency under RF loads which shift the carrier frequency (100 NHz) of older transmitters (without buffering or constant current) as much as 2 NHz.

Possible modifications to the circuit include the reduction of input sensitivity or the broadening of frequency response in order to transmit larger biopotentials such as ECG or

EMG. Resistive attenuation works well for amplitude reduction, and higher frequency response is available by decreasing (or eliminating) the 1000 pFd capacitors on the base of

T3 transistors and by decreasing the size of the integration capacitor (1000 pFd) to obtain higher multiplexor rates. The present average sampling rate of 1000 samples per second

(sps) for each channel could be increased to 2000 sps with an associated factor of 2 loss in dynamic range. Paralleling the inputs of two channels separated by three intervening channels provides a sampling rate of 4000 sps which would be suitable for most ENIG data requirements.


This project was supported in part by Independent Research funding at the Naval Ocean

Systems Center. The authors thank Drs. E. E. Kuhns, E. P. Cooper, and S. H. Ridgway for their continuing support in this effort and Drs. S. H. Ridgway and C. S. Johnson for their comments and suggestions.


1. Beenken, 11. G., and F. L. Dunn, Short distance telemetry of physiological information, IRE Trans. Med. Electr. PGME 12, 1958, pp. 53-57.

2. Kofos, A., Multi-channel radio transmission of bioelectric data, Proc. 3rd Int. Conf.

Med. Electron. 1960, 1:54-55.

3. Ellis, D. and F. Schneidermeyer, Multipack a three-channel system of physiologic/psychophysiologic instrumentation, Amer. J. Med. Electronics. 1962,


4. Hambrecht, F. T., P. D. Donahue and R. Melzack,A multichannel electroencephalogram telemetering-system, Flectroenceph. Clin. Neurophysiol. 1963,

15 (2):323-326.

5. Marko, A. R. and M. A. McLennan, A seven channel personal telemetry system using pulse duration modulation, Proc. 16th Ann. Conf. Med. Biol. Engng. 1963, 5:154.

6. Wehner, A. P. and F. J. Hahn, Implantation of a miniature multichannel physiological data acquisition system in a canine, 5th Int. Cong. Med. Electron., Liege, Abstr.

1963, p. 27.

7. Adey, W. R. , J. Hanley, R. T. Kado and J. R. Zweizig, Design considerations in multichannel telemetry, Proc. 19th Ann. Conf. Eng. Mod. Biol. 1966.

8. Roy, O. Z. and J. S. Hart, A multichannel transmitter for the physiological study of birds in flight, Med. Biol. Engng. 1960, 4:457-466.

9. Reite, M. and J. D. Pauley, Multi-channel implantable biotclemetrv, problems, pitfalls and rewards, Proc. 3rd Int. Symp. Biotclem. 1976, pp. 59-62.

10. Pauloy, J. D., M. Reite, R. W. Weeks and F. M. Long, A hybrid multichannel implantable biotelemetry transmitter, Proc. 31st Ann. Conf. Engng. Med. Biol. 1978, pp. 132.

11. Fryer, T. B., The use of hybrid integrated circuit techniques in biotelemetry applications, Biotelometry 1977, pp. 193-216.

12. Long, F. M. , Microcircuits - what they are, what they will do, Proc. 1st Inst. Conf.

Wildlife Telem. 1977, pg. 1.

Figure 1. 1 200 hour battery pack.

Figure 2. AA size GTE LiSOCl


cell discharge curve under 3 ma load.

Figure 3. Seven channel hybrid telemetry transmitter.


Kenneth O. Schoeck

Space & Missile Test Center

Vandenberg AFB, California

George M. Kobylecky

Federal Electric Corporation

Vandenberg AFB, California


When recording and reproducing telemetry data, signal dropouts are a prime concern to most tape users. The effect of particular depths and lengths of dropouts on data quality must be known before the acceptability of a tape for a specific application can be determined. This paper discusses the correlation of tape dropouts with data quality when recording predetection and post-detection telemetry data and IRIG timing on wideband instrumentation tape, as well as methods of reducing dropouts.


Tape dropouts are reductions in signal amplitude of the magnetic tape reproducer output.

The effect of a dropout on data quality is dependent on the length and depth of the dropout. The other major factors are the way the data is recorded (1) and the type of data being recorded. (2) This study was initiated to determine the best tape dropout criteria to use in evaluating SAMTEC wideband telemetry tapes and explore methods of reducing dropouts.


Dropouts are caused by contaminants on the tape surface (oxide particles, dust, etc), coating non-uniformity and tape damage. Dropouts due to the first two causes may be on the tape as received from the manufacturer. The third type is usually a result of improper human or recorder handling. The signal reduction or dropout is an effect caused by the record and/or playback head being spaced away from the nominal surface of the tape. The amount of this loss is given by the formula (3): where:

La = 20 log


= 54.6 a/8

exp (2 B a/ 8

La = spacing loss in db a = head to tape spacing


= recorded wavelength =

This formula says that for every wavelength separation of the tape to the head, there is a

54.6 db loss in signal amplitude. For an upper bandedge signal, the wavelength is only

0.06 mils, so a very small contaminant can result in a large reduction in signal amplitude.


Tape procured by the General Services Administration (GSA) for Government use is presently specified to have from 10 to 40 dropouts per 100 feet or less, depending on the class of tape and the track being tested (4). A new tape category is being added specifying five dropouts or less on center tracks and 10 or less on edge tracks. GSA defines a dropout as a 50% reduction in output of a 1 MHz sinewave for 10 microseconds. An additional dropout is counted for each additional 10 microseconds that the signal stays below 50%.


Since the Space and Missile Test Center (SAMTEC) has been experiencing problems with both data and timing quality due to dropouts on new tape, we wanted to determine how the

GSA dropout specification related to our actual data requirements and what dropout criteria was best suited to our specific recording applications. Our most critical data was recorded and both dropouts and data quality were evaluated simultaneously. The data selected for these tests was as follows:

Pre-detection PCM, 500 KBPS Bi-0/

Post-detection PCM, 1000 KBPS Bi-0/

IRIG A Timing, 10 KHz carrier frequency

Equipment setup was as shown in Figures 1, 2 and 3. The bit error probability (BEP) tests were basically as specified in IRIG Document 118-73 (revised July 1975), Appendix C.

Noise was added to the PCM signal until a BEP of 1 X 10 -6 was obtained when dropouts were not present. Timing was recorded per IRIG specification.

Upon playback, bit/timing errors and dropouts were displayed on a stripchart recorder and the length and depth settings of the dropout detector varied until the best correlation was obtained. Our dropout detector has variable depth from 0-100% and variable length of 5,

10, 20 and 40 microseconds, AGC amplifiers are used to keep the nominal recorder output constant. Results are given in TABLE I.

TABLE I - Dropout Correlation Results





Dropout Depth (%)




Dropout Length (usec)





SAMTEC uses several types of wideband magnetic tape recorders, which can generally be categorized as those using vacuum chambers to isolate the tape in the head area from reel disturbances and those using pinch rollers. On SAMTEC recorders, it has been found that a major factor effecting dropouts is the type of recorder used. TABLE II shows the average number of dropouts per 100 feet of tape when tested using various vacuum chamber (VC) and pinch roller (PR) recorders. The results are an average of 8500 feet of tape on a total of 31 tracks of five (5) different recorders. It should be noted that the dropout correlation tests discussed previously were conducted using pinch roller type recorders.

TABLE II - Dropouts per 100' of Tape by Recorder Type

Recorded on PR Recorder Recorded on VC Recorder

Reproduced PR


Reproduced VC


Reproduced VC


Reproduced PR


It can be seen from this table that pinch roller recorders exhibited more dropouts than vacuum chamber recorders in both the record and reproduce process.


Dropouts were tested on two 7-track pinch roller recorders at a tension of 8 ounces. When the tension was reduced to 5 1/2 ounces, dropouts increased significantly. Results are shown in TABLE III.

TABLE III - Tension vs Dropouts per 100 Feet of Tape















Tension (oz)

8 5-1/2

92 141

9 13

12 19

25 82

22 39

54 78


In an effort to improve tape quality, tape cleaner/winders are in operation at SAMTEC.

Both new and used tapes are cleaned prior to use. New tapes averaged 31 dropouts per

100 feet. Degaussing and re-recording the once used tapes reduced dropouts to 25 per 100 feet, a reduction of 19%. Cleaning new tape reduced dropouts to 12 per 100 feet or 61%.

These results are an average of 8500 feet of tape on 7-tracks of 7 recorders.


G. R. McKelvey has shown that data quality can be improved by optimizing the recorder for the type of data being recorded (1). SAMTEC has been able to significantly reduce the effect of dropouts on timing by optimizing the record electronics for the timing signal. This was made practical by dedicating certain tracks to timing. Record bias is set for 2 db overbias at 50 KHz, rather than at upper bandedge (2 MHz at 120 ips). Record level is then set just below saturation (clipping). Normal IRIG reproduce equalization is satisfactory so that playback is compatible with any reproduce electronics. We have notfound it practical to optimize the data recording because of the variety of formats required on a day-to-day basis.


A more effective use of magnetic tape can be obtained by tailoring the certification criteria to the data requirements of your particular facility. A criteria should be established that will insure that data is not significantly degraded, but not so stringent that useable tapes are discarded.

Although dropouts are caused by contaminants on the tape surface, coating non-uniformity and tape damage, the way the recorder reacts to the dropout depends on several factors.

The type of equipment being used to record and reproduce the data has a major effect on

the total number of dropouts. On SAMTEC recorders, pinch roller recorders exhibited 3 to

4 times as many dropouts as vacuum chamber recorders. Tape certification should be done on or correlated with the “worst case” recorders. Also, dropouts can be minimized by insuring that tension is not set below manufacturer’s specifications. An increase above that value should not be made without first consulting with the manufacturer on the effects on tape handling and head wear.

At a minimum, tape should be run across the recorder heads once prior to use. Dropouts can be reduced further by cleaning the tapes prior to use.


1. McKelvey, G.R., “Recorder Parameters Affecting Bit Error Rate,” Proceedings

International Telemetry Conference, 1972.

2. King, D.A., “Comparison of PCM Codes for Direct Recording,” Proceedings

International Telemetry Conference, 1976.

3. Mee, C.D., “The Physics of Magnetic Recording,” John Wiley & Sons, Inc., New

York, 1964.

4. Interim Federal Specification W-T-001553 (GSA-FSS), Tape , Recording,

Instrumentation, Magnetic Oxide-Coated; December 21, 1979.

Figure 1 - Predetection Test Configuration

Figure 2 - Post-detection Test Configuration

Figure 3 - IRIG Timing Test Configuration




R. C. Montgomery

Manager, Tape Engineering

Sangamo-Weston, Inc.

Sarasota, Florida


Current trends in digital magnetic recording are expected to lead to increases in speed, linear density and track density. These increases generate conflicting requirements on error control hardware. This paper discusses the conflicts expected and current and future methods of solution.


Tracy G. Wood

Ampex Corporation

Redwood City, California



A radically different magnetic recording approach is proposed to solve the very high density and very high data rate requirements of the 1980’s. In direct contrast to the industry’s traditional approach of head-per-track longitudinal recording is a multi-track rotary helical recorder using narrow tracks. This technique is described. These narrow tracks, in conjunction with a novel development called “Automatic Scan Tracking” (AST), make possible the development of a family of recorders with data rates up to one (1) gigabits per second with user packing densities of five (5) megabits per square inch and a bit error rate performance better than one (1) part in a million (106) without aid of error detection and correction codes.

Landsat D High Density Tape Recorders

John H. Montgomery

Martin Marietta Corporation

Denver, Colorado 80201


High density tape recorders are being provided by Martin Marietta Corporation to facilitate Landsat D data acquisition and processing. High rate data form the Landsat D

Thematic mapper and from the multispectrial scanner are to be stored on high density recorders of 42, 28, and 14 track configurations. Successive stages of data processing will utilize higher error-free reproduce techniques and tape systems will provide ease of operation with a variety of differing record and reproduce data rates.

The design and implementation of these recorders will be discussed.



Peter C. Morran and Joan E. Bebb


One Space Park

Redondo Beach, California

The Tracking and Data Relay Satellite System (TDRSS) provides data relay and tracking services for other (user) satellites. The ground station performs all the modulation/ demodulation functions for, simultaneously, as many as 19 forward (to user spacecraft) and 32 return (from user spacecraft) data channels at K, and at S-band. TDRSS is transparent to the data itself; it passes synchronized bit streams directly to/from NASA.

The data channels may exist on up to 4 communication satellites and are carried on 6 single access and 3 multiple access space antennas. TDRSS relay services are available

24 hours/day, 7 days/week. Individual services using a single TDRS may be maintained for up to 24 hours. Using 3 TDRS’, a user satellite may be serviced continuously.

The TDRSS ground station is a highly automated collection of communication electronics.

Automation is achieved through the use of a large complex of dedicated computers -

9 DEC PDP 11/70’s and a dual processor Univac 1110. From the broadest perspective these computers perform three major functions:

• management of the Ground Segment resources - ground antennas, demodulators, bit syncs, etc.

• computation of user tracking/trajectory data

• telemetry, tracking and control of the TDRS’s themselves

The problems encountered in designing the system to provide these functions were related to rapid response scheduling and management of equipment to provide relay to and from

NASA and to and from the Ground Communications Equipment (GCE), rapid trajectory and tracking computations, ground supported Autotrack for the TDRS single access antennas, and the TT&C functions of the TDRS. Further, controlling the behavior of the

Ground Station under abnormal (failed equipment) conditions posed problems.

Trades and analysis led to a Ground Station configuration with a functionally distributed, partially redundant ADPE, a shared pool of GCE and separate failover concepts for GCE and ADPE.

This configuration provides a satisfactory compromise between a highly redundant, expensive, conceptually simple system and a system affording fast recovery from error at the expense of complex resource management obligation.

* This work is being performed by TRW DSSG under contract to Western Union Corporation.

MMS Command and Data Handling

R. L. Kelley

H. A. Raymond

Fairchild Space and Electronics Company, Germantown, MD


The Command and Data Handling System which is part of the NASA Multi-mission

Modular Spacecraft (MMS) represents a versatile time division multiple access approach to processing spacecraft commands and retrieving data on real-time, delayed or preprogrammed basis. This paper traces the command signals from operator keyboard to execution within the spacecraft and telemetry/data signals from on-board spacecraft measurement to display on the operator’s CRT page. Key technical features throughout this end-to-end signal processing loop are described and discussed.



John R. Sesak


Robert M. Bowman


General Dynamics Convair Division

P.O. Box 80847

San Diego, CA 92138


The rapid growth in world demand for satellite telecommunications and the limited number of positions in the geostationary arc are leading inexorably to larger, higher capacity communications satellites. This trend, coupled with the projected weight and volume capability of the Space Transportation System (STS), will lead to satellites in the 80s weighing 5000 kg and measuring 50 to 100 feet across. By the end of the century these figures could increase again by an order of magnitude. Such large, low-density structures tend to have closely spaced, low-frequency dynamic modes. At the same time, multibeamfrequency reuse antennas (MBFRA) projecting narrow-spotbearns require pointing stability within a hundredth of a degree or so. The combination of low structural natural frequency and more stringent pointing requirements imposes the need for an entirely fresh approach to dynamic control of communications satellites. This paper outlines such an approach.

A modern optimal control methodology is advanced that provides decentralized modular control for large communication satellites. The fundamental property of the control algorithm is its ability to stabilize certain subsets of vibration modes without disturbing others.

This decoupling action allows the control task to be implemented in a modular or building block fashion so that different modal subsets are stabilized by separate controllers.

Decentralization according to functional task is also possible such that noninteracting rigid body and elastic body control is achieved. Thus, the technique provides a solution for the problem of rigid body control in the presence of low frequency elastic modes that are in the rigid body controller bandwidth. The design methodology, termed Model Error

* Engineering Specialist, Control Dynamics

** Manager Advanced Space Programs

Sensitivity Suppression (MESS), is a derivative of modern optimal control and estimation theory. Several examples illustrate the capability of the design algorithm.


The finite number of positions in the geostationary arc, coupled with the ever-increasing demand for telecommunication systems, will inevitably lead to larger, more complex communication satellites, with ever more stringent pointing requirements. Multibeamfrequency reuse antennas (MBFRA) will require pointing accuracies on the order of



degrees. The advent of the Space Transportation System (Shuttle) will allow the construction of ever larger communication satellites in the attempt to meet this demand.

These large systems will present a new class of problems for the control engineer, for these satellites will be of low density (mass to volume ratio) which tends to produce vibration modes of extremely low frequency. These low frequency modes will be closely spaced and will be located within the rigid body controller bandwidth, which compounds the control problem. Therefore, high performance design for this new regime of space systems requires control methods which differ from the usual frequency domain techniques of classical control theory.

In addition to low frequency vibration modes, the large size and possible modular construction of these systems will create another class of problems to be addressed. As shown in Figure 1, the basic space system could be modular in nature, with new modules being added during system life. Thus, another aspect of the control problem becomes evident: the control system must have inherent add-on capability to cope with this modular nature. As new modules are added or structure geometry is changed, the controller must be capable of reflecting these changes.

Now consider the large size of the advanced structures. This large size makes centralized control somewhat troublesome due to the long signal paths for actuation and sensing signals. Control of a large number of modes will involve a large number of signal paths.

To adequately cope with these problems, a decentralized control methodology is advanced which allows noninteracting control with distributed micro processing. Such an approach uses the large size of these systems to positive advantage. Noninteracting decentralization also mirrors the possible modular nature of these structures in that, for many cases, the same control hardware would be employed, and computer control gains would be

Figure 1. Large communication satellite concepts.

recomputed to account for changing structure geometry. Consequently, additional control hardware would be needed in only the new or add-on modules.

The Model Error Sensitivity Suppression algorithm, Ref. 1 through 9, is amenable to decentralized control, providing a solution for the control problem just discussed. The basic controller which is derived from optimal regulator theory is shown in Figure 2.

Figure 2. Active controller structure.


Simply stated, the flexible spacecraft control problem involves multivariable control in the presence of modeling error. If the linear, constant-coefficient differential and algebraic equations (1) and (2) x0 = Ax + Bu + v y = Cx + w



provided an accurate model of the dynamic processes governing flexible spacecraft behavior, we would implement the theories in our textbooks and the flexible spacecraft control problem would be solved. (Here, x is the state vector; u is the control vector; y is the observation vector; A, B, and C are constant matrices; and v and w represent white,

Gaussian noise.) Our analysis task is to find ways to correct for the deficiencies of this model. In other words, our approach is to reduce the model error sensitivity of the system.

It is helpful to classify the model errors as follows:

1. Model reduction (truncation) errors a. Truncation of known modes for implementation of the estimator b. Omission of unmodeled modes; truncation of unreliably computed modes

2. Parameter errors

3. Nonlinearities

4. Unmodeled disturbances

On the basis of studies in the control literature, Ref. 10 through 19, it appears that the significance of these errors is in the order indicated: truncation, parameter errors, nonlinearities, and disturbances. For particular systems these priorities may vary.


MESS is a method adapted from Ref. 1 that very effectively resolves the two most critical model error problems: truncation of known modes, and parameter errors. We speak of model truncation error sensitivity suppression and model parameter error sensitivity suppression. In effect, this method permits the suppression or elimination of truncation and parameter error sensitivity in the estimation and control system. This is accomplished without the augmentation of the state vector or the introduction of special signal compensation, such as comb filters. This paper addresses model truncation error.

References 1, 4, and 7 address parameter error.

We first partition the state equations for the modeled system into separate “controlled states” x c

and “suppressed states” x s

as shown by equations (3) and (4).



Our first task is to suppress the “control spillover,” Ref. 18 and 19, B control the states x c

via B c s u so that we can u and minimally excite the suppressed states x s


We consider only the low-order problem x0 = A c

x c

+ B c u y = C c

x c but we modify the usual performance index



(7) and substitute


By penalizing the control spillover B make B s s u, we can suppress this term. In many eases, we can u approach zero without seriously constraining B c u.

In practice, one obtains desired stability characteristics by incorporating an "-shift in the problem, Ref. 20, introducing

(9) with



This refinement is important for practical application, although not necessary to an understanding of the concept. Analogous procedures are used in estimator design which is designed by duality theory, Ref. 20 through 22.

The final control law is realized via estimated state feedback

(11) where xˆ c

is an estimate of the state vector x c

Defining estimation error e c


and K is the optimal controller gain matrix.

(12) the control vector, equation (11), becomes u = -K(x c

-e c

) and it is possible to write the state dynamics, equation (3), as



Similarly the estimator dynamics (error form) satisfy

(15) where G is the optimal observer gain matrix.

Assembling equations (13 through 15) in matrix form yields


The excitation of suppressed states occurs through the matrix B suppressed states occurs through the matrix GC s s

K, the sensing of the

; but these matrix entries are precisely those which are constrained by the algorithm. For certain classes of systems, these terms can be made zero, Ref. 23. Figure 3 illustrates the decoupling action of the controller.

Figure 3. Controller decoupling action.


Although an elastic body has essentially an infinite number of normal mode coordinates, it is possible to model with acceptable accuracy only a comparatively modest number (surely less than 100). In most applications, the unmodeled states can be safely truncated and ignored in control system design. However, there remains the problem of designing a controller for a physical system which is characterized by a number of states and is generally too large to be accommodated in a state estimator operated by a flight computer.

Moreover, it may not be desirable to develop a centralized estimation and control system even if it was possible. A decentralized concept which permits many independent estimation and control systems to assume individual control responsibilities for a subset of the system states has four advantages: (1) feasibility with small flight computers, or even individual micro-processors; (2) reliability, since subsystem failure need not imply system

failure, and redundancy of model coverage is easily implemented; (3) adaptability, since design changes or model revisions can be accommodated by changes in individual subsystem controllers; (4) economy, since the greater design flexibility permits consideration of more economical alternatives.

The model error sensitivity suppression method can be incorporated into a decentralized control system quite easily, as illustrated by the following example. Suppose that an elastic body (fully characterized by an infinite number of coordinates) is adequately described in terms of a model having 12 degrees of freedom, as defined by 12 normal mode coordinates for the uncontrolled body. If each of these coordinates is judged to be dynamically significant, but a twenty-fourth order state estimator is judged to be infeasible or undesirable, one can construct instead three eighth-order estimators (or four sixth-order estimators, or any other such combination). Each of the three subsystem controllers can be charged with eight states, or four normal modes, grouped in any convenient manner

(perhaps sequentially, or geographically, or by similarity of mode shape). The critical requirement is, of course, that any one of the subsystem controllers exert only minimal influence on those states with which it is not charged. This requirement can be met by assigning each subsystem controller not only eight controlled states (x suppressed states (x s

); applying the method of the MESS algorithm.


) but also sixteen

Figure 4 illustrates the concept for two decentralized controllers. Application of the MESS algorithm tends to decouple the original system into component subsystems as shown.

Figure 4. Decentralized control - two subsystems.

Each subsystem is stable when isolated, and the magnitude of subsystem interaction is constrained; i.e., each controller is “blind” to the other controllers and “sees” only its own subsystem. The remaining portions of the global system dynamics remain “invisible” to it.

This decoupling effect of the MESS algorithm eliminates the need for coordination and information exchange among the controllers. The decoupling between the subsystems is

not absolute but for many systems can be minimized to the extent that each controller views the actions of the other controller as a slight disturbance.

The closed-loop equations for the two controller systems are given below.


In these equations, x


is the state vector associated with controller I, and x


is the state vector associated with controller II. The Greek “dels” indicate matrices constrained by the control and observation algorithms. Smallness in these matrices entries implies decoupling.

Zero entries imply total decoupling.



Figure 5 illustrates the large space structure used as a baseline vehicle for test applications of the model error sensitivity suppression method. The control task is the damping of the vibratory modes by two colocated sensor/actuator pairs at corners labeled A and B in

Figure 5. Each pair is capable of torquing in roll and pitch and sensing local angular rates about these axes. For the studies presented, the platform is modeled by its four lowest frequency elastic modes with zero damping. Except for the decentralized altitude and elastic body controller, the three rigid body modes are omitted in order to facilitate computations. A state estimater, designed by dual MESS, is employed in all examples.


This example illustrates that the platform is not an overly benign system. Modes 3 and 4 are neglected during the design process (truncated from the model) and an optimal controller/estimator is designed for modes 1 and 2. Figure 6 illustrates the unstable result — all modes, controlled and suppressed, are unstable.


The design algorithm is now applied to the platform.

Figure 7 illustrates the results. The suppressed modes receive very little excitation from the controller and oscillate at low amplitudes. These modes ring because they are modeled with zero damping; the controller is not to damp these modes.

Figure 5. Large space platform for design examples.

Figure 6. Attempted design by modal truncation unstable.

Figure 7. Suppression of two modes by the MESS algorithm.


This example illustrates the capacity for decentralization that is inherent in the MESS algorithm. As shown in Figure 8, we wish to address the problem of elastic body modes whose frequencies are within the rigid body controller bandwidth. First we design a rigid body controller using MESS. (For this example, a position sensor was placed in the center of the platform to sense rigid body motion.) Figure 9 illustrates the rigid body controller action. The elastic modes are suppressed and ring with low amplitude.

Figure 8. Elastic modes in the rigid body controller bandwidth.

Figure 9. Decentralized rigid body controller response.

The next step is the application of the MESS algorithm for elastic body control. Figure 10 illustrates the elastic body controller action. The rigid body mode rings at very low amplitude and frequency it is stable).

The final step of the procedure is the simultaneous placing of both the rigid body and elastic body controllers on the platform. Figure 11 illustrates the results and demonstrates that the MESS algorithm can provide noninteracting rigid body and elastic body control, although the modal frequencies are in the rigid body controller bandwidth. Figure 12 shows the open and closed-loop pole-zero plots.

Figure 10. Decentralized elastic body controller response.


For large structures, the modal frequencies of controlled and suppressed modes will be closely spaced. As a result, the controller should not be overly sensitive to small frequency differences. To test this sensitivity for MESS designs, identical frequencies of 1.0 radian were placed in a modified model of the platform and a MESS design was performed. Figure 13 illustrates these results. The algorithm is capable of providing control with identical modal frequencies.


A practical system must not be overly sensitive to parameter variations. To examine the closed-loop parameter sensitivity, a MESS controller for three controlled modes and one suppressed mode was designed. Figure 14 illustrates system performance for parameter variations of standard deviation for 30% from the baseline vehicle. (All matrix entries were varied randomly according to a normal distribution with 30% standard deviation.) This study and others,

Ref. 1 and 4, have indicated that MESS controllers are not overly sensitive to parameter error. A more exhaustive study including the effect of actuator number was undertaken in

Ref. 24.

Figure 11. Decentralized rigid body and elastic body controller - two controllers.

Figure 12. Pole-zero plots for the decentralized controller.

Figure 13. Suppression of modes with identical frequencies.

Figure 14. Parameter error study. All system parameters varied according to 30% standard deviation.


Ultimately, every controller will have to function in the presence of unmodeled modes. To evaluate the effect of unmodeled modes, a two controlled and two suppressed mode controller was designed, and three unmodeled modes were added to the system. Figure 15 illustrates the results. Stability, suppression, and controlled mode stabilization were maintained in the presence of unmodeled modes.

Figure 15. Truncation error study — three unmodeled modes.


Although more experience is required before all the capabilities and limitations of MESS are fully appreciated, it seems evident that this approach permits a practical solution to the error sensitivity problem. For a class of physical systems, at virtually no cost in control system complexity, this method provides substantial reduction in control and observation spillover for truncated modes having known characteristics. On the basis of preliminary empirical evidence, the resulting system performance seems to have relatively little sensitivity to uncertainties in these modal characteristics.

The MESS approach is particularly advantageous in a decentralized control mode of application in which each subsystem controls certain modes of deformation with minimal spillover into other subsystem modes.

These advantages of the MESS technique make it an ideal candidate to satisfy the stringent requirements for control of the large communications satellites of the next two decades. It seems unlikely that conventional techniques will be able to satisfy the pointing requirements of large, multibeam antennas attached to large, lightweight structures having several closely spaced low-frequency dynamic modes. We therefore expect such systems to implement decentralized control schemes incorporating some version of the Model Error

Sensitivity Suppression technique described above.


1. Sesak, J.R., “Sensitivity-Constrained Linear Optimal Control: Analysis and

Synthesis,” Ph.D. dissertation, University of Wisconsin, 1974.

2. Sesak, J.R., “Control of Large Space Structures via Singular Perturbation Optimal

Control,” presented at AIAA Conf. on Large Space Platforms, Los Angeles, Calif.,

Sept. 27-29, 1978, paper no. 78-1690.

3. Sesak, J.R. and Ahn, S.M., “Singular Perturbation Optimal Control of Large Space

Structures,” presented at IRIA/IFAC Workshop on Singular Perturbations in Control,

LeChesnay, France, June 5-7, 1978.

4. Sesak, J.R. and Higgins, T.J., “Sensitivity-Constrained Linear Optimal Control via

Forced Singular Perturbation Model Reduction,” Proc. 1978 IEEE Conf. on Decision and Control, San Diego, Calif., Jan. 10-12, 1979, pp. 1145-1146.

5. Sesak, J.R. and Coradetti, T., “Decentralized Control of Large Space Structures via

Forced Singular Perturbation,” presented at AIAA Aerospace Sciences Meetings,

New Orleans, La., Jan. 15-19, 1979, paper no. 79-0195.

6. Sesak, J.R., Likins, P.W., and Coradetti, T., “Flexible Spacecraft Control by Model

Error Sensitivity Suppression,” J. Astronautical Sciences, Vol. 27, No. 2, Apr.-Mar.

1979. Also 2nd VPI/AIAA Symposium on Dynamics and Control of Large Flexible

Spacecraft, Blacksburg, VA, June 21-23, 1979.

7. Benhabib, R.J. and Iwens, R.P., “Control of Large Space Structures Using

Equilibrium Enforcing Optimal Control,” presented at AIAA Conference on

Advanced Technology for Future Space Systems, Hampton, VA., May 8-10, 1979, paper no. 79-0927.

8. Joshi, S.M. and Groom, N.J., Design of Reduced-Order Controllers for Large

Flexible Space Structures. Proceedings of the 16th Annual Allerton Conference on

Communication, Control, and Computing, October 4-6, 1978.

9. Joshi, S.M. and Groom, N.J., “Controller Design Approaches for Large Space

Structures Using LQG Control Theory,” presented at 2nd VPI/AIAA Symposium on

Dynamics and Control of Large Flexible Spacecraft, Blacksburg, VA., June 21-23,


10. Likins, P., “The Application of Multivariable Control Theory to Spacecraft Attitude

Control,” Proc. of IFAC Multivariable Technological Systems Symposium

Fredericton, N.B., Canada, pp. 11-20, 1977.

11. Seltzer, S.M., “Attitude Control of Orbiting Large Flexible Structures: An Overview,” presented at Annual Rocky Mountain Guidance and Control Conference, Keystone,

Colorado, 1978.

12. Larson, V., Likins, P., and Marsh, E., “Optimal Estimation and Control of a Solar

Electric Propulsion Vehicle,” IEEE Trans. on Aerospace and Electronic Systems,

Vol. AES-13, No. 1, pp. 35-48, Jan. 1977.

13. Skelton, R.E. and Likins, P., “Techniques of Modeling and Model Error

Compensation in Linear Regulator Problems,” Advances in Control and Dynamic

Systems, Vol. XIV, ed. C.T. Leondes, Academic Press, 1978.

14. Skelton, R.E., “Control by Model Error Estimation,” Ph.D. dissertation, UCLA,


15. Skelton, R.E. and Likins, P., “Orthogonal Filters for Model Error Compensation in the Control of Nonrigid Spacecraft,” J. Guidance and Control, Vol. 1, No. 1, pp. 41-

49, 1978.

16. Skelton, R.E., “An Algorithm for an Approximation of the Minimal Controller

Problem,” J. Guidance and Control, Vol. I, No. 1, pp. 90-93, 1978.

17. Skelton, R.E., “Some Limitations of Model Reduction and Controller Design

Methods for Large, Uncertain Dynamical Systems,” AAS Paper 78-103, presented at

Annual Rocky Mountain Guidance and Control Conference, Keystone, Colorado,


18. Balas, M., “Modal Control of Certain Flexible Dynamic Systems,” Siam J. Control and Opt., Vol. 16, pp. 450-462, Aug. 1978.

19. Balas, M., “Feedback Control of Flexible Systems,” IEEE Trans. Automatic Control

AC-23, pp. 673-679, 1978.

20. Anderson. B.D.O., and Moore, J.B., “Linear Optimal Control,” Prentice-Hall, 1971.

21. Bryson, A.E. and Ho, Y.C., “Applied Optimal Control, “Blaisdell Publishing Co.,

Waltham, Mass., 1969.

22. Kwakernaak, H. and Sivan, R., Linear Optimal Control Systems, Wiley, 1972.

23. Coradetti, T., “Orthogonal Subspace Reduction of Optimal Regulator Order,” presented at AIAA Guidance and Control Conference, Boulder, Co., Aug. 6-8, 1979.

24. Kammer, D.C., and Sesak, J.R., “Actuator Number Versus Parameter Sensitivity in

Flexible Spacecraft Control,” presented at 2nd VPI /AIAA Symposium on Dynamics and Control of Flexible Spacecraft, Blacksburg, VA., July 21-23, 1979.

25. Kokotovic, P.V. and Yackel, R.A., “Singular Perturbation of Linear Regulators: Basic

Theorems,” IEEE Trans. on Automatic Control, Vol. AC-17, pp. 29-37, Feb. 1972.

26. Sannuti, P. and Kokotovic, P.V., “Near-Optimum Design of Linear Systems by a

Singular Perturbation Method,” IEEE Trans. on Automatic Control, Vol. AC-14, pp.

15-33, Feb. 1969.


This appendix contains the mathematical model for the structure, matrices A, B, and C.

The control gain matrix K, and the observer gain matrix G, are included for a threecontrolled-mode, one-suppressed-mode controller (Figure A-1).

Figure A-1. System matrices.

The procedure used to calculate the optimal gains is outlined in Table A-1. The internal driver minimization form of the algorithm is used to alleviate the need for calculating any inverse matrices of As, Ref. 1 and 2.

The state weighting matrix Q c

is taken to be an identity matrix of proper dimension. The same is true of the control weighting matrix R c

. The suppressed mode weighting matrix R equals $1, where I is an identity matrix and $ is a scalar equal to 10 10 the performance index of equation (3) Table A-1 is further modified by the "-shift

(equation 3 - text). Alpha for the regulator is 0.35 and " for the estimator is 0.40.

Therefore, the design procedure is: choose the suppressed modal system, form the performance index of equation (3), and further modify equation (3) by the "-shift technique. Mode (4), the highest frequency mode, is suppressed for the example.


. As noted in the text,

Typical system responses of the four actuator systems at nominal values are shown in

Figure A-2.

Figure A -2. System responses of the four actuator nominal system.

Table A -1. Performance under modification for internal driver minimization.

DMSP Block 5D-1 Computer Controlled Spacecraft

Lt. Col. Stephen M. McElroy


Louis Gomberg


Roger Te Beest



The Defense Meteorological Satellite Program (DMSP) Block 5D-1 satellite is the first of a new generation of DMSP long life satellites to utilize onboard programmable computers for spacecraft control functions. During ascent and orbit injection the computers perform the navigation, guidance, and control functions autonomously; during on-orbit operations, they perform attitude determination and control, command and control, and miscellaneous other control functions, with only modest interaction from the ground.

Four DMSP Block 5D-1 satellites employing these computer controls are currently on orbit and operational. On-orbit experience shows that performance has exceeded all expectations with respect to reliability and satellite life-time. In addition to providing the control functions for which they were designed, the computers have provided additional benefits by allowing the control systems to be reprogrammed from the ground to overcome hardware failure and degradation in other on-board components.

This paper describes the DMSP mission, gives a brief overview of the integrated spacecraft system configuration, and provides the details of the control systems used in the various mission phases. The hardware and software portions of the control systems are described and some examples are provided showing how the reprogrammable capability allowed several orbital anomalies to be overcome and satellite life extended.

1 SAMSO, Los Angeles, CA

2 RCA/AED, Hightstown, NJ



Peter Foldes

General Electric Company, Space Division

Valley Forge Space Center, P. O. Box 8555, Philadelphia, Pa. 19101

1. Introduction

With the rapid development of spacecraft systems the trend is toward more sophisticated and larger on-board antennas. Many of these antennas employ an array of radiating elements, either as part of a feed system illuminating a larger aperture optics or as the direct radiators. The main purpose of using multiple radiators instead of a radiator in conjunction with a large aperture is to introduce a much larger degree of freedom in achieving desirable antenna characteristics. If an antenna system contains n independent radiating elements then the amplitude and phase of each of these elements can be selected for desirable results.

For medium complexity antennas it is usually adequate to set the element excitations as part of the basic design and not change them during the operational lifetime. With increasing complexity and lifetime requirement such an approach is no longer economical.

Thus the more sophisticated an antenna system is the more complex its reconfigurability and control requirement become.

The most important applications for antenna reconfigurability may be listed as follows;

1. Change of coverage pattern shaped to compensate for satellite location changes, and to follow long or short term variations of communication traffic.

2. Change of EIRP footprints to reduce rain related fadings.

3. Lock beams to fixed ground locations irrespective of mechanical mispointing of the spacecraft antenna.

4. Produce pattern minimums toward interfering sources.

5. Compensate for ionospheric beam refraction or polarization rotation effects.

6. Compensate for thermally or dynamically introduced antenna shape changes.

In the following, three examples will be presented to illustrate some of the above applications for synchronous satellites. The selected examples will be: a noncontiguous coverage, multibeam, global communication antenna, a contiguous coverage multibeam,

regional communication antenna and a very high gain, multielement microwave power transfer antenna. The common feature of these antennas is that their instantaneous performance requires electronic control of the amplitude and phase of a large number of radiating elements.

2. Noncontiguous Coverage Multibeam Antennas

Definitions — The purpose of a comsat antenna in synchronous orbit is to provide coverage of specified traffic areas with a given antenna gain. The total field of view of such an antenna covers a certain angular region; for instance 17.3E, the angular diameter of the earth disc. A noncontiguous coverage antenna forms a number of such subangular regional coverages, which partially blanket the larger angular region. The cross section of these subangular regions and the corresponding antenna beams are generally irregular, since they must conform to boundaries of traffic regions, continents, countries, etc. These are called shaped beams and are typically formed by a number of circular cross section component beams. In some cases a single, not necessarily contiguous shaped beam is adequate to fulfill the mission requirement. These single shaped beam systems have only one terminal between the antenna and the transmit or receive system of the spacecraft, no matter how many component beams are utilized. In other cases a number of shaped beams form an overall noncontiguous coverage system. These multiple shaped beam systems have one transmit and one receive terminal for each of the shaped beams. Such beams could simply subdivide the overall angular regions into smaller ones thus increasing the available minimum gain in each of the coverage regions. However, more frequently it is necessary to reduce the sidelobe radiation of each beam into the coverage region of all other beams, thus provides a beam isolation between beams. This allows the reuse of the same frequency spectrum in each of the beams, to extend the total communication capacity potential of the system. Isolation between beams can also be provided, even when they occupy the same angular region. For this case the isolation is based on polarization orthogonality, or polarization isolation.

When two or more shaped beams must have a given beam isolation between them, e.g., a given level difference at a given spatial direction within their respective coverage regions, then the beam isolation between these identically polarized and identical frequency radiations can be increased by increasing the angular separation between the contours of the coverage regions. The angular region between two such coverages is called guard region. Its width can be conveniently characterized by normalizing it to the width of the component beams from which the shaped beams are synthesized. In most noncontiguous coverage systems the width of the guard region must be minimized, but otherwise no specification is given for this region.

The defined shaped beam configurations can be realized on a fixed or on a reconfigurable basis. In the latter, the reconfigurability may be necessary to change one or more of the shaped beam cross sections, the EIRP distributions within the beams, split or combine beams, etc.

Figure 1

Noncontiguous earth coverage beam topology with 144 component beams of

" = 1.3E cell diameter, forming 7 shaped beams. (Dotted lines show contours of max. coverage region, solid lines show boundaries of component beams associated with shaped beam.)

Example) 1 - Figure 1 shows a beam plan composed of 144 component beams, grouped into seven shaped beams. Each of the shaped beams is formed by a number of main component beams, which can be excited up to the maximum (0 db) power level. In Figure 1 the main component beams are bounded by dotted line. Additionally each of the shaped beams uses a number of auxiliary component beams, typically at the periphery of the shaped beams, excited in the -14 db to -26 db range and used to control sidelobes. The combined main and auxiliary component beams are bounded by the solid line in Figure 1.

Figure 2 shows the application of this beam plan to a synchronous atellite over the Atlantic at 34.4EW longitude. Only those component beams are excited close to 0 db level which cover meaningful traffic areas. The resultant seven beams are not contiguous but the uncovered areas are ocean surfaces or regions of relatively low traffic. Seven reuses of the

1 P. Foldes, M. Berkowitz, “Reconfigurable Multibeam Antennas for Satellite

Communications,” 3 rd World Telecommunications Forum, Geneva, 23-26 September 1979.

bandwidth are thus possible with a single polarization and fourteen if orthogonal polarizations are employed.

Figure 2

- Possible beam settings with 7 beam, noncontiguous coverage antenna. Intelsat terminals are located on map.

Figure 3 shows the beamforming network (BFN) necessary to realize Beam 1 of Figure 1.

The network contains 16 variable phase horns and a binary power divider tree of 15 variable power dividers (VPD’s).

Figure 3

- Layout of beamforming network (BFN) for Beam 1 on Figure 1.

Method of Control - The provision of proper amplitude and phase at the aperture of the radiating elements is based on the use of two type of phase shifters in the beamforming network.

The control of the VPD’s is based on the use of a pair of adjustable phase shifters, which are driven by n = ± 45E and K 45E respectively relative to a nominal 45E position. The circuit containing two hybrids and two such phase shifters is exhibited on Figure 3. For such circuit the transfer phase is independent of the power division ratio, which is

The phase shifting element can be any device with low loss and frequency independency within the desired frequency band. If switching speeds are relatively low (in µs range) then for C band and higher frequency applications ferrites are preferred because of their relatively low loss and wide bandwidth. The circuit can be implemented either in waveguide or in stripline.

Figure 4 shows the photograph of a four-way VPD implemented in the 5.925-6.725 GHz frequency band using stripline hybrid elements. Such a unit has a total of six hybrids and six ferrite phase shifters. The power division increment ratio is a function of the number of states available in the phase shifter drive. For the circuit shown on Figure 5 63 states are used, which results in 1.452E phase shift resolution. Such resolution provides appr. .4 db power division step around the -3 db power division condition.

Figure 4 - Top and bottom view of four-way stripline variable power divider (VPD).

The control of the output phase of the radiating elements require the setting of the corresponding variable phase shifters (VPS). In the described implementation these are

± 180E phase shifters with 63 phase states resulting in 5.81E phase setting resolution.

In summary, it can be noted that the amplitude and phase control of the described 16 element feed array requires 46 phasers to control by 31 commands. The 15 VPD’s are determined by 15 states out of 945 total states, the 16 output phasers are determined by 16 states out of 1008 total states and the total control bit stream correspond to 1963 possible states. For the complete system shown on Figure 1 the number of possible states is about

17700. The reconfigurability time is related to this number, but it is influenced by other factors. For instance, if the system has to be reconfigured without interrupting traffic, then the finally desired beam shape may be reached via a number of intermediate steps to reduce system transients. The setting of the complete 144 elements antenna needs appr.

279 commands. This can be implemented by a 15 bit data stream. Under these conditions the maximum average command rate is about 200 commands per second,thus the necessary 279 command to change the entire system can be executed in about 1.4 sec. The phase shifters reach their new states within 15 µs after the 0 to 1 transition of the strobe line.

The automated control of the BFN requires a Central Electronics (CE) unit which is internal to the BFN. This is driven directly from the spacecraft TTC interface or for development tests from a microcomputer. Figure 5 shows the block diagram of a typical


Figure 5 - Block diagram of BF14 Control Electronics.

This unit accepts commands from the TTC or microcomputer, verifies commands via telemetry, provides command editing capability if an erroneous command is received, controls the VPD and VPS drives and power supply to execute commands, and verifies that the commands have been executed via telemetry.

The CE is driven by a control computer via the TTC system or directly during developmental testing. Typical implementation contains a CPU (such as INTEL 8080) with associated 16K of random access memory for the discussed 16 radiating element system. The system also needs a teletype terminal and an ROM monitor with a bootstrap loader. Additional peripherial equipment includes a high speed paper tape reader and tape punch and a custom developed interface to the CE.

3. Contiguous Coverage Multibeam Antennas

Definitions - Contiguous coverage multibeam antennas have a formation of subangular region beams, which in total cover completely a larger overall angular region. The subangular regions may be illuminated by independent component beams or shaped beams formed from component beams, where independence (decoupling) is achieved by polarization orthogonality, frequency channelization, sidelobe suppression or by a combination of these methods. While noncontiguous coverage systems provide communications between isolated traffic regions, like large population centers, cities or even specific earth stations, thus they are ideal for trunk type traffic, the contiguous coverage systems can provide complete area coverage, thus they are ideally suitable for the more general, multiple access type traffic. The basic design problem for these type of systems concerns with the maximization of total communication capacity in a given allocation frequency band under the restraining condition that the cost of the channel mile/minute must be minimized. In an ideal contiguous coverage system the total communication capacity should not be uniformly ,distributed. Instead the provided communication capacity must match the density of traffic or approximately the density of population. In such a traffic matched contiguous coverage system the provided bandwidth per capita is constant on a time average basis. In contiguous coverage systems generally it is not possible to provide the full allocation bandwidth in any given shaped beam. Instead the subfrequency bands (channels) and polarizations must be arranged among shaped beams in such a manner that the isolation can be maintained and the allocation spectrum reused the maximum possible times for a given isolation. The planning of such arrangement is called beam topology design.

Table 1 shows a few topology plans covering some of the most important beam topologies.

Example - The meaning of the various characteristics listed in Table 1 and the associated antenna coverage control requirements will be explained in an example. For this purpose beam Plan 5 was selected, because it illustrates most generally the problems associated with contiguous coverage, multibeam antennas. Figure 6 shows the beam plan for this case. In this plan the total allocation frequency band AF is divided into eight subbands

(channels, B and additionally two orthogonal polarization are used, thus a total of 16 independent channels are created. With the availability of 16 such channels shaped beams

Table 1

Topology Characteristics of Studied Beam Plans each using four component beams and a particular combination of frequency channels can be generated which result in a 4 x 4 element grid. Such a grid of “grand cells” can be repeated indefinitely over the region to be covered. Four channels are avilable in each of the component beams. For instance if an overall base bandwidth of 2048 MHz is divided into 8 channels of 256 MHz each, then a maximum of four such channels are available, for a total bandwidth of F


= 1024 MHz, per component beam cell. The cells can be so grouped that two (so-called “empty”) component beam cells will always separate shaped beams using the same channel on the same polarization.

Figure 6

- Channel designation and beam topology using 16 orthogonal channels. Four main and ten (optional) auxiliary component beams form each shaped beam

(quadrupled). Center frequencies of channels are separated by multiples of T from each other

Figure 7 shows an example of such channel combinations. The numbers indicated in each cell are the channel codes, and horizontal and vertical polarizations are indicated by uncircled and circled numbers respectively. Identical channel combinations form a square grid in which the elements repeat at four component cell periodicity. When such a grid is superimposed on the U.S. for " = .5E, cell diameter, then for the lower 48 states, N = 11.6

reuses of the spectrum are possible. (Half of the total bandwidth, or four channels, is available in each cell. The same channel on orthogonal polarizations never occurs in the same cell, Consequently the polarization isolation is independent of propagation conditions. Four-way power dividers are required to form the shaped beams. The power amplifiers in this system can be directly connected to the input of the radiating elements to minimize losses or can be placed before the beamforming network to eliminate the need for highly linear (solid state) amplifiers. The disadvantage of the scheme is that the overall band (2048 MHz) is fragmented into relatively narrow, 256 MHz channels and the communication capacity is uniformly distributed over the coverage area whether required or not.

Figure 7

- Contiguous coverage of 68 component beam antenna based on plan in Figure 6.

Spectrum reuse: 11.6. " = .5E. Uncircled and circled channel numbers represent vertical and horizontal polarizations.

The beam plan given in Figure 7 has a total of 68 component beams from which 87 shaped beams are formed. The number of shaped beams, N


is larger than the number of component beams n, because of the overlap between shaped beams. This overlap is a very important feature of Plan 5, because it allows great flexibility in network planning and eliminated the need to operate earth stations at the steep slope of the shaped beams. In order that the shaped beam regions can be connected with each other a switch of considerable complexity is necessary. The complexity of this switch rapidly increases with the number of independent shaped beams. As a step toward simplification, some of the fractional and under utilized shaped beams (e.g. those which have four component beams) can be combined. Under these conditions the system may have only 64 shaped beams and the number of spectrum reuses is reduced to 8. This still provides 16.384 GHz of effective communication bandwidth, or about 74 Hz bandwidth for each citizen of the U.S.

Note that due to the combination of some of the shaped beams the communication capacity distribution within the system is already nonuniform. This nonuniformity can be further increased to favor large metropolitan areas. Figure 8 shows the modification of the original

Plan 5 such that certain channels are omitted from less populated cells, and utilized in more populated cell. The percentages indicated refer to the available bandwidth. relative to the allocation bandwidth.

Table 2 shows the most important antenna characteristics achievable with such a system in the 18/28 GHz frequency band, using an antenna diameter D


at 20 GHz. The table shows the already discussed characteristics and also the loss less gain, G o

G eff

= G o

- "


, and the carrier to interference ratio, C/I.

of the satellite antenna,

Figure 8

- Modification of Plan 5 for nonuniform communication capacity distribution.

Summary of Characteristics for C o

(" stm

Table 2

= 16.384 Ghz Effective Comuication System


= -1.2 db, "


= -.5 db)

* "


= edge of coverage reduction of gain relative to ideal maximum due to beam shape







satellite loss, " sat

, atmospheric loss, " atm

and earth station antenna losses

Figure 9 shows a simplified block diagram of the system. Each transmitter feeds four horns and each horn transmits signals from four transmitters. The four-way VPD circuit is similar to that in Figure 3.

These VPD’s are usable for concentration of traffic as described on Figure 8 or for rain fading compensation. For instance if one of the four cells of the shaped beam represents the center region of a large metropolitan area with heavy day time traffic then all or most of the transmit power of a given channel can be concentrated in this cell during the day.

During the night, when traffic is less concentrated, the transmit power can be more evenly distributed. Plan 5 allows a maximum of 1:4 traffic concentration.

Figure 9

- Block diagram for contiguous coverage multibeam antenna based on Plan 5, 68 component beams, 64 shaped beams. Complete connectivity is available from any shaped beam to any other shaped beam.

Alternatively, the BFN can be used to allocate more power to cells with heavy rain attenuation. On the average each cell of the shaped beam has 25% of the available

downlink power. If one cell has 9.54 db rain attenuation and the other cells no rain attenuation, then the allocation of 75% power to the rainy cell and 8.33% power to the clear weather cells results in 4.77 db fading in all four cells. Thus the BFN control reduced the peak fading margin (peak power) requirement by 4.77 db. Since control of the BFN takes of the order of 1 sec, rain fading control can be implemented with time constants consistent with nominal rain rate variations, provided closed loop automatic fading compensation is implemented. Such a system requires the continuous monitoring of C/N in all the cells of the network.

The beam plan shown in Figure 7 utilizes .5E diameter beam cells. In such a system an earth station at the contour of the coverage will experience a pattern slope, S in the order of 1 db/.05E. For .05E pointing error of the spacecraft antenna the signal at the earth station then varies by 1 db. This variation has to be added to the required fading margin.

When a nonuniform communication capacity distribution system, then S is further increased. The effect of the large slope can be reduced by reducing the pointing error of the beam. This can be done electronically, by expanding the four element network feed into a 16 element network per shaped beam. In this case 12 auxiliary horns are added around the original four main horns. When the antenna is pointed into its nominal position these horns are not excited. When the antenna is mechanically mispointed, then a small amount of power is fed into the appropriate auxiliary horns to steer the beam. When the mechanical pointing error of the beam is small it may be adequate to utilize the original four main horns for beam pointing corrections. Such control requires the simultaneous adjustment of all VPD’s in the system by an error signal generated by one or more ground beacons.

The system exhibited in Figure 9 has 64 uplink and 64 downlink shaped beams. To connect them, each uplink beam must have capability of being connected to any downlink beam. On the average (for uniform traffic distribution) the 256 MHz uplink channel may be divided into 64 x 4 MHz wide band channels (slots) which then can be routed toward any downlink beam. Many possible implementations of this switching function are possible. Figure 9 shows one in which all the slots are of equal bandwidth, their sequence is optionally rearrangeable and the actual switching is done at a low intermediate frequency. For such a case about .5 M diodes are needed. In more complex systems a number of different slot bandwidths are desirable to accommodate different traffic channels. Control of these switches is similar to that of a standard telephone exchange cross bar switch. Since individual switch channels are of 4 - 32 MHz bandwidth the network configuration involves relatively slow switching functions.

4. The Single Beam, Very High Gain, Multielement Antenna

Definitions - As the directivity of the antenna increases the inaccuracies in its pointing and in its shape play an increasingly important role. The previous examples involved .5E - 1.3E

component beam cell sizes. Under these conditions control of beam shape and pointing was accomplished with phase shifters in the antenna BFN, which could be controlled relatively slowly. When the antenna beamwidth is reduced by an order of magnitude or more, control of the beam shape and direction requires different techniques. For such situations the phase of each radiating element must be continuously measured and adjusted on an instanteous basis so that structural (array element location) uncertainties are compensated for. The control system locks the very high gain, space-originated antenna beam to a specifically selected earth receive terminal. In such systems the apertures of the space and earth located antennas are comparable and the design is optimized for maximum power transfer efficiency.

The Solar Power Satellite has such requirements. At 2.45 GHz, with a 1 km diameter space antenna (array), the footprint diameter on the ground is about .01E. In the following some of the control problems for such an antenna are discussed.

Example A

2 ,3

- The purpose of the phase control circuit for space antenna of the microwave transmission system is to focus as much as possible of the microwave power radiated from space to the rectenna located on the ground.

Figure 10 shows the simplified block diagram of the system. According to this figure the overall system consists of a ground and a space segment.

On the ground a transmitter and antenna complex generates a pilot signal, which is radiated toward the space antenna. In space the subarray elements of the overall space array antenna receive the pilot signal in a phase corresponding to their location relative to the ground antenna. By comparing these phases to the phase of one of the subarrays

(typically the center subarray) the phase differences at the individual subarrays are determined. Then the relative transmit phase of these subarrays is set to the conjugate of the received phases assuring that the downlink signals from all the subarrays arrive in phase to the pilot antenna. The correct operation of the system is monitored on the ground by a set of monitor stations. The output signals from these stations are used to calculate fine corrections, which may be necessary to compensate second order systematic pointing errors.

2 R. C. Chernoff, “Large Active Retrodirective Arrays for Space Applications,” JPL,

NASA Contract NAS 7-100, October 32, 1977.

3 W. C. Lindsey, “A Solar Power Satellite Transmission System Incorporating Automatic

Beamforming Steering and Phase Contgrol,” LINCOM Corp., Pasadena, Calif., TR-7806-0977,

June 1978.

Figure 10

- Simplified block diagram of the phase control system for retrdirective SPS antenna.

A key function in the operation of the above described system is the determination and the conjugation of the relative phases of the pilot signals of the subarrays. This requires the generation and distribution of a reference phase for the conjugators. In the selected system the reference phase is the phase of the A o

subarray and this phase is distributed over a transmission line tree to the other subarrays. In order to eliminate variations in the reference phase due to electrical length changes in these transmission lines the signals to be conjugated are sent back to the next higher level node on the phase distributing network. This is equivalent to performing all conjugations at the A o

subarray. In such an arrangement the phase distributing lines are used back and forth, thus their line length changes are not affecting the conjugation process.

Figures 11 and 12 show more detailed block diagrams of the system. The operation of the system can be explained by following a typical signal through the circuit.

On the ground (Figure 11) a pilot generator at f nominal f f = f


- f




= 2460 MHz is amplitude modulated by a

= 76.5625 MHz - 77 MHz signal. The carrier is suppressed and the remaining

= 2383 MHz and f


= f


+ f


= 2537 MHz tones are distributed to the transmitters of three antennas. These antennas are 10 m diameter steerable paraboloids, which are located in the apexes of a triangle, approximately 1.3 km from each other and symmetrical relative to the center of the grand receive antenna (rectenna). These antennas are used for fine positioning of the downlink beam.

Figure 11

- Ground segment of the control system for retrodirective SPS antenna based on three ground pilot signal transmit sources.

Figure 12

- Space segment of the control system for retrodirective SPS antenna, based on phase conjugation and returnable phase reference phase distribution on the spacecraft.

At the pilot antennas the level and phase of the transmitters can be adjusted in such a manner, that the effective phase center of the three element array appears to be adjustable from the spacecraft antenna. This adjustment is achieved by the pilot location control subsystem, which is using input signals from the monitoring antennas of the downlink beam.

The radiated two tone pilot signal is propagated in the uplink via the atmosphere and ionosphere. Due to practical propagation conditions the actual propagation path will be somewhat refracted from the ideal path applicable for vacuum. Furthermore, the refraction caused by the ionosphere will be larger for the f tone, because of the frequency dependency of the propagation constant in the ionosphere.

At the spacecraft (Figure 12) the 1 km diameter antenna consists of N = 10 associated with a Receiver, Conjugator, Regenerator and Transmitter.

5 subarray elements, each with a nominal 10 m x 10 m linear dimensions. Each of these subarrays are

Generally the uplink frequency f


and the downlink frequency f the phase of the received wave at f



are different. Typically

is determined from the modulating tones at a lower intermediate frequency, which then is multiplied back to the f


downlink frequency. The phase of the end result shall be ideally the conjugate of the received phase at f



The conjugation can be done either exactly or approximately. An exact conjugation requires that the i-f phase from the receiver is modified by a factor so that after conjugation and multiplication it becomes exactly the desired value at f



Alternatively the conjugation can be done approximately and the generated systematic beam pointing error can be corrected by a three element pilot antenna array on the ground.

The presented example uses this later method. The details of the simple conjugator circuit is shown on Figure 13. This circuit receives the previously calculated o/ reference receiver and the two tones from the receiver which produced the signal to be conjugated. The first pair of mixers and band pass filters generate the lower and higher side bands.


phase from the

When these signals are mixed in the second mixer and the upper side band is kept the phase of the resultant signal is where p = 2f




, f


= is the uplink frequency, r o antenna to the reference subarray and )

1 under consideration.

is the distance from ground pilot

is the differential path length to the subarray

Figure 13

- Phase conjugator circuit using modulated (two tone) uplink signal.

When this signal is transmitted through a frequency multiplier with a multiplication factor of q = f




, then the phase of the downlink signal becomes

Thus the output path length differential is the negative of the input path differential, multiplied by a factor of q= f




. For f


= 2450 MHz and f


= 2460 MHz = .995935.

For an array misalignment of )2 = .15E this generates a 383.6 m shift of the beam center on the ground relative to the ideal q = 1 case. With the three pilot antenna arrangement this error is reduced by a factor of 20 to 19.2 m, which is negligible.

The phase of the signal from the r o phase for the entire antenna.

receiver at the frequency 2f


is used as the reference

This signal is used for two purposes.

1. It drives the t o

transmitter system of the A o


2. It drives the 19 conjugators, which belong to the next lower level subarrays in the phase distribution network. The principle of the phase distribution network is exhibited on Figure 14.

The operation of the t o

transmitter system is fairly straightforward (see Figure 14). The 2f

1 signal is taken through a limiter amplifier and a 2 n multiplier. When the actual value of f


= 76.5625 MHz (instead of the nominal 77 MHz)then 2 n = 12. The multiplied signal at frequency f


= 2500 MHz is distributed to all the transmit modules of the A o


With the provided circuitry the signals from all transmitter modules within the A o

subarray are properly phased to the reference signal and are ready to be transmitted to the rectenna.

The phasing of the transmitters at the next lower level of the phase distribution tree is somewhat more complicated. In order to describe this operation the processing of the

Figure 14

- Layout of the reference phase distribution network over the 1 km diameter space antenna.

received signal at the A


subarray must be followed. (See Figures 12 and 14.) The subarray, diplexer and the receiver of this module is identical to the corresponding components of the A o

module, except for small deviations caused by implementation inaccuracies. The output from this r


receiver is used for two purposes.

1. It is sent to the C


conjugator, which is physically located back at subarray A o

2. It is sent to regenerator R


, which is located in the vicinity of the r




The signals from receiver r

1 signal at subarray A




. These signals enter into a coaxial transmission line of approximately

= 250 m long via an i-f diplexer tuned to f a on the other end of the line to separate the bidirectional signals. The signal from receiver r reaches conjugator C


, which also receives a signal from receiver r provides at frequency 2f

1 and also to regenerator R


at frequencies f a

and f b

and f

represent the phase 0/1 of the received b

frequencies. A similar diplexer is used signal carrying the conjugated phase is sent back on line L

1 o

. The conjugator

a conjugation of the phase of the signal from r

. The input circuit of transmitter t


to the t


receiver. The


transmitter complex

is the same as t o



The signal from receiver r

1 conjugator C




also enters into regenerator R


together with the signal from

. By the use of proper combination of these signals the original 0/ io

phase at

frequency is recovered and made available for all the conjugators attached to the A


module. With the selected tree layout the phase regenerator R

1 conjugators (C

11 antenna.

, C




drives a total of 23

, 23). A total of 19 x 23 = 437 regenerators are used in the entire

The described phasing operation is then repeated at the A kR all the transmit signals have the required conjugated phases.

and A kRm

subarray levels until

Note that in the circuit given in Figure 12 the i-f frequency for the conjugators is 2f and 8f



respectively as the phase is distributed to consecutive lower layers of the three.

This method avoids the use of dividers in the circuit thus it is free from ambiguity problems.

, 4f


The essence of the above described example is that the antenna control is based on the use of the retrodirective principle using non ideal phase conjugators for reasons of network simplification. The resultant pointing error is compensated by the use of three pilot transmit antennas on the ground. These antennas allow the change of the phase center of the transmitted beacon wave to the satellite.

Example B - The phase of the transmit elements of the space antenna can be set also on the basis of measuring individually the instantenous phase of each of the transmit elements and then command a differential corrections to each phase shifters in the PA drive circuit on the spacecraft.

Figure 15 shows the simplified block diagram of such a system. This, so-called “Multitone

Phase Computing” system is based on the measurement of the relative phase of each subarray at a selected ground receiver by the use of a signature tone associated with the subarray. The required phase correction of the subarray transmitter is returned by the use of an uplink control channel. Both frequency and time division is used to reduce the complexity and required frequency band for the tones. Approximately 25 MHz bandwidth is needed for a one second phase updating period.

Figure 16 shows a somewhat more detailed block diagram for the implementation of the N tone phase computing system. There is a single telemetry-control antenna in the middle of the rectenna. The telemetry station has an N tone receiver system. The power beam at a spacecraft transmitter is modulated by a low level tone. A typical tone on the ground is received and detected and its phase is compared against the phase of an arbitrarily selectable reference tone. A typical system may use 100 different tones and 100 time division channels for the required 10,000 array phase signatures. After all the tones are received and their relative phase is stored a phase calculator determines the corrections necessary at the phase shifters of each of the spacecraft transmitters. After this information is generated the correction command is transmitted via the command link to the individual

Figure 15

- Principle of multitone phase computing SPS phase control system.

Figure 16

- System block diagram of phase control circuit in phase computing mode of operation using N tone downlink signal.

variable phase shifters on the spacecraft. The accuracy of such a system is determined by the total time necessary for an updating. This time is influenced by the loop delay, information capacity of the TTC channel, time necessary for phase computation and time needed for a phase shifter adjustment. With practically realizable command channel bandwidth an updating time in the order of 1 sec. is realizable. This is expected to be adequate to follow the structural variations of space antennas even in the 1 km diameter category at 2.45 GHz.

5. Conclusions

The basic system configurations and control of characteristics have been reviewed for three specific examples involving spaceborne multibeam. or multielement antennas.

It was shown that the shape of the patterns and their locations relative to the spacecraft can be controlled electronically. In configurations, which are using a feed array and an optical aperture component the control involves the variation of the amplitude and phase on the individual radiating elements of the feed array. In configurations, which use an array of radiating elements or subarrays the control typically involves the phase distribution only.

However, in both cases the actual control is implemented by setting phase controlling elements in the BFN and the amplitude control does not require attenuators. The control can be done in the order of seconds even in fairly complicated circuits and can be implemented on an open loop or closed loop basis. When the phase of the transmitted signal is controlled at the input of the high power amplifier, then no power loss penalty is associated with such control techniques, but a very large number of transmitters are needed for the buildup of significant antenna gain. When the phase and amplitude control is accomplished after the power amplifiers and before the input to the individual radiating elements, then a power loss penalty must be paid for such a control. This loss is increasing with the number of elements in the array.

The control of the discussed type of antennas can be utilized for a large number of practical purposes. These include the change of coverage footprints to allow satellite location changes, follow traffic variations or compensate rain caused attenuation difference. The electronic control of the antenna can be also used to compensate mechanical pointing errors or the effect of structural shape variations.


This paper is based in part on work performed under the sponsorship of the International

Telecommunication Satellite Organization (INTELSAT). Any views expressed are not necessarily those of INTELSAT. The author wishes to express his appreciation for this sponsorship.


Richard Gran

Michael Proise

Alex Zislin

Grumman Aerospace Corporation

Bethpage, New York 11714


Control of large communication satellites becomes most difficult when the structure gets large enough that the structural motion severely impacts the ability to stabilize the RF antennas. This structural/control interaction means that the control engineer can no longer use “benign neglect” of the structural vibrations, but must design a control that has a bandwidth that exceeds the lowest structural vibratory frequency. This in itself is not a problem as long as the sensors and actuators are colocated. Eventually, the antennas have to be controlled independently and the assumption of colocated sensors and actuators is no longer reasonable. This begins the problem. In this paper, the various approaches that have been proposed for controlling large flexible spacecraft when the structural frequencies and the control frequencies overlap will be described. A new approach to the design of such systems will be described, and a reasonably complex example of a large satellite control will be described. The presentation will show a movie that was produced to illustrate the control of this structure and the consequence of using the approach described in the paper.


The communication satellite (comsat) control problem is unique. The specification for these controllers simply demands accurate steady state performance in the presense of random perturbations. No settling time, overshoot or other transient performance criteria are imposed on the controller. Solar induced torques, gravity gradient torques, geomagnetic torques are the environmental disturbances on the satellite that cause the attitude to be perturbed. The sensors have their own inherent noise plus they pick up internal vibration noises on the satellite. The effect of these noises is to torque the satellite in a random way that depends upon the control gains and the type of actuator that is used. All of these disturbances may be very accurately modeled using stochastic modeling techniques. The result is a set of dynamic models that consist of linear systems excited by white noises that can be used to design a control system that has the best possible

response. This “best” response may be achieved by minimizing the uncertainty in the attitude and rate of the rigid body of the spacecraft.

The procedure for determining the best gains is to compute the covariance matrix for the dynamic model of the spacecraft excited by all of the noises that are expected. This covariance will depend on the gains in the control system feedback loops (which are to be determined). The appropriate variances are then minimized to determine the best gains.

The control problem then follows the steps:

• Characterize the dynamics of the satellite by developing linear models of

Rigid body

Momentum exchange devices (or other controller dynamics)



Flexible appendage dynamics

• Specify and characterize the disturbance torques

Solar torques (due to solar array and these are at orbital period)

High frequency solar effect due to variations in solar wind etc.

- magnetic torques created on the spacecraft. In the designs discussed here, these torques are deliberately created to provide magnetic control.

Reaction jet torques created by the thrust uncertainty and c.g. uncertainty

Sensor noises characterized by the power spectral density functions (PSD’s)


• Specify the control configuration, i.e. what will be measured, what gains will be used, and what actuators will be employed. Generally “full state” feedback will be used to provide that control with the best gain and phase margin.

• Find the values of the gains that minimizes the variances of the appropriate dynamic states (attitude or rate).

• Verify the design using the structural dynamics model for appendage and spacecraft vibration modes

Notice that this design procedure does not yield a design that has been dubbed a “modern modal controller” in that the resulting control system simply consists of a set of gains that were selected for the best noise response. This structure is very simple and easy to implement. The modern modal controller on the other hand assumes that the measurements are processed through an optimal filter (a Kalman filter) which determines a best estimate of the spacecraft attitude, rate and other dynamic states. It is the estimated states that are then used for controller feedback. This design has many disadvantages. It is very sensitive to the assumed dynamic model (particularly when the model contains the flexible motion).

The design is usually such that the gain and phase margins are poor. The designs require a model of the spacecraft dynamics be built in the computer (for the filter) so the design is more complex and difficult to implement.

The design approach we have outlined above is described in the following section, and a typical spacecraft design that illustrates this approcah is desribed in the last section.



Consider the stochastic system defined by x0 = A x + B u + c w y = M x + v where: x is the system state vector of dimension n y is the noisy full state measurement of x (i.e. M is nonsingular) w is a unit variance white noise excitation that is independent of v v is the n dimensional white measurement noise with E( v v T )= R S(t) u is the control

A,B, and C are the state, control and noise influence matrices respectively

The covariance matrix P = {( x- E(x) )( x- E(x) ) T satisfies a differential equation


P = A P + P A T + C C T if the control u is not random. If the control system is assumed to be linear, then the control u is given by

(2) u = K y = K (Mx + v).


The control in (3) is clearly random because of the measurment noise v. Thus (2) is no longer the covariance matrix equation. The correct equation is obtained by substituting (3) into (1). Thus we get

P = (A + BKM) P + P (A+BKM) T + BKRK T B T + CC T (4)

If equation (4) is written in such a way that the quadratic in K is made explicit on the right hand side, the result is

P (B K + PM T R -1 R (BK + PM T R

+ A P + P A T + CC T - PM



) T

-1 MP .

To determine the gain that minimizes any linear combination of components of P is now easy since the quadratic term in K that appears on the right hand side of (5) is always


greater than or equal to zero (since R is always positive definite). Thus the smallest that P can be is when the quadratic is zero (P when P

small or equivalently the smallest P in steady state

is zero). Hence the optimal minimizing gain is given by

(6) B K = - PM




or K = - (B









-1 nd the covariance matrix P is the solution of the Ricatti equation

P = A P + P A


+ CC


- PM




MF (7)

It is obvious that this solution is not what one would get were one optimizing a performance measure of the form

(8) because the solution to (8) with the dynamics (1) is given by u = K' xˆ where K' is not the same as the K in (6) and xˆ is the optimal estimate of x obtained from the Kalman filter estimate of the state x. In the implementation we are proposing here, the control does not correspond to t he

performance index (8) at all - in fact there is no performance index like

(8) that gives the same control as was derived in (6). This is partly because of the pseudo inverse usedin deriving the optimal gain in (6) (i.e. the (B






term) and partly because any attempt to leave out the control weight R in (8) leads to infinite control gains (in order to make (8) correspond to a minimization of P , the covariance matrix, requires that R be zero).

Because the introduction of a filter into the control loop of an optimal system destroys the gain and phase margins that the optimal design has (infinite gain margin and at least 60E of phase margin) it is clearly more desirable to utilize the optimal stochastic control gains given by (6). In so doing, a comsat controller that leaves structural dynamics out of the design will have the property that the added phase lags introduced by structural frequencies close to or within the control band will not cause loss of stability. The next section gives an example of a design using this approach.



The control system considered for a typical comsat uses three axes stabilization with three orthogonal reaction wheels. The attitude is measured using a horizon sensor and the angular rates sensed by three orthogonal rate gyros. A magnetic coil around the solar array is used to eliminate solar pressure torques and jets are incorporated for station keeping.

The control configuration is shown in Fig. 1 (a single axis); it uses full state feedback

(position and rate) with the feedback gains chosen so as to minimize the rms angular rate and attitude. This is a classical approach to the problem formulated above when we are minimizing both disturbance torque and sensor noise effects while maintaining gain and phase margin that makes the design insensitive to parameter variations.

Since in its basic mode of operation the geostationary communication satellite requires no rapid slewing maneuvers and associated settling time requirements, the control problem is one of maintaining the attitude against very low frequency disturbances, such as solar, gravity gradient, magnetic and thermal torques, as well as the noises introduced by the horizon sensors and gyros. To compute these disturbances we use the transfer functions between the vehicle rate and the disturbances and sensor noises given by



Using an assumed white noise characteristic for the attitude N

A the resulting satellite stability, on a one-sigma basis is given by

and gyro noise N


in (9),


From Eq. (10), the satellite peak LOS rate levels resulting from cyclic torque disturbances

(orbital rate) becomes


The LOS attitude and rate errors are plotted in Fig. 2 as a function of the control system bandpass (T c

) for given attitude and gyro noise levels, and different torque disturbances to vehicle inertia ratios (T dmax

/I v

). Note that as the bandpass T c

increases, the peak LOS errors due to torque disturbance decreases while the errors due to gyro and attitude sensor noise increases. Using the nominal gyro noise level and the attitude noise and disturbance torque to inertia ratio results in an optimum closed loop bandpass range of 0.004 to 0.04

rad/sec. Also shown on the figure is a lower limit on bandpass of 0.314 rad/sec (0.05 Hz) to prevent interaction of the control loop dynamics with any structural modes.

The net result of this simplified conceptual design is a specification for the bandpass of the control loop of 0.04 rad/sec. This results in a 0.20 deg/hr peak rate stability error and

0.008 deg peak attitude error. This conceptual design considers only a single axis. We extended the design to include all three axes with a magnetic unloading loop. Also included in this analysis is the solar array dynamics, the sloshing of the RCS propellant, and the solar, magnetic, and gravity gradient disturbances. To determine how large the solar array can be before control performance is degraded, a simulation was used that, contained all of the nonlinear actuator and sensor characteristics.

The primary torque disturbance acting on the satellite is due to solar pressure on an assumed 600 ft 2 solar array area with a large imbalance moment arm of approximately

0.1 ft. The resulting torque disturbance (1x10


ft-lb) is cyclic at orbital rate acting about the X (roll) and Z (yaw) axes out of phase by 90 deg. This cyclic torque results in every increasing cyclic momentum storage requirement which must be unloaded over the 9-year mission period. An air core magnetic unloading system has been identified as a baseline approach for this application. Magnetic unloading can have a significant effect, however, on stability of the spacecraft because of undesirable cross coupling between the coil’s magnetic dipole and the earth’s field which cannot be compensated for during any given part of one orbit. Over a long period, however, (one orbit or more) the average reaction

wheel speeds will be kept within acceptable levels. Another problem which further

complicates this problem in geostationary orbit is the relatively low levels of the earth’s field (< 120 gamma). This makes it extremely difficult to sense the level of this field in the presence of the solar array dipole and other ferro-magnetic equipment distributed throughout the vehicle. For these reasons a magnetometer is not part of the control configuration for the unloading system. The exclusion of a magnetometer would not be critical if the earth’s field could be predicted accurately, as it can during quiet periods of the sun. It is during solar storms that the problem must be addressed. An analysis was therefore performed to determine the effect that solar quiet and storm periods have on the stability of the satellite.

A closed loop optimal control was developed that used reaction wheels and magnetic unloading as a combined controller set with a simplified quiet days earth’s field model.

This design minimized a cost function defined by (13) for a set of linear first-order differential equations given by Eqs. (14) and (15) (the A matrix and B matrix for this design are shown in Fig. 3). Notice that this design is a more detailed version of the conceptual design above in that it includes the magnetic control and all three axes coupled together.

(13) x = Ax + Bu u = Kx

(14) where: j = cost function which was selected to attempt to match the bandwidth specified.

x = state vector (12x1) - the first six states are the rigid body pos. and rate and the second six are the wheel speeds and mag. loop cur.

u = contol vector (6x1)

Q = weighting matrix on states (12x12)

R= weighting matrix on controls (6x6)

A,B = matrices which define system dynamics (Fig. 3)

K = gain matrix for optimally controlling the satellite ( 6x 12).

With the resulting optimal gain matrix (K) obtained from this analysis, a covariance analysis was then performed to determine the variance on each of the system states (x) for a combined set of random disturbance levels on the system (solar torques) and measurement noises (attitude and gyro noise levels).

The resulting closed loop characteristics and one sigma variances on the system parameters are presented in Fig. 4. Significant points to be made about these results are:

• Reaction wheel closed loop bandpass is below 0.05 rad/sec as identified in the conceptual design above

• Combined LOS (pitch, roll, yaw ) RSS error does not exceed 0.015 deg (3F) well below the 0.10 deg allowable

• Wheel momentum storage levels can be held to below 3 ft-lb-sec (1F) on an indefinite basis. This value compares favorably with the 15 ft-lb-sec storage capacity that would be required after two orbits if no magnetic unloading was provided.

Our preliminary analysis included the solar array modes for a two wing array configuration. This array had an estimated first mode frequency of 0.25 Hz. To verify the frequencies estimated by simplified preloaded string and beam analysis, the array was modelled in NASTRAN (Fig. 5). Central mast stiffness and mass and blanket mass data were obtained from the manufacture specification for the solar array. With a 20 lb preload, modes were calculated and are summarized in Fig. 5). The lower modes are primarily due to blanket motion rather than beam bending. Frequencies are lower than those predicted by pretensioned string theory due to edge flapping of the blanket.

Figure 6 represents a three-axis time history digital simulation run that demonstrates the ability of the attitude control design to hold the accuracies required over an orbit. The simulation included effects such as reaction wheel friction, gyro and attitude noise, solar disturbance torques, magnetic unloading, solar array structural bending and momentum cross coupling due to orbital rotation rate.

A three axis satellite simulation program called SATSIM was used in the final design verification. The time history of the attitudes and the momentum wheel angular momenta are shown in Fig. 6. Also shown is the disturbance torques that perturb the satellite over an orbit. The attitude traces clearly show the noises in the simulation, and the fact that the maximum rates are well below the required pointing rates and attitude requirements. To determine the effect of solar array modes that move into the control passband, the solar array size was altered in such a way that the first mode frequencies moved well within the control passband. The result of a simulation of this parameter variation showed that the control system remained stable with little or no performance degradation even when the modes were two orders of magnitude smaller in frequency. This result is a consequence of the colocation of the sensor and actuator set. From the point of view of classical control, such colocation means that the poles and zeros of the infinite number of structural modes

must alternate on the imaginary axis (when there is no damping). This alternation means that if full state feedback is used on the rigid body their is no way for the gains to cause the root locus to move into the right half plane. Thus the infinite gain margin property of the optimal design carries through when infinite flexible modes are added.


In a study that we performed for NASA Marshall Space Flight Center (Ref.1), we showed that the problem of control of the flexible motion of a large space structure, whose dimensions were on the order of 100 meters, is possible. This result may be easily applied to communication satellites of the size that would be expected in the 1990’s. The key conclusion of that study, as well as of this, is that the control problem is best handled by measuring as many states as possible and by using “full state feedback” from each of the measurements. In this study we have shown that such a design approach yield significant dividends in the “robustness” of the resulting design (i.e. gain and phase margins). In the

MSFC study we also concluded that such an approach gives reasonable insensitivity to structural dynamics parameters even when sensors are dispersed about the structure (in that study, 52 strain gages were used to determine the motions at 52 dispersed points on.the structure). As part of the presentation accompanying this paper, a movie of the

MSFC structure was shown. This movie shows both discrete and continuous control systems designed for that study.


1. Gran, R., Rossi, M. and Moyer H. “Optimal Digital Control of Large Space

Structures” AAS Journal April 1979 .

Fig.4 Results of Covariance Analysis and Closed Loop Poles

Pole Locations - Reaction Wheel Loops -0.093,-0.091,-0.04,-0.037,-0.043± 0.043j

- Magnetic Unloading Loops -0.00023 ± 0.000245j and -0.00011


Attittude (deg.)

Attitude Rate


Wheel Momentum


Uncertainty in Various States (rms)

Roll Pitch











Fig. 5 Results of NASTRAN Analysis of Solar Array and the First Four Modes

These modes, mode shapes and mode masses were used in the SATSIM

Fig. 6 Results of Satellite Simulation (SATSIM) for the Comsat Design

Future Modular Data Handling Concepts for

Large Space Platforms

G. P. Thompson

ESA/ESTEC, The Netherlands


With the ever-increasing competition for geostationary orbital positions, designers of applications satellites are looking more and more to the use of large multipurpose platforms to fly simultaneously a variety of payloads. This paper looks at the challenges these platforms provide in the field of data handling for individually launched and docking satellites, and reviews current European concepts and technologies which are being developed to meet these challenges.


J. D. Barry

Hughes Aircraft Company

Space and Communications Group

Los Angeles, California 90009


Intersatellite optical communication systems may be based on a number of conventional laser sources allowing wavelength and operating parameter selection, and system optimization to a particular satellite system. Laser sources which are presently available for satellite use, due to the current status of technical and engineering developments, include the HeNe gas laser operating at 632.8 nm, the GaAlAs diode laser operating at 800 to 900 nm, the Nd:YAG laser operating at 1064.2 nm and 532.1 nm, and the CO


laser operating near 10600 nm. The applicability of each laser source to a satellite communication system is determined by on orbit requirements and attitude stability of the satellite, by the performance characteristics and operating needs of the laser source and its associated support equipment, and by the optical propagation properties of the laser system. Two general levels of the optical communication must be considered; initial satellite to satellite acquisition and possible reacquisition, and the transmission of data at the required rate. The general characteristics and system properties must evolve from the interrelated factors and occasionally contradictory requirements. It is the purpose of this presentation to provide insight into the applicability of a laser communication system to space and to indicate the dominating factors in the selection of a laser system.



Lester F. Eastwood, Jr., Ph.D., John A. Maynard, and Samuel I. Green, Ph.D.

McDonnell Douglas Astronautics Company

McDonnell Douglas Corporation

St. Louis, Missouri


This paper describes principles of direct detection laser communications link design.

As background, it reviews the status of communication systems operating at

McDonnell Douglas. It summarizes key modulation techniques, describing how they have been implemented at data rates up to one gigabit per second (Gbps) and how minor component changes can support eight Gbps. The paper describes, in detail, methods for calculating performance of important direct detection modulation schemes: pulse gated, pulse polarization, and pulse delayed binary modulations; pulse quaternary modulation; and pulse interval modulation. It also compares results of these calculations with measurements. Finally, it presents an example of end-toend communications link analysis.


John D. Wolf, James A. Pautler, Robert Z. Olshaw, and

Lester F. Eastwood, Jr., Ph.D.

McDonnell Douglas Astronautics Company

McDonnell Douglas Corporation

St. Louis, Missouri


This paper describes key design issues and tradeoffs in Nd:YAG laser communications system design. As background, it reviews missions and applications contemplated for

Nd:YAG technology. It also briefly summarizes receiver operation, focusing on the primary beamsteering functions that enable accurate pointing of the very narrow communications beam (5 µrad for a one gigabit per second link). The remainder of the paper considers Nd:YAG lasercom system design methods. The paper reviews typical system specifications (e.g. data rate, margin, and acquisition time) and constraints (such as platform dynamics and laser capabilities). It shows how the designer makes tradeoffs among tracking accuracy, control bandwidths, fields of view, signal to noise ratio and transmitted beamwidth to iterate to a final choice of design parameters, and it describes examples of resulting designs.




Frank E. Goodwin

Hughes Aircraft Company

El Segundo, California


The use of optical heterodyne detection in a communication system requires a local oscillator laser beam to be coincident with the incoming signal on the detector. After detection, the signal behaves in every way like a classical microwave or radio signal which has been detected with a heterodyne receiver. This discussion of the use of optical heterodyne detection in laser communications thus includes consideration of modulation formats as well as the special geometrical requirements of combining the local oscillator and signal. Modulation formats of interest are amplitude modulation, frequency modulation and phase modulation, and both heterodyne and homodyne detection techniques are considered.

The physical and geometric treatment of optical heterodyne detection is given. General equations are derived for the signal to noise ratio of a coherent receiver in terms of the distribution functions of the signal and local oscillator fields and the size of the detector.

The most efficient local oscillator field distribution function is when it exactly matches that of the signal field distribution over the detector surface. A special case of interest is when the signal field is an Airy function and the local oscillator field is uniform. This special case is shown to be feasible with a small penalty in heterodyne mixing efficiency.

An analysis of the heterodyne NEP includes factors from geometrical mixing efficiency, thermal noise, dark current, and electrical load mismatch. The degree of degradation is then a function of the amount of local oscillator power. Practical limits on heterodyne

NEP's are established.




Arthur J. Einhorn

Hughes Aircraft Company

Space and Communications Group

Los Angeles, California 90009


Future demands for more information transfer in space will increase the demand for wider space communications with narrower beamwidths. Both of these requirements suggest using as high a carrier frequency as possible. The wider bandwidth is necessary to increase the information carrying capacity. The narrower beam allows the use of a smaller amount of transmitter power to a distant receiver and concomittenly results in larger antenna gains.

The Carbon dioxide (CO


) laser system meets many requirements for a space qualifiable space communication such as wideband modulation capability, potentially long life and reliability. In this paper we discuss several aspects of the design of a CO


laser communication system for space application in terms of overall functional requirements, system tradeoffs and subsystem component selection. The results require use of state-ofthe-art components which are or can be space qualified.



C. D. Arndt

NASA-Johnson Space Center

Houston, Texas


The concept of using large satellites in neosynchronous orbit to collect solar energy for earth use, first proposed in 1968, is being evaluated by the Department of Energy and the

National Aeronautics and Space Administration. A reference system has been defined to provide a basis for evaluating alternate technical approaches and for assessing environmental impacts. This reference system is described with emphasis on the microwave subsystems and possible alternatives. Other considerations, including study guidelines, system sizing tradeoffs, mass and cost projections, and environmental factors, are discussed. An outline of a ground-based exploratory development program which could follow the present evaluation study (now scheduled for completion in 1980) and answer key technological issues is given. Several space technology projects as further steps towards develoning an SPS system are also discussed.


The solar power satellite (SPS) system first proposed in 1968 by Dr. Peter Glaser is envisioned as a means of using extraterrestrial solar energy to supply power for earth use.

This concept has a large satellite in geosynchronous orbit converting, solar energy to electrical energy either by a photovoltaic process or by solar thermal conversion. The energy is transmitted to the earth by a high-power microwave transmission system. This summary paper describes the present reference SPS configuration, the projected masses, alternative configurations, and some details of a proposed ground-based research program.

The SPS concept is now being evaluated under a joint Department of Energy (DOE)/

National Aeronautics and Space Administration (NASA) program which has four major study areas:

1. System Definition

2. Evaluation of Environmental, Health, and Safety Factors

3. Evaluation of Socioeconomic Issues

4. Comparative Assessment of Alternative Energy Systems

Based upon the results from these studies a decision will be made in 1980 as to whether the solar power satellite program should be continued, and if so, in what manner.

The environmental assessment studies would be included with emphasis on specifically defining the environmental factors and limitations.

A set of evaluation guidelines were established based upon the results from earlier SPS system studies. These guidelines, used primarily in sizing a reference SPS configuration and determining its costs, are as follows:

• Construction materials obtained totally from earth resources

• Geosynchronous orbit

• Microwave system operating frequency - 2.45 GHz (center of a 100 MHz industrial, medical and scientific band)

• Microwave power density not to exceed 23 mw/cm

• Microwave system sized to provide 5 GW output from the rectenna

• Satellite implementation rate of two 5-GW satellite systems per year

• Thirty (30) year satellite lifetime

2 in the ionosphere


A reference system configuration has been defined to provide a basis for evaluating alternate technical approaches and as a guideline for assessing environmental impacts. This configuration, developed through both NASA in-house and contractor system studies, is not intended as a final design. Alternative concepts will also be pursued. The reference system as documented in reference 1 has the following general characteristics:

Generating Capacity

Each solar power satellite system sized for 5 GW D.C. power output

Operational Configuration

Flat solar array with transmitting antenna on one end

Orientation - antenna main rotational axis perpendicular to orbit plane

Energy Conversion-Two Baseline Options

Single-crystal gallium aluminum arsenide Photovoltaic - CR = 2

Single-crystal silicon photovoltaic - CR = 1

Microwave Power Transmission

2.45 GHz operating frequency

One-kilometer diameter planar phased-array transmitting antenna

Klystron power amplifiers

Slotted waveguide radiating elements

Active retrodirective phase control

Structural Materials

Graphite-fiber reinforced thermoplastic

Attitude Control

Argon ion electric thrusters panels

Array Power Distribution

Thin sheet conductors

40 kv to microwave power amplifiers

The SPS sizing of a 1-kilometer transmit antenna and 5 gigawatts (GW) of DC power output at the rectenna was based upon:

(1) achieving a maximum output power due to cost effectiveness and efficient reception by the rectenna.

(2) a thermal limitation of 22 kilowatts per square meter in the transmit array, and

(3) a peak power density of 23 milliwatts per square centimeter in the ionosphere to ensure no nonlinear heating of the ionosphere.

From thermal considerations larger antennas are desirable while the ionospheric limitation is satisfied by smaller antennas. Tradeoffs in antenna size and power-handling capability are summarized in Figure 1. After the thermal and ionospheric limits are adjusted to coincide, the final SPS sizing of a 1 Km antenna and 5 GW output power are obtained.

The solar array sizing is determined by the efficiency chain of the various elements in the system. Figure 2 shows the end-to-end efficiency chain for the GaAlAs and silicon solar cell options. A satellite providing 5 GW out of useful power at the ground and an overall efficiency of approximately 7% must intercept about 70 GW of solar energy.

Energy Conversion

A number of different energy conversion techniques were investigated. These included silicon and gallium arsenide photovoltaics with concentration ratios of 1 and 2, thermal conversion using rankin cycle turbines with potassium as the working fluid and Brayton systems using helium gas, and thermionic systems. The overall conversion efficiency, satellite area, and mass are comparable for the thermal and, the photovoltaic systems.

However the photovoltaic systems have the advantages of simplicity in space construction and higher reliability due to the lack of moving parts. The thermal systems also have the disadvantage of requiring large movable solar reflectors to concentrate the sunlight.

Two photovoltaic systems, silicon with a concentration ratio (CR) of one and GapAlAs

(CR=2) were chosen for the reference energy conversion devices. The silicon cell option has no concentration (CR=l) with a blanket area of 52.3 Km 2 as shown in Figure 3. A tradeoff analysis indicated this simplier no-concentration configuration for silicon is cost effective due to a reduction in conversion efficiency at higher operating temperatures,

increases in structural requirements for stretching the solar reflectors, and degradations in solar cell performance due to shadowing in the solar reflectors. The silicon blankets consists of 50-micrometer (2-mil), thick single silicon solar cells with an efficiency of

17.3% at 25EC. At the design operating temperature of 36.5EC, the efficiency drops to

16.5%. Details of the silicon cell are shown in Figure 4.

A problem associated with the silicon cells is degradation due to electron radiation at geosynchronous orbit associated with solar flares. Solar flares strong enough to create radiation damage in the silicon cells will occur a few times over the thirty year lifetime of the satellite. A laser annealing system usinq CO


lasers mounted on an overhead gantry structure can heat the cells to approximately 500EC. This temperature is sufficient to anneal out the radiation induced degradations to the cells while still being low enough to prevent damaging the cell interconnects and substrate materials. Annealing tests are underway to determine the cell recovery characteristics.

The GaAlAs option is shown in the array configuration in Figure 3 and the cell desion in

Figure 4. A five-trough arrangement provides a concentration ratio of two which reduces the amount of gallium required. The solar cell has a 20% efficiency at 28EC, and 18.2% efficiency at the design operating temperature of 125EC. At 125EC, self-annealing of radiation damage occurs within the cells, so there is no need for a laser annealing system.

Gallium arsenside cells have higher efficiency than silicon, are less susceptible to radiation damage, and can operate at the higher temperatures associated with a concentration ratio of 2 or greater. A problem with these cells which has not been fully resolved is the availability of gallium.

Microwave Power Transmission System

The reference satellite configuration has a 2.45 GHz microwave system for transmitting the power down to the earth as shown in Figure 5. A 1 Km diameter phased array antenna with a 10-decibel (db) Gaussian taper illumination focuses the beam at the center of the ground antenna/rectifying system (rectenna). This power beam has approximately 88% of its energy within a 5 Km radius from boresight, with a resultant beam width of 1.2 arcminutes. Mechanically the antenna is divided into 7220 subarrays, approximately 10 meters X 10 meters on a side, having slotted waveguides as the radiating surface. The DC-

RF power converters, i.e., 70 KW klystrons, are mounted on the backside of the subarrays.

A retrodirective phasing scheme is used to provide phase information to each subarray or each power module (antenna area associated with an individual klystron). This system has a pilot beam transmitter at the center of the rectenna and RF receivers/phasing electronics at each subarray or at power module to process the pilot beam phasing signal. The characteristics and error tolerances of the microwave power transmission system may be summarized below:

Frequency - 2.45 GHz

Output Power to Power Grid - 5 GW D.C.

Transmit Array Size - 1 Km. Diameter

Power Radiated from Transmit Array - 6.85 GW

MPTS Efficiency - 63%

Array Aperature Illumination - 10 step, truncated gaussian amplitude distribution with

10 db edge taper

Error Budget:

Total RMS phase error for each subarray = 10E

Maximum mean phase error at edge of transmit array = 2E

Amplitude tolerance across subarray = ± 1 decibel

Failure rate of dc-rf power converter tubes = 2% (a maximum of 2% failed at any one time)

Antenna/Subarray Mechanical Alignment:

± 3 arc-minutes th the grating lobes constrained to # .01 mW/cm


Subarray Size: 108m


Number of Subarrays: 7220

The antenna pattern at the rectenna for the 10 dB Gaussian taper reference system is shown in Figure 6. The antenna illumination function was optimized to provide the maximur amount of RF power intercepted by the ground rectenna (88%) and to minimize the sidelobe levels. Other illumination functions were investigated, includling uniform, various Gaussian tapers, cosine on a pedestal, quadratic and inflected Bessel, but the

10 dB Gaussian taper provided the best overall performance after considering the maximum power density constraints in the transmit array and rectenna. Studies are now underway on the feasibility of producing a flattened, widebeam pattern at the ground by using a continuously variable phase reference across the transmit array. This technique requires a much larger antenna which is amenable to the solid-state configuration to be discussed later.

Grating lobes will also be present; the intervals or spacings between the lobes are dependent upon the subarray or power module sizes and the amplitudes are a function of the mechanical alinement of the antenna and subarrays. The grating lobes are stationary and do not spatially move with misalignment changes. The locations of the grating lobes for a main beam rectenna site in the central United States are shown in Figure 7. The spacings are determined by the phase control system.; if phase conjugation is performed at the power module (tube) level, then the grating lobe peaks are as shown in Figure 7.

However if phasing is performed at the subarray level , (10.4 meters X 10.4 meters), then grating lobes occur at 440 Km intervals. Based upon an environmental constraint of having grating lobe peaks less than .01 mw/cm


, the mechanical misalinement of the 1 Km array

has to be less than 3 arc-minute if phasing is performed at the power module level and 1 arc-minute if phasing is done at the subarray level.

Rectenna - The around rectenna, which receives and rectifies the nower beam, has halfwave dipoles feeding Schottky barrier diodes. The rectenna is a series of serrated panels perpendicular to the incident beam and covers approximately 75 square kilometers. The panels have a wire mesh screen for a ground plane with 75-80% optical transparency. This mesh is mounted on a steel framing structure, supported by steel columns in concrete footings. The DC electrical power from the diodes is transferred via dedicated aluminum conductors into 40 megawatt DC/AC converters. The power is then transmitted to 200 megawatt transformer stations where the voltage is stepped up to 230 KV, then collected in 100 megawatt groups for interface with the outside transmission system for commercial usage.

Space Transportation

The space transporation system required for the SPS program has four types of vehicles.

• Heavy Lift Launch Vehicle (HLLV)

• Electric Cargo Orbit Transfer Vehicle (COTV)

• Personnel Launch Vehicle (PLV)

• Personnel Orbit Transfer Vehicle (POTV)

The heavy lift launch vehicle (HLLV) is a two-stage, vertical takeoff, horizontal landing, fully reusable winged launch vehicle. It has a gross liftoff weight of 11,000 metric tons with a payload to low earth orbit (LEO) of 425 metric tons. This vehicle, as shown in

Figure 8, uses methane and liquid oxygen in the first stage (booster) and liquid hyrodgen and oxygen in the second stage (orbiter). The booster also has an airbreather propulsion system (aircraft jet engine) for a flyback capability to the launch site. At the present time the Kennedy Space Center is expected to be the launch site.

The COTV moves SPS construction material and cargo from the LEO staging area to the

GEO construction site. It is also fully reusable vehicle with the main source of propulsion being electric power from solar arrays driving argon ion thrusters. There are two electric power options - GaAlAs and Si photovoltaic arrays. Both types of solar cells will suffer radiation damage during each trip through the Van Allen belts and will require annealing.

Supplementary chemical thrusters will be available to provide power during times when the solar arrays are shadowed. One scenerio of a typical COTV transfer to GEO and return is shown in Figure 9.

The personnel launch vehicle (PLV) transports personnel and priority cargo from the launch site on earth to LEO. The vehicle, derived from the present space shuttle system.

has a 75-passenger capacity personnel compartment in the orbiter payload bay. It has a

winged liquid propellant fly-back booster instead of the present solid-state booster. The

POTV carries personnel and priority cargo from the LEO to the GEO construction base.

This vehicle can transport 160 personnel in a passenger module, 480 man-months of consumables in a resupply module, and a two-man flight crew in a trip time of approximately one day. The propellants are liquid hydrogen and oxygen for the two common stage vehicle. The transporation sequences for both personnel and cargo from ground : LEO and LEO : GEO are summarized in Figure 10 using the four types of space vehicles.


Primary construction of the satellite is slated for geosynchronous orbit with a LEO staging base. A number of system studies evaluated LEO versus GEO construction bases, with each orbital location having advantages. The GEO location has continuous sunlight, the construction base can be optimized to build the satellite in its final form, and there is no need to transport large completed segments of the satellite from LEO to GEO. The LEO location has a more benign space environment with solar radiation particles being shielded the Van Allen Belt but has the problem of gravity gradient torques on the large satellite modules (the satellite was divided into eight modules for ease of construction and transportation). As stated previously the GEO site has been selected for the reference system report. The construction time for a satellite is six months, using a 550-man construction and support crew.

In order for the satellite to be built quickly and efficiently the construction process must be automated. Construction materials will be packaged in a very dense form to reduce transportation costs. Automated beam builders will take the rolls of construction materials and extrude long structural beams. These beam builders as shown in Figure 11 are not unlike the simple commercial aluminum gutter fabricators used today. the open-truss structure of the satellite is designed with regular uniform cross-section beams to expedite automatic fabrication. The basic structural element of the satellite has a triangular crosssection and is made of a graphite composite material for the antenna which requires a surface flatness of less than l arc-minute. The solar array does not have the stringent surface tolerance and may be constructed of aluminum beams.

The diagram of a satellite with the silicon solar cell array is shown in Figure 12. The antenna has a primary structure, 130 meters deep, with an octagonal shape over 1000 meters in width and length. The secondary structure is a cubic truss, approximately 10 meters in depth, which provides support for the microwave subarrays. These subarrays may be aligned using Az-El mounts to maintain a 3 arc-minute random flatness.


The satellite mass properties may be separated into three areas: solar array, microwave antenna, and solar array-antenna interface (rotary joint, sliprings and antenna yoke). A summary of these masses for a 5 GW satellite is as follows;

CaAlAs Array Silicon Array

(Millions of Kilograms)

Solar Array

Primary Structure

Secondary Structure

Solar Blankets


Power Distribution, Altitude

Control, etc.

Microwave Antenna

Primary Structure

Secondary Structure

Transmitter Subarrays

Power Distribution

Thermal Control

Information Management

Altitude Control





















Array-Antenna Interface


25% Contingency







Total 34.1


The two largest mass contributors are the solar blankets in the solar array and the klystron/waveguide subarrays in the antenna. The large antenna and solar array structures are lightweight and comprise only 12-20% of the total satellite mass.


The question arises - how does the electricity costs from an SPS system compare with the costs from a conventional power plant? The answer is that any comparison depends mainly

upon the projected fuel cost of a conventional system. The cost data shown in Figure 13 are for an average U.S. plant in 1975. Assume the terrestrial transmission, distribution, and operating and maintenance (O&M) costs are the same (1.24¢ per kwh) for the conventional plant and the SPS system. The conventional plant investment cost is less than

14¢ per kwh of electricity, with the remaining expense being fuel costs. If we project a 5% per year increase in fuel, which is considerably less than the 18.7% per year average increase for the 1970-1975 time frame, then the rates for the two systems will be approximately the same (5.5¢ per kwh) in the year 2000. Larger fuel expenses will vary the conventional systems costs accordingly. The point is that the cost of electricity from

SPS systems should compare favorably with those costs from other energy sources.


While the reference system has been defined in some detail, other alternative power transmission systems are presently being studied. These alternate concepts may be summarized as follows:

DC-RF Converters

• Solid state - The main advantage of low-power solid-state devices is reliability, which should be several orders of magnitude greater than klystrons. Two configurations are being studied - (a) a conventional antenna with the solid-state power converters mounted on the backside, and (b) a sandwich concept which integrates the solar array with the antenna.

• Magnetrons - A scheme of injection-locked magnetrons with the heaters turned off to reduce noise is being studied.

Phase Control System

• A ground-based phase control system, with the principal advantage of reducing spacecraft hardware complexity associated with the present phase conjugation system, is being evaluated. The error budget for the phase reference distribution system within the 1 Km array is now limited to less than 1E.

Laser Transmission

• New systems, including the free electron laser, the electric discharge laser, and optical pumped laser, could possibly replace the microwave power transmission system. The advantage of lasers is a greatly reduced ground receiver size while a disadvantage is the low conversion efficiency.


A ground-based exploratory development program which could follow the present evaluation study now scheduled for completion in 1980 would focus on technology programs to provide definitive information on critical issues. This program would emphasize laboratory testing in key technical areas. The systems definition studies have provided guidelines and performance criteria; the technology development program would ascertain the feasibility and assessment of these requirements. Some typical areas which might be explored include:

1. System Definition Studies

• Preferred Concept Definition - update reference system

• Technology Program Impacts - integration of test and analytical results

• Societal/Environmental/Comparative Assessment Impacts - integration of DOE study results

• Alternative Concepts - solid state, lasers, etc.

• Flight Project Definition - define follow-on space tasks

2. Solar Energy Conversion

• Solar Array Resources - develop resources recovery techniques

• Solar Cell Studies - high efficient Si and GaAlAs cells, annealing techniques

• Cell and Blanket Testing - ground and space testing of solar cells

• Blanket and Concentrator Development - develop and test GaAlAs and Si blankets

3. Integrated Microwave System

• Power Amplifier Development - klystron, solid state

• Integrated System Performance (Tube/Solid State) - define test requirements and conduct tests on integrated tube/phase control/subarray configuration

• Phase Control System - develop and test phase control system using either phase conjugation or ground-based system )or hybrid)

• Transmit Antenna Performance - slotted waveguide performance, materials selection, and subarray development

• Rectenna Element - conversion efficiency, RFI characteristics, manufacturability of dipole/diode (and/or alternate configurations)

Since this is a microwave conference, a brief review of the microwave system requirements would be appropriate: DC-RF conversion efficiency of •85%, rectenna collection efficiency = 88%, rectenna conversion efficiency = 89%, waveguide I one time.

2 R loss


2%, phase error budget of 10ERMS, noise output # 2% failed power converters at any

4. Space Construction

• Operations and Support Functions - large system berthing, assess construction support capabilities.

• Structure Fabrication - prototypes of beam builder and joiner, assembly techniques

• System Installation - concepts for solar blanket, concentrator and conductor packaging, rotary joint installation, antenna/subarray integration.

5. Transportation

• Rocket Engine Investigations - engine definition, design, component evaluation, critical technology assessment

• Thermostructure/Cryo Insulation Investigations - materials development

• Electric Propulsion - design goals and performance criteria, subscale propulsion system development and testing

• Ballistic Booster Recovery - configuration design and testing

• Vehicle Design/Analyses - analysis procedures for vehicle design

• Operations - packaging concepts, launch site operations, payload handling

• Airbreathing Engine Technology - multicycle engine performance and testing

6. Electrical Power Distribution

• Power Processing - prototype of SPS power distribution system

• Switchgear - prototype development

• Rotary Joint - subscale design and testing

• Power Conductors - thin sheet conductor develcpment

• Energy Storage - battery storage for powered-down periods during solar eclipses

• Integrated System Modeling - computer simulations and evaluation of test results

• Space Environmental Impacts - spacecraft charging, plasma interactions, insulation degradation and arcing

7. Structure/Control and Materials

• Structural/Control Studies - structural analysis, coupling modes through rotary joint, flight control systems and performance

• Materials - lifetime characteristics in geosynchronous environment, outgassing, radiation damage.


If one assumes the technology development program successfully answers the key issues and prototype hardware is available, a series of space projects would be the next logical step in the SPS program. A number of space projects have already been suggested including:

• Component Testing Using the Shuttle

This would encompass small “suitcase” experiments such as operation of a prototype SPS klystron on a slotted waveguide radiator. Plasma effects in the

LEO environment could be studied.

• LEO-LEO Microwave System Tests -

A microwave system test involving a 3 meter X 45 meter transmit array and a beam mapping satellite separated by 16 Km has been proposed. The transmit array has two configurations: one, with nine subscale subarrays in a 9 meter X

9 meter arrangement for thermal testing and two, in a 45 meter linear array with

15 subarrays to evaluate phase control system performance. The input power requirement to the tubes is 300-400 kW, depending upon the test configuration.

The power is supplied from a solar array which also provides subscale testing of the SPS solar blanket configuration.

• Inverted Ground-GEO Test -

Verification of the phase control system to electronically form and steer the microwave beam with its required accuracy (±200 meters) would be the goal of an inverted test as shown in Figure 14. The pilot beam, normally transmitted from the ground to the satellite, would now be transmitted from geosynchronous orbit to the ground. A narrow, 1 Km long prototype phased array would receive the phasing signal on the ground and transmit a fan-shaped beam directed to the pilot beam satellite. A small beam-mapping satellite in geosynchronous orbit would measure the uplink beam pattern from the linear array. In addition to a separate ground-based heating facility would be utilized to heat the ionosphere to simulate the operational SPS environmental conditions. This inverted test can be accomplished at a much reduced cost due to the small amount of hardware in space. There is however one problem which is now being studied. Due to Fresnel zone lens effects of the ionosphere on the downlink pilot beam signal, mit may not be possible to obtain independent phase measurements at the 1 Km array on the ground. If the measurements are not independent, then the phase control system cannot provide conjugation of the correct phase and the beam pattern would not be properly formed. This problem recently arose in the test planning of phase scintillation measurements through the ionosphere from existing geosynchronous satellites.

• Demonstrator -

The last project prior to a full-scale SPS commitment could be a

“demonstrator” which provides continuous power from geosynchronous orbit to a small rectenna system. These are varied opinions as to the objectives and

the configurations of a demonstrator. However some of the objectives can be listed as follows:

• Demonstrate power transmission, space-to-ground

• Demonstrate a complete subscale system

Verify designs for microwave system, structures, power distribution, solar arrays, and attitude control systems

• Refine data on environmental effects on satellite


Radiation effects (Van Allen Belt and geosynchronous orbit)

• Demonstrate productivity in space

“Large” satellite construction

• Establish satellite reliability and ascertain any failure modes

Establish operational experience


It is important to minimize the amount of power delivered on the ground due to costs. It can be shown that a system providing 1/20 of the power (250 megawatts) as that from a full-scale satellite (5000 megawatts) will cost about

80% of the full satellite. This small differential is due to applying the same

DDT&E costs and the development of a heavy lift launch vehicle. By limiting the ground output to a few tens of kilowatts of less the total demonstrator costs can be minimized. Small amounts of power can still demonstrate the concept of continuous power from space to technical and lay persons as well as provide valuable experience in space operations and productivity.

It should be emphasized that the space projects have not been thoroughly developed but these projects have received varying degrees of consideration.


1. “Solar Power System - Reference System Report”, DOE/ER-0023. U.S. Department of Energy and the National Aeronautics and Space Administration, October, 1978.

Figure 1 Satellite System Sizing

Figure 2 Solar Power Satellite Efficiency Chain

Figure 3 Solar Array Options

Figure 4 Solar Cell Options

Figure 5 Microwave Power Transmission System

Figure 6 Microwave Power Density at Rectenna

Figure 8 Heavy Lift Launch Vehicle (HLLV)

Figure 9 Cargo Orbit Transfer Vehicle (COTV)

Figure 10 Transportation Between LEO and GEO

Figure 11 Automated Beam Builder

Figure 12 Satellite Structure

Figure 13 Power Systems Cost Comparison

Figure 14 Large-Scale Inverted Test Concept


Richard M. Dickinson

Transmitter Group Supervisor

Jet Propulsion Laboratory

Pasadena, CA 91103


The proposed satellite power system for importing nearly continuous solar-electric power from synchronous orbit, is to consist of a fleet of orbiting solar collector spacecraft, each of approximately 100 km


area. The 6.25 GWe power output is converted to 2.45 GHz microwaves and beamed to earth based rectenna collectors for conversion to high voltage ac or dc power, to feed into the electric utility transmission grid.

The microwave power beam subsystem operation and performance must be closely monitored and accurately controlled in order for the SPS to legitimately function in the international radio regulation framework (radio frequency interference and biological hazard) that may exist in the late 1990’s.

We consider the various design characteristics and equipments of the transmitting active retrodirective array and the receiving rectenna that are necessary to be monitored and controlled in order to adequately manage the spacecraft RF power beam performance. The operating and maintenance strategies for coping with requirements for controlling beam pointing, beam shape, harmonics, noise, grating lobes and ionosphere nonlinearities are discussed. We speculate on the potential degree of precision of beam characteristics that may be required or achieved.

The instrumentation requirements for measuring or deriving the state of performance of the microwave subsystem are enumerated along with the control instrumentation necessary to effect changes in operating configuration, state or parameter values so as to initially achieve, better maintain or to restore performance to the defined acceptable state.

We conclude that there are potentially a range of controllable parameters that may be designed into the microwave subsystem in order to achieve the various radio regulatory requirements. However, extensive design tradeoff studies are necessary to properly select the optimum instrumentation, controls and operating and maintenance strategy. This is because the resulting effects on the overall system efficiency, availability and economics are also potentially severe. For example, the increased insertion loss, weight, power dissipation and decreased RF breakdown margin of the harmonic filters required to achieve protection of existing earth and space based radio services, may cost several per cent additional loss in power beam transmission efficiency.



Ralph C. Chernoff

Transmitter Group

Jet Propulsion Laboratory

Pasadena, CA 91103


An active retrodirective array (ARA) is a phased antenna array in which retrodirectivity of the transmitted beam is produced by electronically conjugating the phase of the pilot signal received by each element of the ARA. ARA’s can be easily modified to function as receiving, as well as transmitting, arrays. Due to their inherent failsafe characteristics,

ARA’s are particularly attractive for microwave power transmission from solar power satellites. Communication satellites and deep space probes are other possible applications.

The “Central phasing” principle, in itself a simple generalization of the phase conjugation principle, avoids the need for structural rigidity implicit in conventional phase reference distribution systems. We describe a phase reference “tree” for implementing central phasing in very large ARA’s, as well as a new kind of “exact” frequency translating phase conjugator which provides both input-output isolation and freedom from squint. The effects of doppler, aberration, impedance mismatches and multipath are discussed. We report on the performance of an experimental two element ARA, and on the design of an eight element ARA now being built at JPL.





W. C. Lindsey and A. V. Kantak

LinCom Corporation

P.O. Box 2793D

Pasadena, CA 91105


A classical method of transfering time and frequency from one network node to another is the Master-Slave (MS) technique. This method, though fairly simple and easy to implement, is plagued with the problems introduced by the variable propagation path delays between network nodes. The master slave returnable timing system (MSRTS) technique presented in this paper offers the flexibility and simplicity of the well established

MS approach; however, it provides for a novel automatic delay compensation feature.

Delay compensation between two nodes is achieved by measuring the delay between the two nodes and then using this knowledge about the delay to automatically update the phase of the transmitting node such that the phase at the receiving node is independent of the delay between the two nodes. A theory and analysis of this system is presented here in a noise free environment and an extension is made to cover the propagation of internally generated noise (clock phase noise) through the network.


Most time and frequency transfer techniques in use today utilize the master slave (MS) technique because of the simplicity of its implementation and in some cases, because of the reduced accuracy requirements. Recently, there have been some attempts made to understand the mechanization of the time and frequency transfer in a mutually synchronized system Refs. 1,2, 3. The steady state frequency and phase errors of the mutually synchronized systems are, as pointed out in Ref. 1, sensitive to the propagation path delays; hence, the necessity of delay compensation arises. Recently, an automatic delay compensation technique was presented in Refs. 4,5. This technique was based upon a returnable timing concept.

* This was performed at LinCom Corporation under Contract No. NAS9-15782 for the Johnson

Space Center, Houston, TX, 77058.

The Returnable Timing System (RTS), described in Ref. 4, is of particular interest because the delay compensation feature of this approach can be combined with the MS system to create the Master Slave Returnable Timing System (MSRTS) described in this paper. This approach allows one to combine the advantages of the MS and the RTS techniques. The

MSRTS uses the basic MS hierarchial structure (tree structure) to transfer time and frequency from one network node to another node with the result that the network frequency is delay independent and the time error is nominally zero. This method is particularly important in many applications, e.g., highly accurate clock distribution systems required in avionics systems, computer communications and large retrodirective phasedarrays such as the Solar Power Satellite.

The basic idea in MSRTS is to send the time-frequency signal received at a particular node back to the sending node. The delay accumulated by this returned signal is used to advance the phase of the master (sending) node thereby canceling the effect of the delay introduced by the path. To understand this idea clearly, we shall start with two nodes. One of these is designated as the master and the other represents the slave node.


Figure 1 depicts the two nodal system. Each node has a signal processing equipment containing a phase locked loop. As shown in Fig. 1, we desire to transmit the phase and frequency of the master signal to the slave station. The master signal and the returned signal from the slave station are used to drive the PLL and the output of the VCO is sent to the slave station via a directional coupler. This directional coupler has the ability to sense the direction of energy flow and hence it can route the received signals at the same frequency in the proper directions. At the far end of the “cable” there is a directional coupler with termination. The termination in the coupler simply reflects some of the incident energy back to the master for the purpose of delay compensation. It should be noted that the mechanization can be brought about by using two diplexers (or four bandpass filters) instead of the two directional couplers. The master signal is modeled as

(1) where 2


is the phase to be transferred to the slave node and T is the frequency to be transferred. The reference signal as the master node is modeled by where (t) is the estimate of the phase 2


. In addition, we define the signals

where J ij

is the delay suffered by the signal from the j-th node to the i-th node. Thus the error signal becomes and with no frequency offsets and no biases, in the steady state we will have 2(t-J


2 ss

so that ,(t) will be zero when



) =


Here 2 ss

is the steady state value of 2(t) and T s

represents the steady state frequency. From

(2) we have

(3) and in the steady state the reference signal becomes

In addition, the steady state phase of the reference signal at point B of the slave node is

(4) where


Thus we observe that when the signal paths is reciprocal )J = 0 and the phase and frequency at the master node (point A) equals the phase of the received signal at the slave station (point B) of Fig. 1.


As shown above, the phase of the master signal is regenerated at the slave node. This signal will be fed to the slave PLL and this PLL will act as the master for the next node in the chain. Figure 2 shows the situation in a block diagram form. The phase of the signal at the input of any slave (in the steady state) is obtained from the phase of its immediate master by properly interpreting the quantities in (4). For example the phase of the signal input to the (n+l)-st slave in the link is given by


Let then

Input Phase to

the n+1st Station

(6) where 2 m

is the phase of the master socillator to be transmitted to the slave stations.

Ideally, with nondispersive propagation paths, we will have )J i,i+l

= J i, i+l

- J i+l,i

= 0 for all i = 1,...,n=l; hence (6) says that the phase input at the (n+l)-st node is the same as the phase of the master node.


The condition )J i, i+1

= 0 for the ideal case assumes that the forward and return path are of equal delay and that the hardware components in the system are perfect. In actual practice this is rarely the case hence it becomes reasonable to assume that each )J i, i+1

variable with zero mean (unless the system implementation dictates otherwise) and a variance, say F

It is also reasonable to assume that there is no correlation between any pair of random variables. In fact, we will assume that they are statistically pairwise independent random variables. Thus, we have the expected values i, i+1




and using (7) we get

Thus the disturbance to the phase input to the (n+l)-st station has a zero mean and a variance of . A simple application of the central limit theorem says that for a large enough “n” we have


(10) and the input phase to the (n+l)-st node is so that



i, i+1


So far we have tacitly assumed that E[)J reason to assume that E[)J i, i+1 ie. ,

] = c i, i+1 i, i+1

] = 0 for all œi = 1,...,n. If there is some

… 0 then the mean of y n

increases linearly with “n”,






In such a case an alternate mechanication of the system described in Fig. 3 may prove useful.

The key points to note in this mechanication is that the directional couplers in Fig. 1 are replaced by diplexers and frequency dividers are included. Also the direction of flow of

“T” signals and “2T” signals is alternated. Going through a similar analysis as before we can arrive at an equation, similar to (11) which represents the steady state phase into the

(n+l)-st node, i.e.,

(15) where



If c i,i+1

= c for all i, then



Comparing (14) with (17), we see that the mean of the phase disturbance has decreased by at least a factor of n while the variance remains the same.





The mean and variance of y n

in (15) is given in (17). Thus

Assuming n large the, central limit theorem applies and the probability distribution of the varialbleM n+1

is model by


Thus it becomes possible to find the

(19) where

Thus we see from (19) how the mean value c probability of M such a case with n+l n

being in (-),)) interval. If c

enters into the equation to affect the n

=0 then (19) reduces to 2 erf( ). In

from (19) that and


The above equation tells us how many nodes (“n”) can be allowed in a chain of nodes for fixed values of “a” and “F” and a selected value (variable) of ). The ratio ()/F) interpreted as the Spread Ratio

2 can be


Id we let )=B and change a to (1-a), then (20) becomes


Here “n” specifies how many stations can be allowed in a chain for a predetermined value of “a” and the system parameter value F before slip of one cycle is expected to occur at the input to (n+1) station. Thus a large number of stations can be allowed if F is sufficiently small before the cycle slipping occurs, alternatively the probability of slipping a cycle tends to zero. Figures 4 and 5 show the plot of “n”, the number of nodes in a chain vs the spread ratio and the standard deviation, respectively.


The above analysis does not consider any form of noise either generated by the system components like the PLL or the channel noise. The analysis presented below considers the propagation of noise generated by the PLL through a chain of nodes. Figure 6 describes the connection and signals at the k-th node. We will adopt a new notation, i.e., (@ the phase of the signal (·). At the k-th node


) means

where in which R oscillator and 2 k models of the signals in Fig. 6 are: k

(t) is the phase noise generated by the k-th

(t) is the phase estimate of the k-th oscillator. The appropriate total phase

If we take the Fourier transform on both sides of the error signal we have



Assuming that proper filtering removes the delta functionsand solving for 2 yields k

(T) from (23)




Then (24) becomes


This is a recursive relation for 2 k

This turns out to be

(T) and can be solved for the n-th station phase output.

(27) where

Thus the system can be represented as in Fig. 7 or it can be modeled as a parallel fed system shown in Fig. 8. From (27) we have


Now assuming that E[2 i

(T)] = 0 for all i and E[R i

(T)R j

(T)] = 0 for i…j we have


This says that on the average, the output phase of the n-th nodal clock equals the master phase. We are interested to find from the power spectral density

(30) where is the single-sided spectral density of the R i

. Thus

(31) where R k have H i

(T) and H k

(T) are defined in (25). Now if we assume that J ij

(jT) = H(jT) for all i and R i

(jT) = R(jT) for all i, then

= J for all i and j we

(32) if for all i. Now assuming that J ij for all k then with S


(T) = 1/T we get

= 0 for all i and j and F k

(T) = F(T)

Assuming further that the loop filters are characterized by then

where . Thus H(jT) reduces to

Substituting H(jT) and R(jT) in (32) one can compute the normalized mean squared random jitter accumulated after n nodes (normalized by the mean squared random jitter produced by node one) is computed by the use of an IBM Computer. The results are presented in Fig. 9. As can be seen, as the damping factor . of the circuit increases, the noise variance at the end of the n nodal chain decreases.


When there are a number of stations to be connected (phase locked) together to transmit phase information from one node to all the remaining nodes, one possible way of connection is to connect all of them in a chain and let the master feed the chain. There are obvious disadvantages of a series connection with the above scheme, hence the well known master-slave hierarchial tree sturcture will be tried next. Fig.10 shows such a tree structure using MSRTS concept for the time transfer.

The master (station 1) in the figure feeds the phase to four physically close equidistant

(from the feeding point) phase locked loops via a four way power splitter. These are level

1 slaves. Each of the level 1 slaves is connected to a four way power splitter as level 2 via the MSRTS. Again a group of 2nd level four PLLs are placed equidistant from the power splitter and each of these is connected to the third level power splitter via MSRTS. This is continued until all the oscillators (nodes) are connected.

For each oscillator in the network there exists only one chain connecting the master to the oscillator and our analysis in the previous section holds.



The solar power satellite system consists of a geosynchronous satellite at an altitude of

37,000 km. The main purpose of which is to collect the solar power, convert it into microwave power and then transmit it to the ground based receiving station. Currently the transmitting antenna on the SPS is conceived to be a retrodirective phased array having an area of 1 km diameter circle. Every retrodirective antenna needs a constant phase reference for its operation. The conventional methods of reference phase distribution are of little consequence because of the huge size of the SPS antenna. However, a phase distribution system using the MSRTS tree concept solves the problem.

Figure 10 shows the phase distribution tree system proposed for the SPS antenna as shown in the figure, the array center locks onto the phase of the pilot beam transmitted from the

(Ref.6) center of the receiving antenna on the earth. This phase is transmitted over the entire aperture of the antenna by the use of MSRTS tree structure. The phase picked up by the array center is distributed to 16 level 1 slave nodes (the figure shows only four). Each of these nodes transmit the phase to 16 level 2 slave nodes. This continues up to level 3 slave nodes, each of which supplies 25 level 4 slave nodes which supply the phase to the power amplifier tubes for conjugation and subsequent radiation of the microwave energy.

Figure 11 shows the phase distribution tree structure. The conjugator (multiplier) and the high power amplifier (HPA) along with the necessary phase stabilizing circuit. The SS

(spread spectrum) receiver shown at each power amplifier tube receives the pilot wave, extracts the phase and supplies it to the conjugator where it is conjugated with the help of the reference phase, supplied by the MSRTS phase distribution tree.


The MSRTS provides a new way of delay compensation which can use already existing

Master Slave systems. Proper design of filters at each station can reduce the variance of noise at the end of the chain of stations. The MSRTS can also be useful in the SPS theory, especially those retrodirective antennas where distribution of the master phase accurately is vital to the performance of the antenna. Also this method can be of use for sensing phase at a remote location for phase measurement purposes.


1. Lindsey, W. C., Kantak, A. V., Dobrogowski, A., “Mutual Synchronization Properties of a System of Two Oscillators with Sinusoidal Phase Detector,” IEEE Transactions on Communications, December, 1976, pp. 1321-1326.

2. Inose, H., Fujisaki, H., Saito, T., “Theory of Mutually Synchronized Systems,”

Electronic Communications in Japan, Vol. 49, Apr. 1966.

3. Gersho, A., Karafin, B. J., “Mutual Synchronization of Geographically Separated

Oscillators,” B.S.T.J., Vol. 45, Dec. 1966, pp. 1689-1704.

4. Lindsey, W. C., Kantak, A. V., Dobrogowski, A., “Network Synchronization by

Means of a Returnable Timing System,” IEEE Transactions on Communications, Vol.

COM-26, June, 1978, pp. 892-896.

5. Yamato, T., Ono, M., Usuda, S., “Synchronization of a PCM Integrated Telephone

Network,” IEEE Transactions on Communication Technology, Vol. COM-16,

Feb.1968, pp. 11.

6. Lindsey, W. C., “A Solar Power Satellite Transmission System Incorporating

Automatic Beamforming, Steering and Phase Ciontrol,” LinCom Technical Report

TR-7806-0476, LinCom Corporation, P.O. Box 2793D, Pasadena, CA, June 1973.





Figure 7.




Figure 8.

Parallel Equivalent of the Phase Noise Propagation in a Chain of Nodes

Connected in MSRTS/


Figure 11. Reference Solar Power Satellite Transmission System.


To Be Announced


The most devastating task ever given to the computer system manager, is that of estimating the manpower and time to develop software. Software development has become almost completely out of control in terms of estimating the manpower and time needed for the complete job, from interpretation of the requirements to the operational readiness date.

Just as the hardware design engineer never reaches the ultimate design and thus never concludes his job, so goes the software designer. The difference is, hardware design managers have grown up in this environment and therefore know the pitfalls and stopping points of producing a good, reliable product. The software design manager is less qualified because the field is relatively new, it is very complex, and computer hardware technology is expanding so rapid that software designers cannot catch up. This paper is designed to provide the software development manager categories and rules of thumb for estimating time and manpower requirements for each category of software development. In addition to establishing the standard milestones such as requirements definition, specification, design, coding, checkout, verification and validation, consideration is given on how to keep the Ford requirement from turning into a Cadillac capability.

Management of Software Development

H. Robert Downs

Science Applications, Inc.

1257 Tasman Drive

Sunnyvale, CA 94086

1. Introduction

Development of software is well known as a very difficult management problem. When the software is a part of a complex telemetry system, the management problem becomes particularly complex; particularly if the software is not well defined before implementation begins.

This paper discusses some of the problem areas in software management with particular emphasis on the requirements and design stages. In a telemetry system development effort, the system designers must decompose system requirements into requirements on various subsystems. This process is particularly apt to lead to software requirements which are not well defined, impossible to meet or inappropriate for the hardware on which it is implemented. This paper addresses methods of performing and managing this type of software development.

2. The Software Development Process

The Navy has produced a software life cycle document which describes a number of steps in a typical life cycle. These are:

1. Conceptual Stage which involves conceptualizing the task to be performed.

2. Procurement Stage

3. Requirements Definition which involves writing the RFP and carrying out the procurement.

which is usually developing a formal specification of the requirements by the contractor

4. Design which involves specifying a careful design of the software to be developed.

5. Coding and checkout

6. Testing

7. Integration

8. Operational Test and Evaluation which is often performed with the testing effort.

which is testing the system in a realistic environment

9. Maintenance

Each of these phases has some unique problems and requirements for managing the effort and carrying out the tasks. This paper is concerned primarily with the development of requirements, design and code steps in the above process, though it is also addressing scene of the conceptual development.

A number of recent publications have made the point that the errors which occur in software and which are discovered in the testing phase are usually the result of design or requirements mistakes. The cost of repairing these errors is much higher when they are discovered after the code is written than if they could be discovered during the design and requirements definition process. Hence it is useful to have a software development process which formalizes the requirements definition and design steps and allows for a formal checking for errors at these stages.

However, the carrying out of a requirements specification and design is a subjective process that requires some creativity and judgement on the part of the people performing these jobs and hence is not an inherently error free or automatable process. The process typically must be iterated and additional improvements made as more knowledge is gained about the system and about the implementation or discrepancies discovered. The design should be updated and formally validated or checked as changes are made to the code or specification. This process will maintain a consistent design description document for use throughout the software development process.

3. Proposed Software Development Approach

This paper recommends a slightly modified approach from the usually recommended approaches which are top-down and bottom-up. The top-down approach implies that one

starts with a high-level view of the process to be solved and gradually fills in the details in this process by working down a hierarchy or tree. Conversly the bottom-up process implies that the developer starts with detailed level system components which he knows must be built and gradually links these together to build the overall process. There is a good deal of debate on what process should be followed and many software development efforts but top-down is the currently preferred method.

This paper suggests a slight variation on the top-down method which could be called the

“side-wise” method. This approach requires producing in parallel, several software development steps. The developer begins with an outline for each of the key software documents including the requirements definition, the design document and the code structure. The process consists of gradually filling out each of these document outlines with details as the relevant structure is known. The process is iterative in that each of the documents is gradually improved to satisfy the previous document in the usual chronological sense but there is a consistency check performed on each of these documents in order to be sure that it meets all of the constraints specified in the previous document.

Of course it is necessary to start with a preliminary requirements specification before any design can built and to start with a preliminary design before any code can be built. But it is not necessary to wait until the final document at the previous level is available before beginning the next level. Development can be a top-down development for each of these documents, that is the design and the code can each be developed in a top-down fashion.

Another key concept is related to how the requirements are developed for what the software must do. In many large system efforts, the system designers expect the software developers to perform a lot of functions which are not very well specified in the beginning.

The software is expected to take up most of the slack in the system design. This approach leads to software which does not satisfy the initial requirements and which overruns its budget and often has other failings.

It is important to provide a precise and realistic statement of what the software is to do.

This statement should describe such items as:

What should the software do when an overload occurs

What is the priority of handling various functions

What if inconsistent operator inputs occur

What reliability (less than 100%) is expected of the software

What hardware or other subsystem failures are allowed for and what should the software do in each instance.

It is important that the above issues are addressed at the system level and not by the software designer or programmer.

A third concept in this software management philosophy is the concept of estimating the cost and resources needed to produce and execute the software. This estimation should be done early in the software development process and requires a continual update as more information becomes available. The estimation is used to manage and monitor the software development process and to recognize problem areas as early as possible. When these problem areas are identified early in the development process, it usually possible to take management action to resolve the problems by either moving the functions to another part of the system or putting more effort or resources into organizing the software.

The final concept to be mentioned in this paper is the use of a formal specification language or document for each of the steps in the development process. There are currently available formal requirments languages, formal design languages, and of course high level languages and procedures for specifying the code. It is recommended that the developer choose a particular procedure and language for producing these documents and continue use of this language throughout the development process. The use of such formal languages will greatly reduce the possibilities of error and allow the developer a great deal more flexibility farther down in the software development process if he needs it.

3.1 Iterating the Software Design

The basic idea in this section is to take a software development process consisting of a requirements specification, a design specification, and a code specification, and to iterate the specifications to the final product. The various documents are iterated in the following way.

1. The requirements specification for software must be iterated with the requirements for the remaining subsystems in an overall system. For example, during system design it may be discovered that some functions can be performed by a front-end signal processor or may be performed within the general purpose computer using general purpose software. The trade that needs to be made is to determine which of the two approaches is most cost effective for the particular system. This requires that the cost of performing the function in the processor including the cost of software development, the cost of additional memory, the processing speed and so forth that may be needed must be estimated. once these estimates have been obtained and the performance that is available from such a system has been estimated, then this candidate approach can be traded off against the alternative approach which is to put the signal processing function in a special purpose device. Such

subsystem tradeoffs should be made as early as possible but should be documented and allowed to stand as alternatives even though one of the alternatives is selected at this point. It may be necessary later in the development process to revise the decision on which alternative to use.

2. The design document should represent a software design corresponding to the requirements specification. In addition to the usual software structure deign, it is necessary in real time and telemetry systems to have estimates of the amount of time allowed for each function to be performed in the software and to have other methods for controlling the timeliness of the functioning of the software. Current software development methodologies are aimed at allowing the development of correct software but do not generally take into account how long the software must execute. The method for monitoring the time that the software takes is through the use of a high level or functional simulation. This functional simulation will be described in more detail in another section.

3. The official coding development can also begin before the design is finalized but there is some risk that the code will not be correct when the desgn is finalized. In order to resolve the problem, it is necessary to have a method for formally verifying that the resultant code in fact, meets the design document. Such methods usually consist of the use of consistency checks and other software verification tools. While it is often not possible to verify completely that code meets a particular design, it is possible to catch most errors and a manual verification procedure can be used to obtain high confidence in the code being an accurate representation of the design.

3.2 Software Estimation

In order to monitor the development and resource usage of software as it is being built and before it is completed, it is recommended that a functional simulation be built. The functional simulation models each of the software functions as a usage of the computing resource. Computing resource inludes processing time, memory, I/O channel usage and other critical resources. For each function an estimate should be made for each of these parameters as the software is developed. The function may be decomposed into smaller sub-functions, the numbers may be modified to meet what is acutally measured or what is obtained from more refined estimates, and then included into the simulation. The use of such a simulation throughout the development process provides a check that the software is not exceeding the available resources and also allows developers to investigate and determine reasonable actions to take if the software does exceed the resources available.

The simulation must be driven by a sample external input which repesents the kind of inputs the system is expected to respond to and the simulation must model as key events the start and stop of each function. Typically the operating system contains a scheduler and data management function. This operating system must be modeled in some detail in order to determine fairly accurately what kind of timing and memory usage is expected from these functions.

One other issue to be mentioned in the section is the use of software estimators. Estimating software is difficult and requires a great deal of experience and analysis. When these estimates are obtained, they are usually on the low side unless the function has been analyzed to the finest level of detail. Also, if the estimates lead to a ntnber which is near the limit of the available resources (either memory or time) , then the cost of producing this software has been shown to be much higher than it would be if it were not near the resource limits. Thus it is recommended that the hardware system be designed to include

25% spare memory space and spare processing power over the expected amount of software usage.

3.3 Formal Languages and Procedures

A typical design language is a kind of “pidgin” language. That is, a language that has some formal structure which allows the user to lapse into imperfect sentential structures or constructs. For example, many design languages for software have been referred to as

“pidgin” design languages. Currently there are more formal design languages becomming available. These languages and other requirements languages are recommended for use during any software development effort. Use of a formal language allows the use of automated tools for checking the consistence of the requirements or design and for comparing requirements, design and code.

If a formal language is not used, it is recommended that a precise definition of the contents of the requirements specification document and the design document be made by making formal outlines and specifying the contents of these documents. The developer can assure himself that no components of the requirements or design have been overlooked; however, he does not gain the advantage of possible automated checks on these documents. As mentioned earlier, the formal documents are expected to be updated throughout the development process. This updating requies a sophisticated configuration control process which makes available to all developers the appropriate and most recent version of each document and which maintains records of previous versions for reference later on.

4. Summary

This paper has recommended some particular procedures for managing the development of software. These are by no means all of the procedures but are of interest because of their benefits to telemetry system software.

The focus of all of these procedures has been on the front-end or requiements and design parts of the software development effort. The reason for this is that most errors that occur during software development have been shown to have ocurred during these processes.

Even though the effort devoted in most software development projects is usually 50% or more on the testing component, the way to solve this problem, and reduce that effort is to apply that effort earlier in the development process and to use formal procedures and tools to ensure that the software which is developed has as few errors as possible before going to the testing process.

This paper has recommended an iterative or “side-wise” development process which uses top-down structured design and coding procedures within it. It has recommended the use of a software estimating technique for determining the amount of software to be developed and the amount of effort required to produce it. It has recommended iterating the requirements, design and code documents to a final product.

Finally it has recommended the use of a formal language or procedures for each of the requirements, design and code documents. The use of a computerized and automated software development product such as the Programmer Workbench (trademark) is useful as an administrative aid in carrying out the development efforts recommended here.


To Be Announced


The software has been designed, coded, and entered into the computer -- now comes the big task of verification and validation. This paper develops several system approaches, technical evaluation criteria, and key milestones for managing the progress, technical excellence and completion of this process. This is the phase of software development that becomes very difficult if a good design and design review has not been performed. In short, a well intentioned capability now becomes a "fair" design with patches installed to make it work. It is the number and type of patches to make it work that causes the software package to rapidly get out of hand. These problems will be discussed.


Roland E. Olson

EMR Data Systems

44517 North Sierra Highway

Lancaster, California 93534


The hardware design and development manager with all of his infinite wisdom, knowledge, and years of experience has always had to be on his toes to know when a hardware design is “good enough” to meet the requirement. Add to this duty, the management of software development which is relatively new which makes it difficult to decide if the software is “good enough” to meet the requirements and you have a manager faced with a man sized task. For example, ask him to decide when a computer problem occurs, if the problem is due to hardware or software, especially when both the hardware and software designers claims his system to be working okay --- this manager is in high demand because only a select few have experience with both hardware and software development and integration. So what about the rest of us? This paper provides the computer system manager such evaluation criteria, questions to ask and decision making rules to follow, when the hardware and software designers say to you --- “My system works fine, the problem is his.”




Richard J. Chan

Work Unit Manager

ISPM NASA Spacecraft Telecommunication System Engineering

TRW Defense and Space Systems Group

Redondo Beach, California


The International Solar Polar Mission (ISPM) is a joint venture of the National

Aeronautics and Space Administration (NASA) and the European Space Agency (ESA) for each to develop a single spacecraft to carry unique scientific experiment hardware packages. The hardware is designed to gather data for the scientific study of solar properties and interplanetary physics out of the ecliptic plane.

The NASA and ESA spacecrafts will be launched in tandem by the Shuttle and the threestage Inertial Upper Stage (IUS) from earth in a ten-day launch window in early February,

1983. After separation from the IUS, the two spacecraft will proceed on parallel but independent trajectories very close to the ecliptic plane to Jupiter. There, the two spacecraft will make use of the large momentum transfer derived from the Jupiter swingby.

The spacecraft targeted southwest of Jupiter will be swung to depart in a northward trajectory, highly inclined to the ecliptic plane, and the spacecraft targeted northwest of

Jupiter will be swung to depart in a southward trajectory, also highly inclined to the ecliptic plane. The ISPM will be the first mission to send spacecraft in an orbit highly inclined to the ecliptic plane and the first to exploit any planet to maximize orbital inclination.

The journey to Jupiter from earth will take 475 to 500 days. The remaining 1700 days of the mission is spent in the inclined orbit. By the end of the mission, the two spacecraft will each have completed passage over both poles of the sun, one pole before and the other pole after crossing the ecliptic plane.

Telecommunication services are required from launch through the out of the ecliptic phases of the mission which correspond to ranges from near earth to 6.1 astronomical units

* TRW has been selected by the Jet Propulsion Laboratory (JPL) to negotiate the ISPM spacecraft contract.

(1AU = one earth-sun distance of 93,000,000 miles). Telecommunication services are provided by the following links: spacecraft/IUS/Shuttle to the Jet Propulsion Laboratory

(JPL) Mission Control and Computer Center (MCCC) via the Tracking and Data Relay

Satellite System (TDRSS); spacecraft/IUS to the JPL MCCC via the TDRSS; IUS third stage/NASA spacecraft/ESA spacecraft to the JPL MCCC via the TDRSS; and NASA and

ESA Spacecraft to the JPL MCCC via the Deep Space Network (DSN).

Key features of the NASA spacecraft telecommunication system design:

• General Features

Signal and modulation formats completely compatible with the TDRSS and the JPL DSN

Configuration makes maximum use of space qualified, off-the-shelf hardware

Full equipment redundancy provides high confidence of mission success


Science and engineering data link performance can be optimized by command selection of downlink modulation indices

Concatenated Reed Solomon/convolutional encoding and Viterbi/Reed

Solomon decoding minimizes spacecraft EIRP

• Antenna Subsystem

- signal autotrack for spacecraft attitude control

- aspect angle (LAA) transmit functions


LAA receive functions and for emergency and LAA transmit functions

• RF Subsystem


Transponders for S-band coherent carrier, command, and ranging signal detection and for S/X-band coherent or noncoherent carrier data and ranging modulation


• Telemetry Modulation Subsystem


Multi-frequency suppress subcarrier modulation or bypass


Design consideration, tradeoffs, implementation, and performance for the NASA spacecraft Telecommunication System are discussed.




W. Morris Holmes, Jr.

Advance Systems Engineering Director for Communication Systems

Space Systems Division

TRW Inc.


Satellite communications in the late 1980s and 1990s must provide reliable high-rate communications between very small inexpensive terminals with routing flexibilities approaching today’s telephone system. The capabilities needed for successful competition with established and evolving terrestrial communications systems can be provided most efficiently using Satellite On-board Signal Processing.

The rapid improvement of high-speed digital technology makes it possible and costeffective to demodulate, process, and remodulate individual data streams with rates approaching a gigabit. System processing capacity of several gigabits (ten in the example described) through a single satellite can be provided.

The satellite communications system described provides communications for very small and very large (trunking) users. Independent combinations of FDMA and TDMA are used in the uplink and downlink designs to minimize terminal costs. Signal routing for small users is accomplished by a digital store-and-forward technique which greatly simplified the terminal receiver, compared to satellite-switched TDMA. Different processing techniques are used for very high data rate users, but complete interconnectivity between all users is maintained. This avoids double-hop routing with excessive transmission delays.

On-board processing allows use of innovative responses to rain attentuation without requiring expensive, large signal-power margins. Terminal synchronization and timing is greatly simplified without a significant increase in satellite complexity, by integrating the synchronization loops with the downlink communication TDMA burst structure.


* This paper reports work contracted under Contract NAS3-21933 sponsored by NASA Lewis

Research Center.

A NASA LeRC funded 20/30 GHz Mixed User Architecture study has resulted in a baseline system design for cost-effective communications in the years 1990 to 2000.

Generation of a useful mixed user TDMA architecture has required a broad overview of system characteristics.

The TRW 20/30 GHz Baseline System consists of a synchronous orbit satellite, 18 large trunking terminals with 12 meter apertures and 10 km space diversity, 25 to 30 small trunking terminals with 6 meter apertures and 10 km space diversity, several thousand inexpensive direct-to-user (DTU) terminals, and a master control station with at least one alternate master control station.

The DTU terminal low-cost design offers the greatest challenge in the 20/30 GHz communication system. Since the number of DTU terminals is very large, the terminal direct cost is a large (perhaps the largest) element of system cost. The DTU terminal performance drives the satellite design, and hence determines a second major system cost element. Because the DTU system is able to avoid using terrestrial signal distribution and routing, and the charges associated with these functions, the DTU system also represents the largest economic value element of the system.

Because of these factors, the 20/30 GHz TDMA Mixed User Architecture design must start with definition of DTU user terminal characteristics. TDMA, control, and onboard processing architectures are designed to maximize system performance with minimal DTU user terminal requirements.

Trunking terminal design and the satellite trunking support components are less critical elements in determining system cost effectiveness. Trunking at 20/30 GHz provides direct cost benefits and prevents saturation of the lower frequency satellite communication bands.

Integration of trunking and DTU systems provides the greatest economic benefit to the

DTU system, however, by increasing the utility of a DTU terminal.

DTU terminals need the ability to integrate their communications channels with a trunking system. A large percentage, perhaps greater than half, of DTU point-to-point communication channels will have their other termination point located in a trunking area.

By shifting signals from the DTU system to the trunking system in the satellite, the more expensive DTU system carries less total traffic. Signal routing is greatly simplified.

Without this interconnectivity it might be necessary to provide DTU terminals at each trunking location to provide integration and avoid double-hop routing.

The baseline 20/30 GHz satellite communication system resulting from this study incorporates on-board satellite demoduation and routing of individual 65 Kbps digital voice-grade circuits. This level of routing flexibility is necessary to provide efficient

communications to the very large number of DTU terminals projected. From an external point-of-view, the resulting system looks very much like a very large, distributed telephone switching system. The circuit interfacing hardware is distributed among all the DTU and trunking terminals. Control and routing computers are at master control station(s). And the switching circuitry which provides full interconnectivity between 30 to 45 thousand circuits is in the satellite.

The satellite (Figure 1) must be fairly large to provide such a capability. The digital onboard processor will be complex and will represent a significant, but not dominant, portion of the satellite. By moving all the switching control logic to the master control station, and by taking advantage of a level of digital technology that is commercially available (but not yet space qualified) today, the digital on-board processor design risk is acceptably low.


The system network configuration required to support DTU terminals efficiently is shown in Figure 2. While terminals handling heavy trunking loads can continue to operate in a net-connected configuration, a star-connected configuration is required for small users to simplify the DTU terminal. Furthermore, the central node of the star must be located in the satellite to avoid additional transmission delays to and from a ground processing station.

The concept selected to provide the configuration of Figure 2 is shown in Figure 3.

Separate multiple beam antennas (MBA) provide fixed-beam coverage of the high-density traffic regional and scanning-beam coverage of low-density traffic regions. The highdensity region trunking users can be interconnected using SSTDMA technicaues similar to those developed for the Advanced WESTAR communication system.

DTU terminals require on-board satellite routing, and this is most economically provided by demodulation, digital routing, and remodulation. Analog routing techniques become difficult when several thousand channels must be independently routed.

DTU terminals need the ability to interconnect with trunking communication systems. This can be provided by locating a DTU terminal at each trunking terminal, or by integrating trunking and DTU signals on board the satellite. We have chosen the latter approach. This is shown by the trunk/DTU interfaces in Figure 3. Integration of DTU and trunking signals on board also solves the problem of serving thin-route trunking locations that cannot economically justify an independent satellite antenna beam in the fixed-beam MBA.

The satellite communication subsystem supporting the concept of Figure 3 is shown in

Figure 4. An SSTDMA section at the top of the block diagram supports the eighteen fixed

locations shown in Figure 5. An on-board processing section with demodulation and remoduation of each signal operates with scanning receive and transmit arrays to cover all other locations in the contiguous forty-eight states. (Hawaii and Alaska coverage is easily added).

The processing section has been sized for a 3 Gbps data throughput. While this seems very large, it could be mechanized with currently available but unqualified digital LSI circuits.

The total system throughput is 10 Gbps. Of this eight Gbps is trunking data and two Gbps is DTU data. Seven Gbps is of trunking data passed by the SSTDMA switch without further processing. The two Gbps of DTU data, plus one Gbps of small trunking and trunking/DTU cross-system data must be processed.

The processing technique used is illustrated by Figures 6 and 7. The memory is loaded with all the data to be processed for one frame period. Routing is accomplished by then reading the memory during the next frame period in a different arbitrary order. The readout order is defined by a system master control station and provides complete routing control of the data. Two memories are required since at any one time one memory is being loaded and one is being read.


A system design has been completed which, by transferring system complexity to a satellite and master control station, provides economic communication capabilities with very expensive DTU terminals.

Figure 1. Preliminary Satellite Design

Figure 2. System Network Concept

Figure 3. 20120 GHz Mixed User System Architecture

Figure 4. Satellite Communication Subsystem Block Diagram

Figure 5. Trunking Station Plot

Figure 6. Processor Memory Loading

Figure 7. Processor Memory Output


by G. L. Dutcher

J. G. Lankford

Hughes Aircraft Company

Space and Communications Group

P.O. Box 92919

Los Angeles, California 90009


Beginning in 1982 communication services will be provided to the U.S. Navy by a series of UHF communications satellites known as LEASAT. The communications payload will be carried on a new spacecraft bus developed as an optimum bus for space shuttle launches; the program is the first of its kind to take advantage of the full 15 foot shuttle payload bay diameter. Several new spacecraft design concepts are employed in this optimized bus.

The communications payload incorporates transponders in the UHF and SHF regimes.

Four distinct types of transponders are employed: wideband, narrowband, relay, and fleet broadcast. The functional characteristics of each type is described in detail. The frequency plan leads to a significant potential for passive generation of intermodulation products, and intermodulation considerations are an integral part of the spacecraft design.


The LEASAT Communication System will provide wordwide communication services for the United States Navy beginning in 1982. The Navy awarded the LEASAT contract in

1978 to the Hughes Communication Services, Inc. (HCSI), which is in turn procuring the satellite system from the Hughes Aircraft Company. The LEASAT system will be the follow-on for the existing GAPFILLER (MARISAT) and FLTSATCOM systems, which have projected end of life dates in the early 1980’s.

System elements provided by HCSI include four geosynchronous orbit satellites and associated telemetry, tracking and command facilities. The four satellites will be located nominally over the continental United States and the Atlantic, Pacific, and Indian Oceans, as shown in Figure 1. Four fixed ground stations, or primary satellite control sites (SCS), will be located at Hawaii, Norfolk, Guam, and Stockton, California. Centralized satellite

control facilities will be located at the HCSI operational control center in Los Angeles. A direct interface exists between the HCSI control center and the Navy Telecommunications

Command Operations Center in Washington D. C. , and between the control center and the SCS.


The LEASAT bus is an outgrowth of the Hughes Syncom IV program, and is optimized specifically for Space Transportation System (STS) launch. This optimization has resulted in an arrangement distinctly different from transitional designs (expendable launch vehicle compatible). The most notable feature of the STS-optimized LEASAT bus is the relatively short axial length when the bus is installed in the STS. This is made possible by the 14 foot diameter spacecraft body which permits installation of the perigee propulsion stage centrally within the spacecraft enclosure, as shown in Figure 2. This compact axial length results in a payload length fraction of the STS capability closely matched to the mass fraction of the STS capability. Since STS launch fees are based on the greater of these fractions, the matched LEASAT system efficiently uses the STS resources and benefits from the resulting launch cost economies.

This full-diameter LEASAT layout also yields very ample payload installation areas, mass properties distributions yielding passively spin stable inertia ratios, ample thermal enclosure surface area (permitting passive thermal balance while accommodating large payload dissipations), and sufficient enclosure surface area for a simple body-mounted solar array.

As seen in Figure 2, the spacecraft is built around the 8000 pound solid perigee motor centrally located within the spacecraft. The 4000 pound bipropellant fuel tanks are mounted symmetrically around this motor, and are attached to the solid motor support structure. Loads from these elements are transferred to a graphite composite tubular truss structure to pickup points at the spacecraft periphery where they are carried through a cradle to the STS attachment points.

An annular shelf is provided at the outside diameter of the spun truss structure for mounting bus subsystem hardware, including batteries, power regulation electronics, attitude control units, spun telemetry and command (T&C) units, and driver units for pyrotechnic and propulsion valve operation. A solar cell array installed on a composite sandwich substrate forms the outer spacecraft enclosure.

Two 100 pound liquid bipropellant axial thrusters are located at the aft end of the spacecraft. These thrusters provide the apogee impulse which circularizes the orbit at synchronous altitude, and also serve to augment the perigee motor burn. A total of six

5 pound thrusters are also used about the spacecraft. These hydrazine thrusters provide both axial and lateral thrust for stationkeeping throughout the life of the mission. The lateral thrusters, when fired in opposing pairs, also provide spinup and spindown control.

The weight and power summaries for the spacecraft are given in Tables I to III. Table I highlights the propulsion requirements for taking the spacecraft from the STS parking orbit to synchronous orbit, and shows how the 2700 pound spacecraft weight grows to over

15,000 pounds in the STS. The allocation of the 2700 pound dry weight to the various space craft subsystems is given in Table II. The overall spacecraft power budget is summarized in Table III. The power subsystem generates over 1 kW of dc power at the end of the spacecraft lifetime, over 75 percent of which is devoted to the communications subsystem.

All communications equipment is carried on a shelf forward of the truss spaceframe.

During launch, the shelf is structurally locked to the truss. On-orbit following lock release, the shelf is despun and earth pointed by means of the bearing and power transfer assembly

(BAPTA) and despin control subsystem. Power and some spacecraft T&C signals are carried across the BAPTA on slip rings. All RF connections from the shelf to the various antennas are carried via waveguide or coaxial cable.

Both the large 14 foot UHF helices and the 7 foot mast for the omnidirectional T&C antennas are stowed during launch to minimize the axial length occupied in the STS. To provide T&C capability during the transfer orbit, the omni antenna is deployed immediately after the spacecraft is ejected from the STS. After the spacecraft is on-station and the communications platform despun, the UHF antennas will be deployed using a passive spring damper system. The deployed antenna configuration is shown in Figure 3.

The helices are deployed far above and forward of the spacecraft body to minimize potential for passive intermodulation (IM) generation on the structure.


Users of the LEASAT system will consist of designated mobile (air, surface, and subsurface) units and fixed stations. These users are largely DOD, including users from the

Navy, Army, Air Force, and Marine Corps. Readers who are interested in the various ground segment terminal characteristics should refer to Reference 1.

The communication subsystem consists of five equipment groups: the fleet broadcast, the

500 kHz channel (wide band), the 25 kHz channel (relay), the 5 kHz channel (narrow band), and the common channel group. A functional diagram of the communication subsystem is shown in Figure 4. A summary of the operating characteristics and quantities of the various types of channels in the LEASAT communications subsystem is shown in

Table IV.



Spacecraft in STS

Solid motor plus ejectable case

Liquid motor bipropellant fuel

Hydrazine fuel used in ascent

Spacecraft on-station

Stationkeeping fuel

Spacecraft dry weight

Weight, lbs













Attitude control

Reaction control (dry)

Liquid axial motor (dry)

Power and harness

Thermal control


Dry hardware

Minimum margin

Allowable spacecraft dry weight

Weight , lbs
















Attitude control

Thermal control

Power and distribution



* Eclipse season heaters.

Weight, lbs









Channel Type No.

Bandwidth (kHz)




Fleet broadcast








Onboard processing

* These are specified minimum values over coverage areas

EIRP * (dBw) G/T* (dB/K)










The wideband channel subassembly consists of four main units: the preamplifier, the receiver, the transmitter, and the output multiplexer. The preamplifier assembly is common to all the channels being received at UHF, which include the narrowband and relay channels in addition to the wideband channel. Uplink UHF signals are received through a separate receiving helical antenna, filtered in a bandpass filter to reduce transmitter leakage signals that couple from the transmitting antenna, and amplified in one of two redundant low noise UHF preamplifters. The redundant outputs of the preamplifters are cross-strapped through a hybrid network and split to separate receivers for the various types of channel groups.

The wideband receiver is shown in Figure 5. The signal from the preamplifter is downconverted to a fixed IF at a center frequency of 25 MHz. The downconversion local oscillator (LO) signal is generated in a reference generator which may be tuned to one of four frequencies by command to accommodate the four frequency plans, W, X, Y, and Z.

Functionally, the same design is used to accomplish the selectable frequency plans and is shown in Figure 6.

The channel crystal filter establishes the channel passband characteristics with a 1 dB bandwidth greater than 480 kHz and a 60 dB bandwidth less than 2 MHz. The filter output signal is fed to a limiter to normalize the signal level for receiver input signal levels over a

46 dB dynamic range. The gain of the limiter is set to provide limiting on front end noise within the crystal filter bandwidth.

The limiter output signal is upconverted to the transmit frequency with a second LO signal from the reference generator. Both downconverter and upconverter LOs are stepped in frequency by the same amount to maintain a constant translation frequency when tuning to one of the WXYZ frequency plans.

The receiver unit is completely redundant, with cross-strapping at the input and output of the signal path. The output from the receiver unit drives redundant UHF transistor power amplifiers. An amplifier consists of a two stage low level module, a single stage hybrid coupled driver, and a paralleled pair of hybrid coupled output stages. The redundant amplifier outputs are selected by a coaxial relay switch. The power level at the switch output is greater than 17.2 dBw.

The output from the switch is coupled to the UHF transmitting helical antenna through the

UHF multiplexer. The multiplexer consists of a series of high Q bandpass filters, each tuned to a particular channel frequency and interconnected to a common coaxial transmission line. The spacing of the filters along the transmission line is designed to achieve isolation between the filters. A six section filter with 3.6 MHz bandwidth is provided for the wideband channel. The filter has more than 100 dB rejection over the

UHF receiver band to prevent transmitter noise and spurs at the receive frequency from degrading receiver sensitivity.


The layout and functioning of the 25 kHz channels are similar to that of the wideband channel. The preamplifier assembly is common to all channels being received at UHF, including the narrowband and wideband channels in addition to the relay channel. Cross strapped outputs of the redundant preamplifier are fed to the relay receiver for processing of the relay channels.

The relay receiver is shown in Figure 7. The six relay channels are downconverted as a group to an IF ranging from 15 to 28 MHz. The downconversion LO is generated in a reference generator that is locked to the system master oscillator in a manner similar to that previously described for the wideband channel. A nearly octave bandwidth IF amplifier follows the mixer in the downconverter to provide gain prior to a large loss incurred in splitting the channels.

The IF amplifier output in the downconverter is split in a six-way power divider and distributed to the individual relay channels. A crystal filter at the input of each IF channel establishes channel passband characteristics with a 1dB an width greater than 24 kHz and a 60 dB bandwidth less than 110 kHz. The crystal filters for the six channels are identical except for the center frequency.

In each of the channels, the filter output signal is fed to a limiter whose gain is set to provide limiting on front-end noise within the crystal filter bandwidth in the absence of an input signal in the channel. Each limiter output signal is individually upconverted to the transmit frequency with a second LO signal from the reference generator. The entire group

of relay channels operates on the same frequency plan, but independently of the other groups of UHF channels.

The receiver unit is completely redundant, except for cross-strapping passive hybrid and power dividing networks. The redundant downconverters are cross-strapped at the input and output, allowing either downconverter to drive all 12 filter/limiter upconverter channels (six redundant pairs). The 19 filter/limiter upconverter channels are connected in two banks of six each, such that a given channel may be selected from either bank independent of the other channels. Each individual upconverter output is hybrid coupled with its corresponding redundant upconverter to provide cross-strapping at the receiver output. The two redundant LO reference generators also are cross-strapped so that either generator may supply the downconverter LO to all of the channels.

The receiver unit outputs drive redundant UHF transistor power amplifiers which are very similar to the wideband units. The power level at the switch output is greater than

15.2 dBw. The switch output in each channel is coupled to the UHF transmitting helical antenna through the UHF multiplexer. In the multiplexer, a three section filter with

0.7 MHz bandwidth is provided for each of the relay channels, 3 through 8. The filter has more than 100 dB rejection over the UHF receiver band to prevent transmitter noise and spurs at the receive frequency from degrading receiver sensitivity.


The basic design of the narrowband channels is similar to that of the relay channel except that there is a common power amplifier for the five channels.

The narrowband receiver is shown in Figure 8. The five narrowband channels are downconverted as a group to an IF of 8 MHz. The downconversion LO is generated in a reference generator in manner similar to that previously described for the wideband and relay channels. The reference generator is phase locked to the 5 MHz reference signal from the system master oscillator to achieve high frequency accuracy. An IF amplifier follows the mixer in the downconverter to provide gain prior to a large loss incurred in splitting the channels.

The output from the IF amplifier in the downconverter is split in a five-way power divider and distributed to the individual narrowband channels. A 4 kHz crystal filter at the input of each IF channel establishes the channel passband characteristics with a 1 dB bandwidth greater than 4 kHz and a 60 dB bandwidth less than 20 kHz.

In each of the channels the filter output signal is fed to a limiter whose gain is set to provide limiting on front end noise within the crystal filter bandwidth in the absence of an input signal in the channel.

The LO signals from each channel are passed through a second crystal filter and recombined in an IF summer. The crystal filter prevents degradation of the noise floor of adjacent channels due to spectral spreading caused by the limiter. The combined group is then upconverted in the receiver unit to the transmit frequency with a second LO signal from the reference generator.

The receiver unit is completely redundant except for cross-strapping passive hybrid and power dividing networks. The redundant IF chains are cross-strapped at the input and output, allowing either downconverter to drive all ten filter/limiter channels (five redundant pairs).

The upconverted UHF output from the receiver unit drives a redundant UHF power amplifier. Due to the stringent intermodulation requirements, a pair of amplifiers is operated in parallel with reduced output power in each amplifier. The switch output level is greater than 5.1 dBw per channel.

The output from the switch is coupled to the UHF transmitting helical antenna through the

UHF multiplexer. A three- section filter with 3.2 MHz bandwidth is provided for the narrowband transmitter. The filter has more than 74 dB rejection over the UHF receiver band to prevent transmitter noise and spurs at the receive frequency from degrading receiver sensitivity.


The fleet broadcast group is shown in Figure 9. The signal is received on the earth coverage receive super high frequency horn antenna, and downconverted in the SHF receiver to the IF required by the fleet broadcast processor. After processing, the resulting signal is routed to the bypass electronics where it is filtered, limited, and upconverted to

UHF for transmission through the channel one 35 watt transmitter. In the bypass mode, the processor is bypassed and the output of the SHF receiver is routed directly to the bypass electonics. The SHF beacon transmitter is included to transmit processor status information through the earth coverage transmit SHF horn antenna.


Historically, one of the most severe problems encountered in UHF communication satellites has been with passive intermodulation products. Weak nonlinearities in the common transmit path will generate signals in the spacecraft receive band.

The severity of the IM problem is determined by both the system definition (number of carriers and frequency plan) and the spacecraft design. LEASAT has 13 downlink channels, and the resulting order of the lowest IM which will fall in a receive channel is fifth order as shown in Figure 10.

Spacecraft design to minimize IMs involves eliminating poor metal-to-metal contacts and ferromagnetic materials in the common transmit path, minimizing UHF currents flowing on the external spacecraft surface, and maximizing the transmit to receive antenna isolation.

The separate receive and transmit helices have been located as high above and forward of the spacecraft structure and as far apart from each other as other system design constraints would allow.


1. Braverman, D. J. and Waylan, C. J., “LEASAT Communication Services,”

Proceedings ICC ’79, June 1979.

Figure 1. Worldwide LEASAT System

Figure 2. LEASAT Launch Configuration

Figure. 3. LEASAT Deployed Antenna Configuration

Figure 4. Communication Subsystem Functional Groups

Figure 5. 500 kHz Receiver Block Diagram

Figure 6. UHF Channel Group Typical Signal Flow

Figure 7. 25 kHz Channel Receiver

Figure 8. 5 kHz Receiver Block Diagram

Figure 9. Fleet Broadcast Block Diagram

Figure 10. LEASAT Frequency Plan Passive IM Products



Lt. Col. J. Baker and Lt. S. Vest

Space Service Division

Air Force Systems Command, Los Angeles, California and

A. S. Gilcrest and T. M. Rodriguez

The Aerospace Corporation, El Segundo, California


The motivations are discussed for considering a dedicated military satellite system that will provide tracking, telemetry, and command (TT&C) services, wideband mission data relay to the continental United States (CONUS), and narrowband mission data relay on a worldwide basis for United States Air Force (USAF) satellite systems. Mission models for the next 20 years are discussed. A concept study is in progress, and the guidelines for this effort are presented.


The USAF currently maintains worldwide facilities for supporting approximately 50 satellites that carry out communication, navigation, meteorological, surveillance, and research and development (R&D) missions. Studies performed in the late 1960s and mid1970s have revealed the potential benefits to be derived from supporting these mission satellites with relay satellites and from the use of CONUS-based ground facilities instead of the present complement of dispersed worldwide ground sites. To date, the development of a Satellite Control System has not been pursued for a combination of reasons, which include insufficient cost benefits, an inadequate technological base, and a lack of mission requirements that could only be satisfied by a spaceborne relay system.

Because of the military’s increased dependence on space assets, there is a greater need for more survivable space systems, a desire for less reliance on foreign countries, an increased potential for mission requirements that can be fulfilled only by relay systems, and an evercontinuing need to more cost effectively satisfy functional requirements. Thus, the USAF has initiated concept studies (1 and 2) to reexamine the entire spectrum of spaceborne satellite control and data relay issues.

* This research was supported by the Space Service Division (SSD) of Air Force Systems

Command under Contract No. F04701-79-C-0080.

In this paper, the needs of a satellite control system that might eventually become requirements for such a system are discussed, and postulated mission models that provide a basis for defining system alternatives are presented. The most significant guidelines imposed on the concept studies are delineated.


General representation of the satellite control system concept is depicted in Figure 1. The main features of the system are the duplex interconnections between a variety of mission satellites with CONUS-based fixed or mobile ground control nodes via relay satellites.

Two-way communications between both fixed and mobile overseas military are shown.

The functional concept provides for total TT&C as well as mission data processing support from CONUS-based ground nodes for all mission satellites supported by the

USAF. The links from CONUS to the mission satellites would support low (to 100 kbps) to medium (to 1 Mbps) bandwidth commanding and tracking signals. In addition, low bandwidth signals consisting of processed mission data could be forwarded from CONUS to overseas fixed or mobile military users. The return links from mission satellites to

CONUS would consist of medium bandwidth tracking signals and wide (to 5Gbps) bandwidth mission data. The system network would also provide for the direct distribution of low bandwidth mission data from mission satellites to worldwide users. Some limited control of network routing functions is postulated from certain CONUS or worldwide users, such as Airborne Command Posts and certain theater-based mobile or fixed assets.

The Satellite Control System concept could support the following current and future DoD needs:

1. Increased Survivability

Both physical and electronic survivability of satellite assets will be significantly enhanced by the Satellite Control System. Physical survivability will be improved by obviating the need for overseas ground-based facilities for TT&C or mission-related functions.

Electronic survivability will be improved by the increased antijam capability naturally derived from the use of 60-GHz (high atmospheric attenuation) or very narrow beamwidth optical beams for the cross-links between the mission and relay satellites. Communications between the relay satellites and the surface of the earth are antijam enhanced by utilizing spread spectrum techniques. In addition, communications between the relay satellites and

CONUS will be very narrow beam, which will further improve antijam characteristics.

Limited network routing control from airborne or ground mobile points will be possible in the event of nuclear warfare.

Figure 1. Satellite Control System Concept

2. Unique Mission Requirements

A given TT&C or mission ground facility has limited access to other than an equatorial synchronous altitude satellite within its own field of view. The problem is especially exacerbated for low-altitude satellites. For instance, the USAF Satellite Control Facility has remote tracking stations at seven worldwide locations, and for satellites at a 200-nm altitude, access duty cycles are less than 7%. Any mission requirement for either continuous access or access at a particular time between the ground and mission satellites can, in general, only be satisfied by a satellite relay system. Mission requirements that necessitate continuous or frequent communication accesses between the ground control point and the mission satellites include real-time or near real-time mission data transfer needs, inability or impracticality of recording mission or telemetry data, or continuous command capability to aid in dynamic contingency situations.

3. Reduced Dependence on Foreign Assets

Changing political alignments and pressures exerted by continuing localized crises dictate that less reliance be placed on assets in foreign countries.

4. Potential Economic Benefits

Globally dispersed ground facilities for either TT&C or mission processing are expensive to build and maintain. Reducing or eliminating the need for such facilities provides significant potential for cost savings. In addition, the unique mission benefits that can be practically provided only by the system described herein potentially result in a costeffective solution to increased military effectiveness. Finally, the cost-effective potential for increasing the survivability of this country’s military space assets is believed to be considerable.


Postulated mission models were formulated for the 1984 to 1989 and the 1990 to 2000 time periods. The basic model functions are depicted in Figure 2 and matrixed in Table I.

Twenty different satellite systems (114 individual satellites) are assumed for all mission models. For the near-term model (user Model No. 1), all ground nodes are assumed to be contained within CONUS. The model permits five ground stations dispersed throughout

CONUS. Mission Model No. 2 covers the 1990 to 2000 time period and includes the worldwide distribution of processed mission data of up to 100 kbps. Mission Model No. 3 is for the same time period and also includes some limited network routing from other than

CONUS ground-based locations. The main distinction between the earlier and later models is the higher assumed mission data rates (5 Gbps versus 1 Gbps maximum) for the later period. The requirement for handling higher mission data rates is somewhat counterbalanced by assuming that by the 1990s certain satellite programs will use onboard data processing of mission data so that links at 100 kbps or less are directly available to worldwide field users. For such satellites, it is assumed that the nonprocessed wideband data must be returned to CONUS only 2% of the time.

For each satellite system, the following model parameters are defined:

1. Number of Prime and Backup Satellites

2. Orbital Altitude Range

3. Telemetry, Commanding, and Raw Mission Data Rates

4. Link Duty Cycles

5. Priority

There are three priorities:

1. Priority 1 - Scheduled activities will occur on schedule 99. 7% of the time.

2. Priority 2 - All scheduled activities must be completed. However, support phasing may be adjusted 50% of the time.

3. Priority 3 - All scheduled activities must be completed. However, support phasing may be adjusted 100% of the time.

The mission models contain additional details on data sources and sinks so that an objective baseline exists for defining alternative options for a Satellite Control System.


Before concept study contracts could be initiated, guidelines and requirements over and above those covered in the mission model had to be formulated. The principal guidelines are summarized in Table II. The concept studies are currently under way, and final results will be available by late 1979 or early 1980.


Because of the military’s increased dependence on space, more survivable space systems are required. The system concept delineated in this paper could provide an additional measure of survivability and at the same time fulfill mission support functions for which there are no practical alternative means of accomplishing. System alternatives for satisfying the postulated mission model requirements delineated in this paper are currently being formulated by the study contractors (1 and 2). Costs will be determined for each alternative. In addition, cost savings that would result from the potential elimination of existing or planned capabilities that would not be required with the employment of a satellite control system will be delineated. Thus, the referenced studies will provide the basis for a cost-effectiveness analysis. The study results will be available by late 1979 or early 1980.

TABLE I. Satellite Control System User Models








1984 - 1989


1990 - 2000

1990 - 2000

Data Sources

TT&C Mission


C 2 from








Data Sinks












Figure 2. User Model: TT&C plus Mission Data


The authors express their gratitude to Col. B. Browning, Space Services Division, AFSC, for his support and guidance. Also, our gratitude is extended to the following individuals who were instrumental in formulating the mission models and initiating the concept study contractual effort: Lt. Col. E. Gruenler, Space Services Division, AFSC, and D. Moore,

L. Rider, and D. Speece, The Aerospace Corporation.


1. Satellite Control and Data Relay System, Contract No. F04701-79-C-0045, Contractor:

Stanford Telecommunication, Inc., with subcontractors Ford Aerospace and

Communications Corp. and Science Applications, Inc.

2. Laser Communications, Contract No. F33615-76-C-1052, Contractor: McDonnell

Douglas Corporation.



Frequency Ranges




TABLE II. General Guidelines Under Consideration

Worldwide Links


Worldwide coverage for satellites exceeding 200-nm altitude

Global broadcast coverage of processed mission data (100 kbps) between 70ES and 70EN latitudes

Downlinks Uplinks

15 Ghz

20 Ghz

40 Ghz

18 Ghz

30 GHz

49 Ghz

7 Ghz

20 Ghz

60 GHz or optical

System must operate in a USAF designated physical and electronic threat environment.

Redundancy for all functions.

8 GHz

30 Ghz

Data Rates

Mission Data





CONUS Ground

Entry Points

Maximum processed = 100 kbps

Maximum raw = 5 Gbps

Maximum = 128 kbps

Maximum = 10 kbps

Five entry regions

Satellite TT&C and its mission data may be transmitted to separate

CONUS entry points.


John Dunstan

Rockwell International

Space Systems Group

Space Shuttle Instrumentation

12214 Lakewood Blvd

Downey, CA 90241


The launch and orbital phase of the Shuttle is comparable to the Apollo flight. The entry phase, on the other hand, presents many new challenges to a reusable vehicle. To explore this area and provide more detailed data than that required for flight, the Shuttle technology flight instrumentation (TFI) system was proposed.

This paper discusses the TFI, which records flight data during the operational phase of the

Space Shuttle. It also deals with pertinent background information, such as Shuttle operation, flight verification, and instrumentation provided for the developmental and operational phase.


It launches like a rocket, hauls like a truck, and lands like an airplane—it is the Space

Shuttle. Our future space transportation, the Shuttle vehicle is the key to opening up near space to productive and economic application of both manned and unmanned exploration.

A quick look at the Space Shuttle and a typical mission will demonstrate the need for

Rockwell International’s instrumentation system.


The Space Shuttle consists of the orbiter, two solid rocket boosters (SRB’s), and one expendable external tank (ET). At launch, the Shuttle weighs approximately four and onehalf million pounds. With the five engines producing about six and one-half million pounds of thrust, the Shuttle starts its flight. On reaching an altitude of 28 miles it expends the two

SRB’s (5.2 million pounds of thrust for 2 minutes). The SRB’s separate from the external tank at this point and parachute down to a safe water landing. The boosters are designed to be recovered, towed back to the launch facility, refurbished, and readied for the next flight

within two weeks. The external tank provides liquid oxygen and hydrogen to the three main engines which are located on the orbiter. The ET separates just before the orbiter enters a normal orbit of about 100 nautical miles. The orbital maneuvering subsystem

(OMS) engines, located in the OMS pods at the base of the vertical tail, provide about

3000 pounds of thrust each. After ET separation, the OMS fires, completing the orbital insertion. A nominal seven-day mission of space exploration in a zero-g and vacuum environment then follows. The mission can expand to as many as 30 days with the addition of more consumables. Upon completion of the mission, the OMS fires to deorbit the orbiter for entry and landing. The thermal protection system (TPS) is designed to act as a thermal blanket to protect the orbiter aluminum structure from exceeding 350EF as it reenters the atmosphere. The orbiter lands at Kennedy Space Center or Vandenberg Air

Force Base.

The Shuttle is made up of six primary systems consisting of: (1) structures, (2) propulsion,

(3) control, (4) maneuvering, (5) crew environmental control, and (6) thermal protection.

To verify each system, extensive ground testing, simulating flight environments and conditions, is being completed in special laboratories. Flight testing is scheduled where modeling or ground testing is impractical.


Instrumentation has been developed in two phases, operational and developmental.

Operational flight instrumentation (OFI) consists of approximately 3000 housekeepingtype measurements. These are mandatory system measurements to provide the flight and ground crew with the state, performance, and condition of the orbiter subsystems.

Development flight instrumentation (DFI) consists of approximately 3500 measurements to verify that the OFI system provides sufficient data to confirm system performance and to verify that the assumptions made in developing the system models are valid. The Rockwell philosophy stressed maximum use of analytical modeling and ground tests, with recourse to flight testing only to verify the modeling.

The OFI system is located everywhere in the orbiter, except for the payload bay area. As shown in Figure 2, three avionic bays are in the crew module and three in the aft fuselage.

The DFI system consists of one avionic rack in the lower crew module and three DFI containers mounted in the payload area. Both OFI and DFI measure thermal, dynamic, structural strain, hinge moments, and other primary subsystem parameters.

The DFI system is scheduled for the first four flights only because scheduled payloads occupy the full payload bay for subsequent operational flights. A typical mission timeline

(Figure 3) shows that in less than 20 minutes the SRB and ET separate, and the orbiter enters an eliptical orbit. After OMS burn, the orbiter enters the planned circular orbit.

Entry and landing take about thirty minutes. This means that in four flights the instrument systems will obtain 80 minutes of launch data and 120 minutes of entry and landing data.

These are the two phases of the mission where the instrument technology is critical. The launch and space environments will be similar to the Apollo and Skylab missions and should present no new problems. The entry phase (Figure 4) presents the most challenging phase for the recoverable orbiter. Many new questions are anticipated in this phase.

To resolve the many questions pertaining to vehicle aerodynamics, flight performance, stability, handling qualities, and weight distribution during re-entry, the OFI measurement system had to be either expanded or complimented with another DFI-type system. Rather than compromise the OFI system, the technology flight instrumentation (TFI) system was proposed to resolve this problem. The ground rules were simple—use existing qualified

DFI equipment, keep it out of the payload area, keep the weight and power down, and record the data with no impact to the OFI system.

Technical Flight Instrumentation

The primary purpose of the TFI system is to support the unique orbiter-related experiments. Payloads are not usually considered orbiter related and will not be part of the

TFI system. The payload bay environment, however, is orbiter related and will be measured by TFI for various payload configurations.

The baseline TFI system (Figure 5) consists of DFI sensors installed throughout the orbiter, DFI signal conditioners, two pulse code modulators-master units (PCM masters) with eight PCM slaves, a frequency division multiplexer (FDM), and experiments in the development stage. The system shown will handle about 1000 analog measurements and

60 wideband measurements. A 28-track modular airborne recording system (MARS) recorder will provide 2 hours of recording at 15 inches per second tape speed. The system can be controlled either by the crew or via uplink command using the orbiter processing system.

The proposed TFI equipment will be installed in the wing carry-through area and the forward nose wheel well (Figures 6, 7, 8 and 9).

Of the 3500 DFI measurements, over 1400 have been baselined for TFI. These are primarily the temperature- and pressure-related measurements to the thermal protection system (TPS). No orbiter subsystem measurements have been requested for TFI.

The 1400 measurements are in addition to the measurements directly added by each experiment. Since the TFI system will handle only 1000 analog and 60 wideband measurements, many will not be recorded on each flight. Measurements to be recorded

first will be the active experiments and then the TFI measurements required to complement and support these experiments.

Orbiter Experiment Program (OEX)

The OEX experiments that are being developed for flight are designed to provide some of the answers in the area of uncertainty in a Shuttle mission — that is, entry heating. Four of the experiments are identified in the TFI block diagram (Figure 5).

Few questions deal with the eight to ten minute launch phase because the technologies developed for the Apollo and Skylab launches apply. The entry and landing, starting after the deorbit burn, present a new set of difficulties requiring detailed system verification.

The temperatures that the orbiter will experience during the blackout area are shown in

Figure 10. The TPS is designed to limit the orbiter structure maximum temperature to

350EF for 100 missions without replacement. The surface temperatures range from 600 to


Aerodynamic Coefficient Instrument Package (ACIP) — The ACIP is designed to provide high-resolution state data required for post-flight research quality aerodynamic analysis.

(The OFI system accuracy will not satisfy this requirement.) The ACIP consists of the triaxial linear and angular accelerometers and the tri-axial rate gyros. It is mounted in the wing root area of the mid fuselage. The instruments are aligned to a baseplate in the lab, and the baseplate is aligned to the orbiter axis during installation. (See Figures 11 and 12.)

Flight data will be recorded on entry from 400,000 feet through roll out. The instrument parameters are shown in Table I. ACIP is installed on the orbiter for all flights.

Table I. ACIP Parameters


Linear Accelerometers






(a y z x




Rate Gyros








±1.5 g

±0.5 g

±3.0 g

±30 deg/sec

±10 deg/sec

±10 deg/sec


183 µg

61 µg

366 µg

3.6 mdeg/sec

1.2 mdeg/sec

1.2 mdeg/sec

Angular Accelerometers







±1.0 rad/sec 2

±2.0 rad/sec


±2.0 rad/sec


122 µrad/sec

244 µrad/sec


244 µrad/sec



Shuttle Entry Air Data System (SEADS) — The air data parameters obtained from the operational air data system (Figure 13A), although accurate enough for flight, will not satisfy the research air data parameters. The data in Table II provides an example.

Table II. Example Air Data Parameters


Mach number

Fit Phase








Accuracy (percent)

Min Accep









Dynamic pressure Entry












Note: Other parameters involved include pressure, altitude, true air speed, angle of attack, air density, and temperature.

The purpose of the SEADS is to improve the air data measurements by providing the freestream condition across the entry speed profile. To provide the required system accuracy with the nose cap’s non-symetrical geometry and large angle of attack, 10 to 40 degrees, 14 pressure ports were added, as shown in Figure 13B. Pressure measurements of

0 to 20 and 0 to 2 pounds per square inch (absolute pressure) are made at each port. This dual pressure transducer configuration provides better than 5 percent data. Six radiometers

(Figure 13C) are used to measure the nose cone temperature in the port areas. Design problems are numerous, and a cross section of the nose cone area (Figure 13D) identifies a few.

The 3000EF temperature imposed on the tubes aft of the ports requires an expansion coil before the support manifold. The dynamic environment has dictated the coil shape and tube size, which is 0.25 inch outside diameter with a 0.015 inch wall. The problems are being resolved, and SEADS is scheduled to fly in 1981.

Shuttle Upper Air Mass Spectrometer (SUMS)

— The SUMS is designed to provide freestream environmental data at pressure levels below the lower SEADS range (above altitude 80 ft). SUMS provides a measurement of atmospheric mass density using a mass spectrometer. The inlet port (Figures 8 and 13E) is just aft of the nose cap and close to the centerline of the orbiter. The instrument is under vacuum until approximately 30 minutes prior to entry. The entrance port valve will be activated, opening the mass spectrometer to atmosphere. Pressure measurements desired range from 20 to 10


Torr. The system will be deactivated by closing the entry port valve upon reaching 80,000 feet and/or 10 -3 Torr.

SUMS is scheduled to fly in 1981.

Shuttle Infrared Leeside Temperature Sensing (SILTS)

— The SILTS experiment consists of an infrared scanner mounted on top of the orbiter vertical fin. Its purpose is to obtain detailed temperature maps of the upper wing area and upper mid body surface area.

(See Figures 14 and 15.)

The infrared camera scans a 40E surface area through two silicon windows. The windows are cooled by flowing gaseous nitrogen to approximately 200EF during entry. The camera pointing system provides for a 2.3 second view through the left window, a 2.3 second view of a reference black body located between the wondows, and then a 2.3 second view through the right window. The scan continues left to right and right to left. The SILTS is activated 5 minutes prior to entry and records data for 30 minutes.

Nineteen temperature measurements located within the TPS in the scanned area provide additional data. Temperature, pressure,and sound pressure level sensors typical of those 19 are installed in the TPS as shown in Figures 16, 17 and 18.


The four scheduled development flights will provide approximately 80 minutes of launch data and 120 minutes of entry and landing data. The instrumentation will provide the measurement data necessary to verify that the assumptions made in modeling were sufficient to verify subsystem mission performance. To resolve what happens when deviations from a normal launch and entry occur, additional experiments, as well as a flight measurement system, will be flown. The technology flight instrumentation system will provide approximately 1000 analog measurements and 60 wideband (dynamics) measurements.

Figure 1. Typical Space Shuttle Mission Profile

Figure 2. Orbiter Avionic System Installation Configuration

Figure 3. Typical Mission Timeline, Payload/Tug Deployment and Retrieval

Figure 4. Orbiter Entry and Return Flight Profile

Figure 5. OEX/TFI

Figure 6. TFI Equipment Location

Figure 7. TFI Equipment Panels

Figure 8. Shuttle Upper Atmosphere Mass Spectrometer, Orientation View

Figure 9. Shuttle Upper Atmosphere Mass Spectrometer, Schematic View

Figure 10. Maximum Temperatures for Entry Trajectory 14414.1

Figure 11. ACIP Installation

Figure 12. ACIP Configuration

Figure 13. SEADS Air Data System

Figure 14. SILTS Experiment and Support Hardware

Figure 15. SILTS Experiment Infrared Camera Coverage

Figure 16. Typical TPS Thermocouple (T/C) Installation

Figure 17. Typical Sound Pressure Installation in Aft TPS

Figure 18. Pressure Ports in the TPS HRSI and FRSI



J.C. Hoagland

Member Technical Staff

Space Systems Group

Rockwell International

Downey, California


During operational space flight, the communications and telemetry subsystem of the Space

Shuttle orbiter uses S-band and Ku-band links to provide, in addition to tracking, reception of digitized voice, commands, and printed or diagrammatic data at a maximum rate of 216 kilobits per second (kbps). The subsystem also provides a transmission capability for digitized voice, telemetry, television, and data at a maximum rate of 50 megabits per second (mbps). S-band links may be established directly with a ground station and both

S-band and Ku-band links may be routed through NASA’s tracking and data relay satellite system (TDRSS). A simultaneous capability to communicate with other satellites or spacecraft, using a variety of formats and modulation techniques on more than 850 S-band channels, is provided. Ultrahigh frequency (UHF) is used for communication with extravehicular astronauts as well as for a backup subsystem for state vector update. Audio and television subsystems serve on-board needs as well as interfacing with the radio frequency (RF) equipment.

During aerodynamic flight following entry, the S-band link can be supplemented or replaced by a UHF link that provides two-way simplex voice communication with air traffic control facilities.


The orbiter’s communication and tracking subsystem (C&TSS) is an unusual combination of complexity and simplicity, specialization and versatility, and off-the-shelf and newly developed hardware. It must interface with not only NASA’s space tracking and data network (STDN), but also with NASA’s TDRSS, the USAF space ground link subsystem

(SGLS), other satellites, crew members performing extravehicular activities (EVA), and the Federal Aviation Agency’s (FAA) air traffic control (ATC) voice communications. In

addition, it must interface with the multiple on-board computers of the data processing subsystem, the orbiter displays and controls, other on-board subsystems, and payloads.

All of the orbiter’s on-orbit communication links (Figure 1) may be employed simultaneously. For S-band operation, either the TDRSS or STDN can be employed, but not simultaneously. The design of the C&TSS was driven by several factors (Table I).

Table I. Design Drivers

Design Driver





Flush or deployable antennas



Objective to make Shuttle a low-cost launching system

Objective to reuse orbiter up to 100 times


• Compatability with existing ground facilities

• Use of off-the-shelf when available

• Limited new component development

• Many environmentally sealed boxes

• Extended environmental testing

• Almost completely redundant communication.

Requirement that two failures not endanger crew or vehicle and a single failure not force mission termination

Requirement to interface with both DOD and NASA ground networks and inter face with a wide variety of payload communication systems

Protruding antennas would burn off during entry

Requirement to be independent of ground support (operate without radiating)

• Two turnaround ratios

• Multiple payload data rates, formats, and operating frequencies

Considable difficulty in meeting performance requirements

Development of Global Posi tioning Subsystem to obtain navigation data from stable ground-reference signals

Design Driver

Power and weight

Long RF coax runs


Objective to maximize vehicle payload capability

Large size of orbiter


• More complex development trades and design effort

• Three collectors Ku TWT

Special efforts to minimize losses and improve antenna and receiver/transmitter performance

Because of the complexity of the C&TSS, this paper is limited to an overview of capability and design. In those cases where companion papers describe subsystems (S-band,

Ku-band, and antennas), details in this paper are further reduced and are listed in

References 1, 2, 3, and 4.


The C&TSS is most conveniently described in terms of subsystems or equipment groupings (Figure 2). Functional descriptions of these subsystems or equipment groupings are provided in the following paragraphs.

S-Band Subsystem

The orbiter’s S-band communication subsystem was designed and manufactured by TRW and subtier contractors. It is comprised of two independent subsystems, the network subsystem and the payload communication subsystem. The network subsystem provides tracking and two-way communication via phase-modulated (PM) links directly to the ground or through the TDRSS and transmission of data directly to the ground via frequency modulation (FM) link. The payload communication subsystem, like a flying ground station, provides two-way communication with unmanned orbiting spacecraft.

Network Subsystem — The network subsystem consist of 8 line-replaceable units

(LRU’s). Those not shown as being redundant (Figure 3) are internally redundant.

Therefore, the subsystem includes two electrically isolated strings (with the exception of the reeds and contacts in the switch assembly), the antennas, and their coaxial cables.

Although some cross-strapping of functional units between strings is possible to improve the capability to withstand failures, this flexibility is limited to minimize orbiter wiring complexity and weight.

As may be seen from the block diagram, the FM and PM functions are separate, except that the switch assembly services both. Not shown are the interfacing multiplexer/demultiplexers (MDM’s) that provide telemetry of configuration data,

performance parameters, and transmission and reception of data to and from the data processing subsystem (DPS) computers. The network signal processors (NSP’s) can route both data to be received and data to be transmitted through communication security boxes for decryption or encryption.

The subsystem provides for several modes and data rates (Figure 4) for both the forward link (ground-to-orbiter) and the return link 1 (orbiter-to-ground). Coding (Reference 5) is used in the tracking and data relay satellite (TDRS) modes to improve bit error rates

(BER). The forward link receiving equipment is capable of handling data at two different rates, spread with a pseudo-random noise (PRN) code rate of 11.232 megachips per second or not spread, and transmitting on any of four frequencies. Spreading is used on the

TDRS mode to reduce TDRS interference to ground-based communications by reducing the power flux density at the earth’s surface. The four forward link frequencies accommodate two return link frequencies and two turnaround ratios (ratios of orbiter transmit to receive frequencies). Two return link frequencies, which operate in the 1.7 to

2.3 GHz band, are used to minimize interference to payload communications. Two turnaround ratios correspond to those used by NASA (240/221) and DOD (256/205).

Two data rates are available for the return link, accommodating, as in the forward link, one or two voice channels, and, in addition, two different telemetry rates. The lower data rate is used when link margins are required, as is the case for a large portion of the time when communicating through TDRSS. In the TDRS low data rate mode, where link margins are on the order of +2 to +3 dB, the power amplifier generates over 100 watts with an effective isotropic radiated power (EIRP) of 17.7 dBW and the preamplifier provides a sensitivity of approximately -125 dBm at the antenna. The antenna gain-to-noise (G/T) value is approximately -27.3 dBEK with the dual beam antennas. In direct-to-ground communication, although the power amplifier and preamplifier are not used, a transponder output of 2 watts and sensitivity of -118 dBm provide much better link margins because of the reduced range and improved (as compared to TDRS) ground terminal performance.

FM Subsystem (Operational) — The FM subsystem consists of three LRU’s (Figure 3).

The FM signal processor and FM transmitters provide a capability for the transmission of data not amenable for incorporation into the limited-rate pulse code modification (PCM) telemetry data stream. The data to be transmitted via FM include television, digital data from the main engines during launch, wideband (to 4 MHz) payload data, and digital data from recorder playback of payloads.

1 The terms, forward and return links, were adopted in preference to up and down to avoid the confusion resulting from the usage of a relay satellite at a synchronous altitude where signals in both directions follow paths going both up and down.

Conditioning and multiplexing for FM transmission occur in the FM signal processor.

Video and wideband digital and analog signals are routed to the FM transmitter with only matching and filtering. Narrower-band digital engine data are placed on subcarriers at 576,

768, and 1024 kHz.

The FM transmitter operates at 2250 MHz with an output power of 10 watts. Both baseband and RF filtering is provided to reduce out-of-channel interference to the PM and payload receivers. Nominal RF bandwidth is 10 MHz.

Audio Distribution Subsystem

The audio distribution subsystem (ADS), designed and manufactured for the orbiter by

Telephonics, provides intercom and radio access functions for the various crew stations and hardline subscribers involved in an orbital mission. It includes facilities for audio processing, mixing, amplification, volume control, isolation, switching, and distribution. It provides paging capability, communication over various alternative audio bus circuits, distribution of caution and warning signals, and communication with ground crews during preflight vehicle checkout.

The ADS is comprised of six audio terminal units (ATU’s), two speaker-microphone units

(SMU’s), one redundant audio central control unit (ACCU), and various interface units; their selection depends on the kind of headset being utilized.

These LRU’s, except for the interface units, and their functional relationship are shown in the ADS system block diagram (Figure 5). This block diagram also indicates the relationship of the LRU’s of the ADS with the radio equipment, recorders, navigation aids

(NAVAID’s), and hardline installations that the system serves. Two additional ATU’s, one on the mid deck and one at the flight deck orbit station, are provided on Orbiter 102.

These are to be eliminated as a weight-saving measure on later vehicles.

The ADS utilizes hardwire baseband transmission of audio signals and time division multiplexed transmission of control signals. The audio central control unit (ACCU) acts as a central switchboard for the system; all audio routing is accomplished in the ACCU under control of switching commands originating at the ATU’s. The ATU switching commands are transmitted to ACCU in the form of serial digital data streams. The switching commands arrive independently over dedicated wires, one pair from each ATU. These control signals include channel selections, independent volume levels for each channel, and keying signals.

The mission requirements for ADS involve not only high reliability but also high speech intelligibility. The ADS audio circuits are designed to process and condition voice audio

signals over a large dynamic range with minimum distortion and minimum introduction of internal noise. In meeting these requirements, close attention was given to possible sources of degradation of the voice signal as well as to all points where the signal-to-noise ratio can be safeguarded. The specific techniques employed include:

• Bandwidth filtering to the range of 300 to 3000 Hertz with 18 dB/octave slopes

(minimum) at each end of the pass band.

• Usage of digital control data rates that have no repetition frequencies in the audio pass band.

• Syllabic-type automatic gain control with symmetrical clipping (followed by appropriate filtering) to standardize signal levels against variations in talking level, variations in microphone sensitivity, and distance from speaker to microphone.

• Controlled sidetone injection and extensive use of sidetone cancellation circuits to ensure that no ring-around conditions can be set up.

The overall speech intelligibility between the orbiter crew and the ground stations is above

96 percent when measured on the Harvard 1000 word test. With a weak radio link margin, and the total communication and telemetry system operating at a BER approaching 10 -2 the intelligibility is in excess of 80 percent.


Television Subsystem

The television subsystem allows visual monitoring from the ground of on-board activities.

It provides the crew with the ability to see areas of the payload bay obscured from direct observation. Television signals originating in the orbiter, near the orbiter, and its payloads can be transmitted to the ground on either of two links, and the FM direct S-band link, or, when it becomes available, the Ku-band TDRS link.

The operational television subsystem (Figure 6), designed and manufactured by RCA, will have up to nine on-board cameras, two large-screen monitors, two portable viewfinder monitors, and the associated switching and control logic.

All TV cameras are black and white, but may be converted to color with the substitution of a color lens assembly (CLA) for the normal monochrome lens assembly (MLA). This CLA contains a rotating color separation wheel to provide a field sequential color signal. Only the two cameras located inside the cabin are to be equipped for color. One of these same cameras may be carried by an EVA astronaut outside the crew compartment. These cabin cameras are the only cameras equipped with viewfinder monitors; the pointing of all other

cameras being either fixed or remotely controlled from the console television monitors


Up to three cameras may be located in the payload bay: one at the forward end, one at the aft end, and one (keel camera) at one of four locations on the floor.

The remaining four cameras may be located on the two arms of the remote manipulating system (RMS). These two jointed-arms are to be utilized in deploying and retrieving payloads. Cameras are provided at two locations on each arm: one at the elbow and one at the wrist. The elbow camera is mounted on a remotely controlled pan-tilt unit to adjust its pointing as desired. This same pan-tilt unit is provided for the forward and aft payload bay cameras. The camera at the wrist is fixed-mounted, but has a viewing light atop it for aid in viewing shadowed areas.

The two black and white CTM’s are located at the aft-end of the flight deck near the television control panel. Each has the capability for split-screen viewing, thus allowing monitoring of up to four cameras simultaneously.

These cameras and monitors are interconnected through a video switching unit (VSU), which performs switching in response to signals from the remote control logic unit (RCU).

Commands are decoded by the RCU and multiplexed on the sync signal along with a camera ID. The camera’s electronics decode these signals and drive the lens and pan/tilt motors. The camera multiplexes its ID number, temperature, pan/tilt angles, and angle rates on the composite video to the VSU.

In addition to its switching function, the VSU multiplexes Greenwich Mean Time from the orbiter’s master timing unit on the downlink or record video and encodes one of two audio channels on the video sync signal.

Whereas the orbiter-supplied cameras produce a field sequential color signal, the TV subsystem is designed to allow a camera that generates a National Television Systems

Committee (NTSC) color signal to be used. The signal format of the composite fieldsequential color video is in compliance with commercial broadcast standards and would produce a black and white picture on a home television receiver.

Ku-Band Radar/Communication Subsystem

The Ku-band subsystem, currently scheduled for installation later in the flight development program, is designed and manufactured by Hughes. It operates as a radar during space rendezvous to measure angles, angle rates, range, and range rate. When not employed in this manner, it can be used as a two-way communication subsystem. It transmits data

through the TDRSS at a rate of up to 50 mbps and receiving at a rate of up to 216 kbps. In both radar and communication modes, it uses a three-foot (0.9 meter) parabolic monopulse antenna that is mounted inside the front of the orbiter’s payload bay and deployed by rotation about a single axis after the orbiter is in space and its payload bay doors are opened.

The deployed assembly (DA), which includes the antenna and considerable electronics, is mounted on the starboard side of the vehicle. The location of the hardware and radar range, communication modes, and maximum data rates are presented in Figure 7.

In both radar and communication modes, acquisition of the radar target or the communication satellite is aided by the onboard computer’s designation of an angle around which a spiral search is conducted. Acquisition, thereafter, is automatic. Manual entry of antenna angles is also possible.

Hardware common to both radar and communications functions includes the antennas (the three-foot dish plus a small acquisition horn), the antenna drive mechanism, drive electronics, traveling wave tube (TWT) transmitter, and receiver front end.

Hardware is packaged in four LRU’s (Figure 8). All hardware, except the communications signal processor, are used for radar. All hardware except electronic assembly 2 (EA-2), are used for communications.

All externally generated control signals are applied to electronic assembly 1 (EA-1) and distributed with internally generated control signals to the other LRU’s. Control of radar functions is accomplished by feeding signals directly from display and control (D&C), but communication control signals originating at the D&C are routed through the ground control interface logic (GCIL), to permit communications functions to also be controlled from the ground.

Because failure of the communications function becomes self-evident rather quickly, provision is provided to automatically switch to the S-band network subsystem.

Table II provides some Ku-band subsystem parameters applicable to both radar and communications.

The parabolic antenna has two uncommon features. It is edge mounted with supports radiating from its mounting point and is constructed largely of graphite-epoxy to minimize thermal distortion. In angle tracking modes, antenna sum and angular error signals are processed. The elevation and azimuth signals are time-multiplexed, eliminating the need for a third processing channel.

Table II. Ku-Band Subsystem Parameters

Narrow Beam Antenna


Peak gain

3 dB beamwidth


Wide Beam Antenna





Common Parameters

TWT output

Receiver NF

Prime power

*Frequency dependent

Prime feed parabolic

38.4 - 38.9 dB*

1.57 -1.680'E



18 dB



50 W

5 dB max.

28 vdc

Although concern had been originally expressed over the wisdom of combining a communications and radar system, it is now obvious that savings in weight, volume, and developmental costs were attained without significantly degrading either function.

Radar Function — As a rendezvous aid, the Ku-band subsystem operates as a pulse doppler, frequency-hopping radar. The relatively long 66-microsecond pulses employed at longer ranges provide reasonable efficiency with the peak-power-limited TWT amplifier.

Short pulses of 22 nanoseconds are used to provide radar operation down to 100 feet

(30 meters). Pulse widths and repetition frequencies are selected to provide unambiguous measurement of both range and range rate for uncooperative (skin-tracked) targets to

10 nmi (18.5 km).

Because the orbiter in space performs the latter part of rendezvous and station-keeping by accelerating and braking along the z axis (the axis that runs vertically through the orbiter), the radar normally searches angles within 30 degrees of a straight upward pointing position. Tracking, however, may continue through larger angles until the beam is intercepted by the orbiter structure.

Radar accuracy characteristics are summarized in Table III. The velocity accuracy requirement led to the choice of a pulse doppler approach. Sixteen doppler filters cover the doppler interval defined by the repetition rate.

Table III. Radar Passive Target Requirements

Range accuracy (3F)

Velocity accuracy (3F)

Angle accuracy (3F)

Angle rate accuracy (3F)

80 feet or 1 percent

1 ft/s

8 milliradians

0.14 milliradians/sec

Like the S-band payload communication subsystem, the radar must operate over a large dynamic range. In addition to concern over receiver dynamic capability, nearby targets could be damaged by excessive energy from the radar. In order to prevent damage, the three levels of output power provided are 50 watts full power with -12 and -24 dB steps of attenuation.

An auxiliary (acquisition) antenna provides a convenient source for a guard signal, which is processed and compared to the narrow-beam sum signal to eliminate sidelobe targets.

Because the auxiliary antenna gain is 20 dB less than the gain of the main antenna and because antenna sidelobes are down 20 dB or more, there is at least a 20-dB difference in the main/ guard ratio for mainlobe and sidelobe targets.

Communication Function — Ku-band subsystem communications provide the orbiter with a highly flexible means of transmitting data at various rates and formats (Table IV). Except for the 192-kbps channel, which is comprised of the orbiter voice and telemetry

(operations data), other rates and bandwidths shown are maximums. From the rate ranges shown, it may be seen that the capability extends continuously from 16 kbps to 50 mbps.

Similarly, the 4.5 MHz analog channel extends downward to dc. The unusual signal design provides quadraphase shift keying (QPSK) of a subcarrier and either QPSK or FM of the carrier.

Table IV. Return Link Data Rates

The problem of mutual acquisition of the orbiter and TDRS has received considerable attention. In one acquisition scenario, the orbiter radiates Ku-band energy at the TDRS through the widebeam acquisition antenna. The TDRS locates this signal and points its narrow (0.36-degree) beam at the orbiter, which searches and acquires with the narrow beam antenna, and then, switches its transmitter to the narrow-beam antenna. In a similar scenario, the orbiter radiates S-band energy through the appropriate antenna and the TDRS points its S/Ku-band antenna at the S-band source. It even appears possible that available orbital parameters will be good enough to allow both TDRS and orbiter narrow-beam antennas to be pointed at each other with sufficient accuracy to achieve acquisition without search.

Failure of the forward link results in a signal being generated in the signal processor, which commands the GCIL to switch the NSP forward link input from the Ku-band to the S-band receiver. This precludes the possibility of the ground losing communication with the vehicle should a Ku-band forward link problem develop while the crew is asleep.

As in the S-band subsystem, spreading of the forward link is used to reduce interference to ground-based communications systems. The PRN code rate is 3.02803 megasymbols per second, less than that used at S-band.

Ground Command Interface Logic (GCIL)

GCIL is an LRU that provides the capability for ground control of many functions of

C&TSS and a portion of the operational instrumentation subsystem. It also provides the logic to allow control of the same functions from either D&C switches (manual commands) or in response to ground-originated commands through either the S-band or

Ku-band links. Ground-originated commands flow through the NSP to the general-purpose computers (GPC’s) of the DPS. Commands are sent to the LRU’s. Then, the command status (from either the GCIL or the LRU’s) is returned to the GPC and routed to the ground through the pulse code modulation (PCM) data stream. The GCIL provides logic to allow the on-board crew, if required, to block ground-originated configuration commands.

On-board commands may also be originated through the usage of any of several DPS keyboards that enter the command directly in the GPC.


The GCIL, in conjunction with the other described equipment, allows a ground crew to operate and monitor the C&TSS configuration, freeing the crew for other activities. It also avoids the necessity of having one astronaut awake at all times just to maintain contact with the ground.

2 The term, GPC, includes all five of the operations (OPS) computers on board, including those previously referred to as guidance, navigation and control (GN&C) computers.


UHF transceivers are provided for the transmission and reception of voice to allow contact with ATC facilities and chase aircraft during landing operation. They are provided during on-orbit operations for the transmission of voice to and the reception of voice and telemetry from extravehicular space-suited astronauts. Both functions are provided by a newly developed EVA-ATC communication subsystem being built by RCA under direct contract to NASA.

EVA/ATC Communication System — The EVA/ATC communication subsystem is designed primarily to support extravehicular activities, but it also provides ATC voice communication capabilities. Thus, allowing it to replace the ARC-150 already used on

Orbiter 101. In the ATC service, it provides two-way RF links on either of two frequencies

(296.8 or 259.7 MHz) with a transmit power of 10 watts. In addition, emergency communication is provided by a 243-MHz guard channel transmitter and receiver.

In EVA service, things are more complicated. To understand the various modes of operation, it is necessary to consider the extravehicular communicator (EVC) equipment carried by the EVA astronaut or astronauts. Basic block diagrams of both the orbiter’s

EVA/ATC transceiver and the EVC are presented in Figure 9.

The EVA unit consists of AM transmitters, AM receivers, a telemetry subsystem, a warning subsystem, and an antenna. This equipment is arranged to operate in several different modes. Mode A is the normal mode used by a single EVA astronaut and Mode B is the normal mode by a second EVA astronaut (Figure 9). Other combinations of receivers and transmitters can provide voice communication if there is equipment failure or interference on normal channels.

In the RF OFF mode, the equipment provides duplex voice operation utilizing an audio input/ output interface to the Shuttle orbiter via a service umbilical.

In Modes A and B, the operating transmitter is modulated by a 5.4-kHz (standard IRIG) subcarrier oscillator, which transmits biomedical data (electrocardiographs). Voice, which modulates the carrier directly, is keyed on by voice-operated circuitry (VOX) or a push-totalk (PTT) switch.

A 1.5-kHz warning tone generator, square wave modulated at 15 Hertz, operates in response to a sensor (external to the EVC) to alert the astronaut to conditions requiring his attention.

On the orbiter side of this communication link, receivers and 500-milliwatt transmitters are provided on each of the frequencies shown in Figure 10. Two antennas are provided: one inside the airlock and one on the bottom of the orbiter. The latter is the same one used with the ARC-150 during aerodynamic flight tests.

The processing within this system strips the electrocardiograph signals from one or two

EVC’s and provides them to the orbiter telemetry subsystem. A two-way voice interface with the orbiter’s audio distribution subsystem is provided, giving astronauts performing extravehicular activities access to orbiter voice communications on up to three voice channels. This enables an EVA astronaut to be in direct voice contact with the ground or the orbiter crew. The astronaut can also have his conversations recorded.

S-Band Payload Subsystem

The S-band payload subsystem provides the capability to communicate with a wide variety of satellites. It will be used for such purposes as checking the operation of an attached onboard or a released payload prior to moving from its immediate vicinity in the orbiter payload bay. Also, it can help with the safing of a satellite before taking it on-board for repair or return to earth. Two separate payload subsystems are available. First, a subsystem to operate with STDN or DSN, and second, a subsystem to operate with the space ground link subsystem (SGLS) network.

The subsystem provides for several modes and data rates in addition to multiple frequency selection for transmit and receive (Figure 11). The receiver and transmitter are packaged in a single LRU called the payload interrogator (PI), which is used for both NASA and DOD.

Signal processing in both directions is performed in either the payload signal processor

(PSP) for NASA payloads or the communication interface unit (CIU) for DOD operations.

Redundant LRU’s are carried for both the PI and PSP, while the CIU is a single unit

(Figure 12).

The PI provides 851 duplex channels for simultaneous reception and transmission of information with a coherent frequency of 256/205 in the DOD mode (20 channels) and

240/221 in the STDN (808 channels) and DSN (23 channels) modes. In addition, provision for six receive-only RF channels and four transmit-only RF channels in the DSN mode

(Figure 13) are included.

Because payloads are intended to communicate with the same ground stations as the orbiter, they also receive on the lower frequencies of the space communication band and transmit on the high frequencies of the same band. This means that the interrogator must receive on frequencies near those employed for transmission by the S-band network subsystem. Transmission frequencies are close to those where the network subsystem

receives data. Careful filtering is employed in both subsystems to minimize cross interference. The upper and lower frequency region is selected in the network subsystem to maximize separation from the payload channel in use (Figures 11 and 4).

Both the receiver and transmitter of the PI are capable of performing sweep for acquisition and automatically terminating the sweep upon acquisition. The receive sweep is selectable for a plus-or-minus 80 to 140 kHz range for signal acquisition within 5 seconds. The transmitter sweep is 540 Hz/s in the DSM or DOD mode for a sweep range of plus-orminus 33 kHz. Also in the DOD mode the sweep rate is 10 kHz/s for a sweep range of plus-or-minus 55 kHz. In the NASA mode, the sweep rate is 10 kHz/s for a sweep range of plus-or-minus 75 kHz.

The receiver section of the PI is equipped to demodulate the subcarrier for data processing by the PSP or CIU, or transfer via the Ku-band through the TDRSS to the ground.

The standard received signals are:

Item a) is compatible with STDN/DSN while Items a), b), c) are compatible with DOD and the Ku-band bent pipe.

The non-standard PI received signals must meet all of the following requirements to be bent pipe compatible for Ku-band transfer: a) PM 2.5 $ $6 $ 0.2 radians, with a minimum residual carrier of -114 dBm for acquisition and -116 dBm for tracking b) The inband data spectral components within ±100 kHz of the receive carrier frequency shall be 26 dB or more below the RF carrier power c) The intelligence contained in the received signal shall be # 4.5 MHz to be compatible with the one-sided 3 dB post detection bandwidth of the PI receiver

In addition to the above, the received signal, with direct carrier modulation by a nonperiodic digital information channel, is accepted provided that; the RMS phase noise component, because of modulation sidebands, is 10 degrees or less, and the maximum allowable number of transitionless bits will not contribute more than 18 degrees to the carrier phase noise.

The transmitter section of the PI is equipped to modulate the selected carrier signals generated by the PSP or CIU for transmittal to a NASA, DSN or DOD compatible satellite.

The standard transmit signals are: a)

$ = 1 ±0.1 radian, SCO 16 kHz, commands = n x 125/16, where n = 2 x and x integer 0 to 8 b)

2.5 $ $ $ 0.2 radian, SCO’s: 65 kHz or 76 kHz or 95 kHz, AM: 500 Hertz or 1000 Hertz.

Commands: 1 K-Baud or 2 K-Bauds

Item a) is compatible with STDN/DSN while items a) and b) are compatible with DOD.

The non-standard transmit signals are a high tone sinewave command as follows for DOD operation: a) Frequency range - 1 to 200 kHz b) Duration - “1” tone 0.1 to 3.5 seconds (TD)

- “0” space 0.1 to 3.5 seconds (SD) c) Type

Interval 0.1 to 6 seconds

- Sequential NRZ keying (Figure 14)

To aid in accommodating the large variation in signal strength, which results from a range that can vary from a few feet to several miles, the receive section of the PI automatically adds protective attenuation. Signals greater than -116 dBm at the orbiter payload antenna will be automatically tracked. Protection for +20 dBm signals are incorporated in the PI.

Three selectable EIRP power levels are provided by the PI. These are: +29 dBm

(maximum), +19 dBm, and -4 dBm.

The PSP and CIU demodulate the subcarriers, provide bit synchronization and frame synchronization. The PSP provides four frame synchronization word lengths (8, 16, 24, and 32 bits). Bit synchronization will accommodate Bi-N-L, -M, -S, and NRZ-L, M, S; one at a time, for a single selected data rate of 1, 2, 4, 8, or 16 kbps.

The demodulated telemetry plus clock and frame sunchronization is routed to a payload data interleaver (PDI). The PDI, a component of the instrumentation subsystem, interleaves the telemetry data with data from up to four attached payloads for eventual transmission to the ground in the PCM data stream via the network subsystem.

The CIU handles one of eleven data rates (0.25, 0.50, 1.2, 4, 8, 10, 16, 32, 48, 64 kbps)

Bi-N-L on a 1.024 MHz subcarrier and a constant data rate between 0.125 and 256 kbps

Bi-N-L on a 1.7 MHz subcarrier. The former data rate is transmitted to the ground via the

PDI while the latter is transferred via the FM subsystem or FM phase of the Ku-band subsystem.

The PSP also receives configuration and payload command messages from the DPS at a burst rate of 1 mbps. It responds to a configuration message by configuring itself to handle data at rates and formats designated. It buffers the commands to phase shift keyed modulate a 16 kHz subcarrier for transmission by the PI or hardline to attached payloads.

The CIU accepts command from the satellite control-facility (SCF) via the network subsystem and DPS or direct on-board generated command, which can override the DPS.

The command is buffered, verified, and changes from Bi-N-L to ternary and modulates a

1.024 MHz subcarrier for transmission by the PI to a detached payload.

Antennas — The antennas associated with each subsystem (except the UHF airlock and the deployable Ku-band) are flush mounted (Figure 15). The locations were chosen to favor the desired direction of, coverage within the constraints of the space available onboard the orbiter.

All flush antennas are overlaid with the thermal protection subsystem (TPS), which covers that part of the orbiter surface that otherwise would be unable to survive the heat of entry.

TPS is thickest on the bottom to a depth over the lower antennas reaching 2.5 inches. TPS has electrical characteristics somewhat similar to polyurethane foam. It has required special attention where the patterns are critical (quads). Basic data on the antennas are presented in Table V.

Table V. Orbiter Antennas



S-band quads

S-band hemi’s

S-band payload


Ku-band subsystem

Quantity Freq













Annular slot

Crossed dipole fed cavity fixed array

Crossed dipole fed cavity

Cross dipole fed cavity

Reason for Selection

High efficiency, broad angular coverage

Beam shaping, gain, efficiency

High-efficiency, broad, continuous coverage

Polarization switching, shaped beam


1 or








L radar

CP comm



Total coverage in the airlock cylinder

High gain, low sidelobes

The four dual beam quad antennas are placed in the roll plane of the vehicle at 45 degrees to the orbiter horizontal plane. The patterns of about 100 degrees in roll provide overlapping roll coverage; fore-aft coverage is on the order of 130 degrees. More details on orbiter antennas and patterns may be found in References 1, 3, and 4.


Communication control panels in the orbiter are not too different from their counterparts in large commercial aircraft and previously manned spacecraft. All panels (not just C&TSS) are designed and manufactured by the orbiter prime contractor, Rockwell International, to ensure commonality of component usage and standard layout and nomenclature.

The S-band control panel (Figure 16) makes use of block diagramming to aid in understanding switch functions; a technique that also shows on the bottom of the Ku-band control panel (Figure 17). The signal strength meter on the Ku-band panel is shared between the S-band PM (network) subsystem, the S-band payload subsystem, and the Kuband subsystem. The striped barber poles, two-state displays, are more reliable and easier to view under various lighting conditions than pilot lights.

Additional information on the C&TSS is available on the DPS cathode-ray display. Most interesting of the data that can be called up is the antenna status display, the center of which is shown in Figure 18. This display shows the volume around the vehicle as if a tube had been placed around the fuselage, split along the bottom, and flattened. The nose is a line across the top of the figure, the tail the line across the bottom of the figure, and both sides of the figure represent the bottom fore-aft centerline of the vehicle. The oddly shaped figures are the outlines of the angles where the Ku-band beam would be blocked by the vehicle. Moving symbols E, W, and S are the directions to east and west TDRS and to the nearest ground station. Obviously, they are not all always on the display. The square around the W indicates the Ku-band antenna is pointing toward TDRS west. The pairs of letters along the top of the display indicate that the S-band quad that best serves those

90-degree quadrants (LL = lower left, LR = lower right, etc.). This center display is surrounded by other antenna status and signal strength data. The GCIL unit that operates in conjunction with D&C’s has been previously discussed.


The C&TSS utilizes both environmental sealed and unsealed LRU’s, with most of those specifically designed for the orbiter being sealed as an aid toward meeting the goal of a ten-year life. A typical box is approximately 7.6 inches (19 cm) high and no longer than

20 inches (51 cm), with width being determined by the volume needed. A flat thermal base provides contact with water-cooled shelves on which or under which it mounts

(Figure 19). Hold-down is by captive fasteners. Connectors are on the front panel. The physical design was greatly influenced by the sealing requirement, conduction cooling, and by the design vibration requirement of 0.03 g


per Hertz.

A summary of subsystem weight and power consumption is provided in Table VI. Power dissipation is about the same as consumption except for those LRU’s that generate significant amounts of RF. Power consumption is not totaled because all of the subsystems are never on at the same time. In addition to the powers listed, there are short-term switching transients as latching relays are switched and intermittent heater-power for TV cameras and the Ku-band DA, which are external to the temperature-controlled crew compartment.

Table VI. Power Consumption and Weight of Subsystems


UHF-ATC transceiver

Audio distribution

Global positioning





Television (operational)

Ground command interface logic



Quantity of




















Radar 429

Comm 489




*Sum of LRU’s that normally operate simultaneously

**LRU’s only—excludes wiring coax, and C&D

***Excludes headsets, interface units, and cables

****All cameras on




[lb (kg)]

18 (8)

56 (25)***

105 (48)

218 (99)

66 (30)

271 (123)

279 (127)

39 (18)

76 (35)

1128 (513)

Although weight was an important design driver, it lost when traded against several other factors. One design consideration that added weight was the 100-mission design goal, which resulted in most of the LRU’s of the RF being sealed. Studies on weight reduction of communication LRU’s revealed the cost was prohibitive.

Most of the LRU’s of the C&TSS are mounted in equipment bays located at the fore and aft ends of the mid deck along with LRU’s of other orbiter subsystems. In these bays, wires to LRU connectors are in trays in front of the equipment for easy repair and modification. Where feasible, redundant LRU’s are mounted in separate bays for damage control.


1. “Special Issue on Space Shuttle Communications and Tracking,” IEEE Transactions on Communications, Vol. COM-26, No. 11, Nov. 1978.

2. Carrier, L.M., and Pope, W.S., An Overview of the Space Shuttle Orbiter

Communication and Tracking Subsystem, Rockwell International, SD 78-SR-0016,

July 1978.

3. Hoagland, J.C., and Pope, W.S., Space Shuttle Orbiter S-Band Communication

Development Problems, IEEE, International Electrical and Electronic Conference and

Exposition, Toronto, Canada, September 1977.

4. Hoagland, J.C., S-Band Communication on the Space Transportation Systems Space

Shuttle Orbiter, Session 21, Part II, Wescon Professional Program, Los Angeles,

California, September 1978.

5. Lindsey, W.C., Synchronization System on Communication and Control, Prentice

Hall, 1973.

Figure 1. Orbital Communication Links

Figure 2. Subsystem Groupings

Figure 3. S-Band Subsystem Block Diagram

Figure 4. S-Band Frequencies, Modes, and Data Rates

Figure 5. Audio Distribution Subsystem Block Diagram

Figure 6. Operational TV Subsystem Diagram

Figure 7. Ku-Band Radar/Communication Subsystem

Figure 8. Ku-Band Subsystem Block Diagram

Figure 9. EVA/ATC Subsystem Block Diagram

Figure 10. EVA Communication

Figure 11. Payload Frequencies, Modes, and Selected Data Rates

Figure 12. Payload Subsystem Block Diagram

Figure 13. Payload Interrogator RF Channels

Figure 14. NRZ Sequence Keying

Figure 15. Antenna Locations

Figure 16. S-Band Controls

Figure 17. Ku-Band Displays and Controls

Figure 18. Antenna Display

Figure 19. Typical Avionics Installation



William E. Teasdale

National Aeronautics and Space Administration

Lyndon B. Johnson Space Center

Houston, TX 77058

Kwei Tu

Lockheed Electronics Company, Inc.

Houston, TX 77058


The primary objective of the Space Shuttle Program is to provide an economical space transportation system that will support a wide range of scientific, defense, and commercial applications in Earth orbit. The Shuttle will be a manned, reusable space vehicle designed to accommodate these applications. The advent of the Space Shuttle will usher in an era of space industrialization and utilization which undoubtedly will result in new products, new services, and new sources of energy.

The Space Transportation System (STS), consists of several elements, the core of which is the Space Shuttle Orbiter vehicle. The basic elements of the STS include an external fuel tank and expendable solid rocket boosters that are necessary for the Shuttle launch process, and an Inertial Upper Stage (IUS) which is necessary for placement of certain free-flying payloads destined for high Earth orbit or planetary trajectories. The Orbiter vehicle is a true aerospace vehicle in that it will launch like a rocket, maneuver in Earth orbit like a spacecraft, and land like an airplane.

The STS capabilities will result in payload cost savings and operational flexibility. First, the Shuttle will have the inherent ability to retrieve payloads that experience early failure immediately after deployment for subsequent relaunch and use. Second, the large capacity and benign launch environment of the Shuttle Orbiter payload bay will relax the weight, volume, and q-loading constraints imposed on future satellites, allowing design simplification and hence reducing costs. Third, the Shuttle, alone or in combination with a retrieval system or an advanced upper stage, will permit the retrieval of satellites at the end of their service life for refurbishment and reuse. Fourth, the Shuttle can carry replacement subsystem modules to failed satellites on-orbit and extend their life, eliminating the need to recover the entire satellites. Fifth, the Shuttle can be used to perform dedicated space missions, subsystem deployment tests, or technology demonstrations on a space-available basis or in the sortie mode.

The STS will provide numerous supporting subsystems for exclusive accommodation of payloads. These supporting subsystems and accommodations will include payload mechanical attachments/cradles; a Payload Deployment and Retrieval System (PDRS); electrical power; fluid and gas utilities; environmental control; communications, data handling and displays; guidance, navigation, and payload pointing; and certain mission kits which will increase or extend STS mission capabilities, particularly in the area of consumables.

A two-level crew cabin area is provided at the forward end of the Orbiter to accommodate the crew and passengers. The flight crew will control the launch, orbital maneuvering, atmospheric entry, and landing phases of the mission from the upper level forward flight deck. Payload handling will be accomplished by mission specialist crewmen from the aft flight deck cabin area on the upper level just forward of the front payload bay bulkhead.

Seating for passengers and a living area are provided on the lower deck. With these accommodations and a more benign launch/reentry environment, space flight will no longer be limited to intensively trained astronauts who are in perfect physical condition, but will be available to experienced scientists and technicians who are of normal physical condition.


One very important type of STS mission will be the placement of satellites into Earth orbit or into lunar or planetary trajectories. The Shuttle Orbiter will be used to place all deployable payloads and their associated propulsive stages (if any) into low Earth orbit.

Satellites whose missions require attainment of high Earth orbit (such as geosynchronous satellites) or require placement into lunar or planetary trajectories will utilize one of several classes of add-on propulsive stages. The currently planned propulsive stages include the IUS and the spinning solid upper stage (SSUS), sometimes referred to as the payload-assist module. Currently, there are two versions of the IUS planned — one for

Department of Defense (DOD) applications and one for National Aeronautics and Space

Administration (NASA) scientific payload applications, and there are two planned versions of the SSUS.

As many as five deployable payload packages may be delivered on a single mission. Upon reaching the desired orbit, the mission and/or payload specialist will conduct predeployment checks and operations. After determining that a satellite is ready, the crew will operate a payload deployment system, which will either lift the payload (and its propulsive stage) out of the payload bay and release it (as in the case of the PDRS), or eject the payload elements from the mounting cradle by means of a latch release/spring eject mechanism.

The final activation of the satellites can be performed by radio command from the Orbiter or from a ground network. The Orbiter will stand by until the satellite is performing satisfactorily before proceeding with the remainder of the mission.

A satellite launched on a previous mission or which fails soon after deployment can be retrieved and returned to Earth for refurbishment and reuse. To recover a satellite, the

Orbiter will rendezvous with it, deactivate it by radio command, maneuver close, and grapple it with the PDRS arm. The recovered satellite will be lowered into the cargo bay and locked into place. The Orbiter will perform deorbit maneuvers, enter the atmosphere, and land, returning the expensive satellite for reuse.

Placement of free-flying scientific laboratories and data gathering devices into low Earth orbit can be accomplished by the Shuttle. This type of satellite launch will not require the use of propulsive stages, but will require placement and alignment of the subject scientific facility. Examples of free-flying scientific laboratories to be launched by the Shuttle are the

Long Duration Exposure Facility (LDEF) and the Space Telescope.


In addition to the utility for placement of satellites into Earth or planetary orbit, the STS will also be used to transport into space a complete scientific laboratory called Spacelab, which is being developed by the European Space Agency.

The purpose of Spacelab is to provide a ready access to space for a broad spectrum of experiments in many fields and from many nations. Spacelab personnel will be men and women who are experts in their fields, and who are in reasonably good health, thereby requiring only a few weeks of spaceflight training. In the pressurized module configuration, Spacelab will provide facilities for as many as four laboratory specialists to conduct experiments or perform processing functions in such fields as medicine, manufacturing, astronomy, pharmaceutical , and materials. In the five-pallet configuration, the entire STS payload bay will be filled with structural pallets upon which will be mounted numerous scientific experiments or applications hardware elements to be exposed to the Space environment. Control and management of the pallet systems will be exercised from the STS by payload and mission specialists using various Orbiter and Spacelab control and display interfaces provided for the mission. When only pallets are to be flown, essential subsystems requiring environmental control will be carried in an igloo which will provide a pressurized and thermally controlled environment for them.

Spacelab will provide an extension of the experimenter’s ground-based laboratories with the added qualities which only space flight can provide, such as a long-term gravity-free

environment, a location from which Earth can be viewed and examined as an entity, and a place where the celestial sphere can be studied free of atmospheric interference.


As has been described, the types of mission cargoes being considered for the Shuttle range from a full payload bay complement of deployable satellite/booster payload elements to a dedicated sortie mission in which the payload bay is filled with an environmentally controlled laboratory and experiment pallets. The types and quantities of data to be processed and/or throughput by the Orbiter will vary widely depending on the composite payload cargo. The Shuttle may be required to handle up to 50 Mbits/s of high rate experiment data while simultaneously performing systems monitoring and control functions, system checkout functions via an onboard cathode ray tube (CRT) keyboard interface or via one of the uplink/downlink operational data links with ground controllers.

The Space Shuttle Orbiter is being equipped to provide a variety of data-handling services for both attached and detached payloads. Scientific data from attached payload sensors and experiments can be transmitted to Satellite Tracking and Data Network (STDN) or

Satellite Control Facility (SCF) ground stations by the Orbiter S-band communication subsystem, or relayed through a tracking and data relay satellite (TDRS) at S-band or

Ku-band. The Orbiter can also record and store scientific information sent over hard line from attached payloads, or relay text and graphics data sent from the TDRS system ground station (at Ku-band).

Engineering health and status data from both attached and free-flying payloads can be processed, displayed and recorded on-board, or sent to the ground.

Ground-initiated commands for both attached and detached payloads can be transferred through the Orbiter communication system and data processing system (DPS). Commands to both attached and detached payloads can also be initiated by the Orbiter crew via the

Orbiter DPS.

The Orbiter will initialize a payload guidance and navigation system or update its state vector using onboard data or information transmitted from the ground. Guidance data from an attached payload may be monitored and recorded by the Orbiter crew, or processed by the Orbiter DPS for closed-loop pointing, using the orbital maneuvering system (OMS).

Up to five safety-critical status parameters can be hardwired from an attached payload to the Orbiter. The Orbiter crew will monitor these parameters and can take the necessary timely remedial actions. These parameters plus others are also monitored and can be

recorded as part of the Orbiter systems monitoring function. Payload caution and warning

(C&W) data can be transmitted to the ground through the Orbiter.

An onboard processing, display, and transmission capability will provide by the Orbiter as a service to both attached and detached payloads. The Orbiter data processing and software subsystem will furnish the onboard digital computation required to support payload system management. The system management function will be used during prelaunch and orbital phases for payload checkout and status monitoring (passive).

Functions in the computer will be controlled by the crew through main memory loads from the mass memory. Flight-deck stations for payload management and handling will have provisions for data displays (CRT’s) and keyboards for monitoring and controlling payload operations.

The Shuttle Orbiter S-band communications subsystem provides tracking and communications via phase modulated (PM) links direct to ground or through the TDRSS and transmission of data direct to ground via a frequency modulated (FM) link. On the PM link, two digital voice channels, each source encoded with delta modulation at 32 Kbps, are time-division-multiplexed (TDM) with the 128 Kbps telemetry channel to form a 192

Kbps composite TDM bit sequence prior to phase modulating the carrier at frequency

2287.5 MHz or 2217.5 MHz. The telemetry channel can accommodate up to 16 Kbps from a detached payload and up to 64 Kbps from attached payloads. On the FM link, the

Orbiter FM signal processor can accept (1) real-time attached payload data in either analog or digital form (analog from 300 Hz to 4 MHz, digital from 200 bps to 5 Mbps) or (2) payload recorder playback data (from 25.5 to 1024 Kbps) to frequency modulate the link carrier (2250.0 MHz). The Orbiter Ku-band system can operate as a radar during space rendezvous maneuvers or can be used as a two-way communication subsystem, transmitting the attached or detached payload data via the TDRS system. There are a total of three channels that can be accommodated via the Ku-band link. Except for the 192

Kbps channel, which is comprised of Orbiter voice and telemetry as described in the

S-band PM link, the other two channels can accept attached or detached payload data.

From the data rate ranges shown in figure 1, it may be seen that the capability extends continuously from 15 Kbps to 50 Mbps. Similarly, the 4.5 MHz analog channel extends downward to direct current. The unusual signal design provides quadriphase shift keying

(QPSK) of a subcarrier (8.5 MHz), and either QPSK or FM of the carrier (15.0034 GHz).

The Shuttle Orbiter S-band Payload Integrator (PI), Payload Signal Processor (PSP), and

DOD Communication Interface Unit (CIU) will be used for communications with freeflying payloads. For the STS-to-NASA payload link the PSP can accept up to 2 Kbps command data from the Orbiter general purpose computer, modulate command data onto a

16 KHz subcarrier and deliver the resultant signal to the PI. The PI then phase modulates the carrier (2025-2120 MHz) with the command subcarrier prior to transmitting to the

payload. For the NASA payload-to-STS link, the PI can receive and demodulate from the payload, a carrier (2200-2300 MHz) phase modulated by either a 1.024 MHz subcarrier

(standard mode) or a base-band signal with a bandwidth less than 4.5 MHz (bent-pipe data). If the 1.024 MHz subcarrier is transmitted, the payload telemetry data which phaseshift-keys the subcarrier can be 16 Kbps or less (in submultiples of 2). The PSP will demodulate the subcarrier and reconstruct telemetry data and associated clock prior to sending to the Payload Data Interleaver (PDI) for further processing. If the bent-pipe data

(nonstandard mode) is transmitted from the payload, the PI will demodulate the carrier and forward the baseband signal to the Ku-band system for transmission to the ground via

TDRS system (without onboard data processing or display).

On the STS-to-DOD payloads link, the PI will accept ternary frequency shift keying/ amplitude modulation command data from the CIU and phase modulate the command signal onto the carrier (1760-1840 MHz) prior to transmitting to the payloads. For the

DOD payload-to-STS link, the PI can receive and demodulate from the payloads a carrier

(2200-2300 MHz) phase modulated by a 1.024 MHz subcarrier and/or 1.7 MHz subcarrier. The 1.024 MHz subcarrier can be phaseshift-keyed by the payload telemetry data up to 64 Kbps, while the 1.7 MHz subcarrier can be frequency modulated by three payload analog subcarriers. The CIU will then demodulate the 1.024 MHz and/or 1.7 MHz subcarrier and process the demodulated baseband signals.

Figure 1 illustrates overall payload/Orbiter communications system and data rates.


The beginning of frequent scheduled flights by the Space Shuttle, to and from Earth orbit in the 1980’s, will mark the coming of a new age in space. The Shuttle will turn formidable and costly space missions into routine, economical operations generating maximum benefits for people everywhere. Moreover, the Shuttle will open space to men and women of all nations who are reasonably healthy and have important work to do in space.

The Shuttle Orbiter will be capable of carrying a payload or several payloads, and their associated airborne support equipment, totaling up to 65,000 pounds, into Earth orbit. The crew will generally consist of two pilots plus one mission specialist and one payload specialist. The duties of the latter crewman will include checkout and deployment of freeflyer payloads, or the management of experiments to be performed on space laboratory type missions.

Among the multifaceted uses of Space Shuttle during its operational life, which will extend beyond the 1990’s, will be a wide range of applications of the environment of space and of space platforms. STS application will be achieved through operation of satellites, satellites

with propulsion stages, space laboratories, or combinations as appropriate to the specific objectives and requirements. The Shuttle will also provide a laboratory capability to do research and to develop techniques and equipment that may evolve into new products, services, and sources of energy. It is important to note that the Space Shuttle will not be limited to uses that can be forecast today. The reduction in the cost of Earth-orbital operations and new operational techniques will enable new and unforeseen solutions of problems.

The STS will be used to carry into space nearly all civilian and military payloads of the future, including automated scientific space probes and Earth-orbiting solar and astronomical observations. Commercial and agency application payloads will include Earth resources sensing, communications, meteorological, and geodetic satellites.

Current plans are for the Shuttle to conduct up to 60 missions per year with a mission mix of perhaps 40 percent for the DOD and the remainder for commercial and scientific applications. The majority of the flights will be concerned with deploying, retrieving, and servicing satellite payloads.

The current NASA mission model for STS covers space activities in four major areas: (1)

General Science, (2) Applications, (3) Technology, and (4) Space Industrialization. In addition to the basic NASA mission model, there will be missions for non-NASA civil programs such as domestic and extranational space activities relative to synchronous satellites for communications, meteorology, Earth resources, oceanography, and air traffic control. DOD payloads and space activities will be based on a separate DOD mission mode.

The General Science Program includes three major areas of investigation: Physics and

Astronomy, Lunar and Planetary, and Life Sciences. Projected payloads will enable the continued study of the Earth’s space environment, the structure of the universe, and the effects of the space environment on living systems. The Technology Program is designed to exploit the Shuttle capabilities to extend ground-based technology to the space environment. The Applications Program is directed toward exploitation of the benefits of space in such an area as Earth resources, meteorology, and communications. The objective of the Space Industrialization Program is the utilization of space for economically beneficial industrial activities such as the processing of pharmaceutical, electronic, and metallurgical products.

Numerous free-flying paylods for near-term STS missions have been defined and are currently being developed by various commercial, NASA, and DOD organizations. Some of the candidate payloads being planned for early missions are summarized as follows.

Tracking and Data Relay Satellite

The TDRS, which will be launched using Shuttle and a DOD-version IUS, is a highcapacity communication and data relay satellite that will be used to support NASA spacecraft (including the Shuttle Orbiter) in low-to-medium altitude Earth orbit. It is intended that the TDRS system comprised of two operational satellites, one or more onorbit spares, and a ground station located in White Sands, New Mexico, will eventually replace most of the existing STDN ground stations.

Jupiter Orbiter Probe (JOP)

The JOP/Project Galileo spacecraft is to be launched into planetary trajectory via the STS using the IUS. The JOP will eventually be placed in orbit about the planet Jupiter to make remote measurements. Attempts will be made to determine atmospheric structure, elemental and isotopic quantities, and cloud characteristics. The JOP will also be used to refine previous measurements of the Jovian magnetosphere.


The Telesat payload, which will be launched into geosynchronous orbit using the STS, will be an upgraded version of the currently operational Canadian satellite communications system which provides relay facilities for television, radio, voice, data and facsimile for all of Canada. Primary service areas include an east-west trunk between Montreal, Toronto,

Winnipeg, and Vancouver and service to northern remote areas.

Intelsat V

The Intelsat V, fifth-generation Comsat satellite for commercial communications application, will be placed in geosynchronous orbit via the Shuttle and a SSUS-A upper stage. Intelsat V will serve as part of an international commercial point-to-point communications network providing telex, facsimile, telegraph, data, television, and twoway voice transmission worldwide.

Teal Ruby (Space Test Program P80-1)

The Space Test Program P80-1 or Teal Ruby System is a satellite comprised of equipment necessary to support three experiments onorbit for a minimum of one year with a threeyear goal. The experiments include the Teal Ruby experiment, which will provide infrared multispectral Earth background data; the Ion Auxiliary Propulsion System, which will provide onorbit checkout of a millipound thruster for satellite position-keeping; and the

Extreme-Ultraviolet Photometer, which will provide sky and Earth background mapping.

The P80-1 space system is to be inserted in a 160 nmi parking orbit with a 57E inclination by an operational flight of the STS around March 1981. It will then be placed in its

400 nmi operational orbit by an upper stage consisting of two solid rocket motors. It will be operated onorbit for its three-year mission by the Air Force Satellite Control Facility.


The Space Transportation System is designed to provide an economical means of supporting a wide range of scientific, commercial, and defense-oriented space missions and will be used to carry into space nearly all civilian and military payloads of the future.

The STS will reduce the cost of Earth orbital space operations while enabling the use of more than a decade of technology advancement in the area of new products, new services, and new sources of energy.

Figure l.- The overall payload/Orbiter communications system and data rates.




Gordon R. Bolton

European Space Agency (ESA) ESTEC

Noordwijk, The Netherlands


This paper describes the Spacelab command and data management system and its support capabilities for various types of experiments in terms of data processing, display, recording, and multiplexing.


The European Spacelab is one of the several parts of the Space Transportation System.

When carried into space by the Orbiter, it provides features and services for the user community working in space research and space applications. Like the Space Shuttle,

Spacelab is designed to be reflown many times. Its purpose is to extend the Space Shuttle capability in a manner suitable for the user community. With the Space Shuttle, it can be regarded as a short-stay space station which can stay in space for a duration of a nominal 7 and a maximum of 30 days.

Spacelab is designed as a general-purpose laboratory having features and services similar to those provided on the ground. These services include power conditioning and distribution, environmental control, command and data management, and system software.

In order to meet the requirements of users, Spacelab can be assembled from several modular units to meet a particular mission requirement. The overall set of Spacelab configurations is fully described in Reference 2.

When flown in the orbiter, the Spacelab receives support services in the form of electrical power and energy, heat dissipation, and telecommunications.

The overall data path is shown in Figure 1. Data are generated by the payload either accommodated in the module or on the pallets of Spacelab. These data are then acquired

* This paper is an updated version of the paper published in the IEEE Transactions on

Communications, Volume COM 26, No. 11, November 1978 (Reference 1).

by the Spacelab data management system and multiplexed in low-rate housekeeping and high-rate scientific data streams for transmission by the orbiter. This orbiter transmission is controlled by the orbiter crew through the orbiter avionics system. The orbiter can communicate at Ku-band with the Tracking and Data Relay Satellite System (TDRSS). All ground communication with the orbiter/Spacelab pass through the one ground station of the TDRSS. The overall coverage is not 100 percent. Due to Earth geometry and beam blockage by the orbiter and Spacelab payload structure, the coverage in Earth-orbital space is between 50 and 85 percent over a 24-hour period. To bridge mission periods with no downlink capability, a digital recorder is included in Spacelab to record payload data with provisions to interleave its recorded data into the data system when communication with ground is restored.

The TDRSS ground station communicates the payload data (as opposed to the orbiter/Spacelab operating data) to the payload operations control center (POCC), with the rest of the data being communicated to the mission control center (MCC). In the POCC, ground-based scientific personnel will have a direct involvement in the flight operation.

Because of the high data rates, the POCC will also require further communication links to other remote payload centers either by land lines or by satellite links.

The overall telecommunications block diagram is shown in Figure 2 and is taken from

Reference 3. This shows the command and data acquisition paths to and from the user and the ground. It will be seen that commands to the payload/experiment can be generated by the:



• Orbiter avionics and/or crew

• Spacelab avionics and/or crew

• Payload/experiment itself

Similarly, data can be acquired by each of the above. Various data paths exist and are used for different purposes.

The uplink command capability consists of 2 kbits/s from the MCC to the Spacelab avionics via the TDRSS and the orbiter communications and avionics systems.

The downlink is either through the Spacelab/orbiter telemetry interface (which is limited to

64 kbits/s) or the payload high data rate telemetry downlink which can be operated up to a maximum of 50 Mbits/s.

As a service to the user, Spacelab is able to distribute uplink commands throughout the payload area and also issue commands at a specified time or when certain required conditions exist. Similarly, a capability is provided which can acquire data in the low- and high-rate data streams from the payload area. This is accomplished by the Command and

Data Management Subsystem (CDMS) of Spacelab. The overall block diagram of the

CDMS is shown in Figure 3. It will be seen that the CDMS not only provides the above services, but also is able to allow the input, display, and manipulation of data as well.

The on-board units of the CDMS system are given in Table I with their basic functions.

These units are shown in Figure 4.

The CDMS also includes a voice intercommunication assembly and a closed-circuit television extension of the orbiter subsystems, which are described in Reference 2.


The CDMS consists of a data processing assembly (DPA) for low-rate data acquisition and command distribution and a high-rate data assembly (HRDA) for high-rate data acquisition.

The DPA contains experiment and subsystem parts. The subsystem part is for subsystem operation and is independent of the experiment part. This separation is to ease payload integration and to prevent interference among the experiments and the subsystems. Each part of the DPA consists of a computer (3.2 x 10


typical operations per second, 64K

16 bit words core memory), an IOU unit (6 data bus couplers each with a micro-machine) and a digital data bus (1 Mbit/s) routed through Spacelab with standard interface units

(RAU’s). Peripherals shared between both parts of the DPA are three keyboards and displays, and a mass memory unit (8 x 10


words). The DPA’s role is to acquire data and to distribute timing and commands with the data bus rate of 1 Mbit/s and to interface with the orbiter-payload multiplexer/demultiplexers (MDM’s), the PCM master unit (PMU), and the master timing unit (MTU).

The HRDA consists of a high-rate multiplexer, a high-data-rate recorder (HDRR) and a highrate demultiplexer on the ground. The HRDA’s role is to acquire data directly from the users and to time-division multiplex these data into a composite data stream of up to

48 Mbits/s. This data stream is interfaced with the orbiter Ku-band communication system and subsequently transmitted to the ground via the TDRSS. On the ground, the demultiplexer is used to derive the original user inputs to the multiplexer. In the multiplexing operation, some low-speed data acquired by the DPA can be merged into the data stream. The multiplexer also adds digitized voice and timing data to the bit stream as required.

The HDRR has been added to the system to act as a storage buffer to cover periods during which TDRSS transmission is interrupted. This recorder can record 1 to 32 Mbits/s for a period of 20 min to over 10 hr. This storage buffer is intended to be emptied during the next TDRSS transmission period by the recorded data being interleaved and multiplexed with new real-time user data. The user may also record experiment data for on-board storage. In the case where the mission does not demand a rate of 32 Mbits/s and can be accommodated by a 1 Mbit/s rate, the payload recorder of the orbiter is used in place of the HDRR. This also has to be the case in the pallet-only mode since the HDRR can only be flown in the module.

The operational software required to run the DPA and control the HRDA is supplied with the subsystem DPA. This operating system is called SCOS (subsystem computer operating system) and controls all the units described above. Since during launch and entry all the

CDMS are switched off, the CDMS has to be activated/deactivated from the orbiter. This is accomplished by a manual activation/deactivation sequence which allows the subsystem

DPA to be controlled via the orbiter payload MDM’s. When the subsystem DPA is operating, it then has control over the experiment DPA, the HRDA, and other parts of

Spacelab. The operational software for the experiment DPA is called ECOS and contains special functions for payload operation. Both these operating systems provide support to applications programs written in assembler. Above this, SCOS supports application programs written in HAL/S and ECOS supports them in FORTRAN.

As has been explained, the CDMS is divided into two areas for subsystems and experiments so that payload integration is eased and interference is reduced. In the frequency domain, it has also been necessary to design the DPA for low-speed command and data acquisition and the HRDA for high-speed data acquisition. The two assemblies will now be described with regard to data communications, crew/orbiter and payload interfaces, and hardware and software techniques.


The CDMS low-frequency communications involve data transfer within the DPA, between the DPA and the orbiter, and between the DPA and the user. As has already been explained, the DPA is divided into two similar parts, subsystem and experiment; but for simplicity, only the experiment portion will be discussed, since the subsystem portion has no user interface and is concerned with the management of the Spacelab subsystems.

Internal Operation

The internal data communications within the DPA permit the computer to dialog with its peripherals:

• Remote acquisition units (RAU’s)—which handle the user interface

• high rate multiplexer (HRM)—which accepts DPA data for incorporation into the high speed downlink

• Data display systems (DDS’s)—which provide a crew interface with the DPA

• Mass memory unit (MMU)—which contains computer program memory loads and display data

Note that the DDS’s and the MMU are shared with the subsystem DPA. The overall DPA configuration is shown in Figure 5.

Internal data are transferred on redundant half duplex 1 Mbit/s serial data buses which are controlled by the computer and managed by the I/O unit (IOU). Data are transferred over the buses in a Manchester-II format in 16-bit words preceded by a 3-bit time, non-

Manchester synchronization pattern and followed by an odd parity bit.

The peripherals are controlled by instructions transferred via the relevant data bus from an

IOU coupler. Each coupler consists of a discrete component micro-machine having its operation defined by instructions stored in a 512 word read only memory. Connection between the coupler micro-machine and the serial data bus is effected by a bus interface unit (BIU), which converts outgoing messages into the previously described Manchester format and incoming messages to standard TTL logic levels. The BIU is also constructed using a micro-machine, and performs a number of error checks on the incoming data.

This data transmission method is used for communication between the CPU memory and the RAU, HRM, DDS, and MMU.

Orbiter Interface

Three communications paths are provided for data transfer between the DPA and the orbiter (see Figure 5). Serial data buses are again used which are controlled by the orbiter, supported by the IOU. These paths are between the DPA and Orbiter MDM’s, which permits a dialog between the CDMS computers and the orbiter general purpose computers

(GPC’s); the DPA and the orbiter PMU for the transfer of engineering telemetry data via the orbiter to the ground; and with the orbiter MTU and the DPA, which provides time reference for the CDMS and the users.

Payload Interface

The DPA interface to payload hardware is provided by the RAU. Its capabilities, which are under software control, are as follows:

• 64 ON/OFF command outputs. The ON state outputs 20 mA at approximately 5 V.

• 128 flexible inputs arranged in blocks of 16. When acquired as discretes (one to eight blocks may be transferred as one to eight words). When acquired as 8 bit resolution analog inputs, channel pairs may be transferred as single words (single mode) or 1 to 8 blocks of 16 channels may be transferred as 8 to 64 words (scanning mode).

• 4 serial output channels providing 1 to 32 NRZ-formatted words, including associated burst clock.

• 4 serial input channels acquiring 1 to 32 NRZ-formatted words on user request with concurrent IOU command.

• 4 User time clocks (UTC’s) and associated 4 pulse/s UTC resets. The clock is 1024 kHz referenced to the orbiter MTU.

The ON/OFF commands are used to control user hardware, and they may be individually set or reset by asynchronous commands from the DPA. The operating system also provides facilities for pulsing these outputs. For some applications where the DPA has to communicate with an “intelligent experiment” containing a dedicated experiment processor

(DEP), synchronous data transfer on serial input and output channels may be required. In this case a discrete ON/OFF command channel may be used to indicate to the DEP the phase of the periodic input/output loop (PIOL) by the insertion of a pair of triplets at an appropriate position in its tables.

Flexible inputs may be read into the DPA at sample rates from 1 to 100 samples/s by the insertion of appropriate triplets in the PIOL tables. The third word of each triplet executed by the IOU coupler is an instruction that is sent to the RAU’s, but only recognized by the specific unit addressed. This word defines to the RAU the channels to be acquired and mode, analog or discrete. The maximum total acquisition rate for flexible input data is approximately 500 kbits/s.

Serial output channels provide the means for transferring large amounts of data to a user.

These can be used for initial program load of a DEP or the results of the processing by the

DPA of data acquired from the DEP, the orbiter, or another experiment. The transfer may be synchronous or asynchronous and consists of messages of 1 to 32 words. Except for very simple cases, the first word will contain information on the nature and magnitude of the message. The maximum practical data transfer rate for all users is approximately

100 kbits/s.

Serial input channels provide the means for acquiring digital data from users. On user request, one to 32 words may be transferred by the next arriving IOU serial input command. Clearly, synchronization of user data availability with the PIOL acquisition is a problem that can be solved either by “oversampling” or by synchronizing the DEP activities with the PIOL as outlined above. The maximum practical data transfer rate for all users is approximately 50 kbits/s.

The UTC is continuously available from experiment RAU’s whenever the experiment module is ON. It consists of a precision 1024-kHz clock derived from the orbiter MTU and reset pulses occurring at 4/s. After initialization either with GMT from a serial output channel or by an ON/OFF command, the user can maintain absolute or relative time within his experiment for tagging his data as required. In the event of IOU time coupler or timebus failure, the accuracy of the 1024 kHz will be degraded and the UTC reset pulses will cease.

Hardware/Software Techniques

Synchronous and asynchronous communication between the DPA computer and the user via the data bus and RAU are managed by the PIOL. As has been outlined before, the

PIOL consists of command tables assembled at system generation time which define a sequence of RAU input and output activities allocated over 100 10 ms time slots. In each time slot, all but one command is for a synchronous transaction. The remaining (last) command is a dummy, normally skipped, which can be loaded with an asynchronous command specifically generated in response to a user application program request. For input data, the result of this asynchronous transaction is passed to the requesting program, whereas synchronously acquired data arrive in predefined locations in the PIOL table with no specific alert to a data user. Execution of PIOL 10 ms segments is initiated by the CPU in response to the 10 ms interrupts generated in the time coupler by the decoding of the

MTU 100 pulse/s IRIG-B GMT input, and proceeds under RAU coupler control until an

“error” or “end of work” is signaled to the CPU by the general IOU interrupt.

This method of data acquisition is very efficient and has the additional benefit that data obtained by the PIOL may be output via the PMU for downlink simply by having an appropriate overlap of the PIOL table and the telemetry buffer in core. Since the multiplexer also has an interface with the RAU bus and responds to a subset of the RAU instruction repertoire, it is possible to reformat data acquired from RAU’s and output them to the multiplexer simply by the inclusion of suitable triplets in the PIOL. Thus the DPA can operate as low-speed sub-multiplexer with no significant CPU overhead. Output rates to the multiplexer of up to 50 kbits/s are feasible.

The DPA has three identical CIMSA 125 MS computers as part of the assembly. Two of these computers are used in the subsystem and experiment parts of the assembly. The third is used as a back-up, primarily for the subsystem computer. This redundancy is continued through the assembly in the two IOU’s, each IOU having dual redundant functions. Each

IOU can drive either of two data buses that are connected to each RAU of the subsystem or experiment assemblies. The data display systems are also redundantly connected to both the subsystem and experiment IOU. The overall reliability of the system is high (0.95) in the module mode and a little less (0.93) in the pallet-only mode, the weak point being that of the one MMU. The original use of this mass memory was to store all basic subsystem software in the unlikely case of a failure in the initial load of the subsystem computer after arrival in orbit. Since the mass memory can store over 120 complete 64K core loads, it can also be used to support the experiment computer. This has resulted in the MMU tending to be used as an on-line device rather than a back-up device.

The data display system introduces a new dimension into space technology in that the data display unit (DDU) has a tricolor (green, yellow, red), penetration-type, cathode ray tube with a 12-in. diagonal screen.

The user is expected to call up various displays via the keyboard into the DDU screen for experiment evaluation and control. Display formats that are stored within the DPA are supported by SCOS and ECOS as appropriate.

Much has been said about the software of the CDMS. The two operating systems, SCOS and ECOS, will be fully developed with the hardware. Much thought was given as to how the user and operator of Spacelab would develop application programs for the CDMS and then check out and validate hardware/software operation. This process of software checkout was one of the evaluation criteria when the computer was selected. It was required that the on-board computer should have a comparable ground computer which was, in most respects, common (especially in real-time behavior) with the on-board machine, but of course with the added advantage of having a much larger set of peripherals and versatility to produce software.

The ground machine for the 125 MS is the MITRA 125S. These machines have identical instruction sets, operating speeds on instruction and instruction-mix level and input/output architecture. The MITRA 125S has been used in the early stages of hardware/software development in the Spacelab program. It has been shown on the engineering model of

Spacelab the the final insertion of the on-board machine into the DPA can be done with few problems.

By this process of commonality, it is planned that users and complete payloads will be checked out with the use of ground support equipment which will use the MITRA 125S

and that cross-product software on host computers will only be used in the early stage of application software development. By giving the user the chance to accurately simulate the real-time on-board computer with the use of a ground equivalent, the time taken in final integration will be reduced over the case where no such common computer existed.


The CDMS high-frequency communications involve data transfer within HRDA from the user to the orbiter. The performance of the HRDA has to be compatible with the constraints of the Ku-band signal processor (KUSP), the orbiter, and the TDRSS. The combination of the orbiter/ TDRSS communication capability for Spacelab is an ability to handle simultaneously the following:

• One channel in the range 16 kbits/s to 2 Mbits/s (NRZ-L, M, or S) or 16 kbits/s to

1.024 Mbits/s (biphase L, M, or S)

• One channel of either 16 kbits/s to 4 Mbits/s (NRZ-L, M, S) or 2 Mbits/s to 50

Mbits/s (NRZ-L, M, S) or one analog/video channel.

It is the purpose of the HRDA to make this capability available to many users in Spacelab.

The HDRA comprises the following equipment:


• The HRM to time-multiplex the input data and to perform the routing of the composite output data stream to one of the two recorders and/or one of the three KUSP digital inputs.

• The HDRR to store data at rates up to 32 Mbits/s during mission periods without (or degraded) downlink capability.

• The payload recorder (orbiter equipment) to store data rates up to 1 Mbit/s during mission periods without (or degraded) downlink capability where the HDRR is not required or not flown.


• The HRDM to demultiplex the composite data stream to recover on ground the same channels as presented at the HRM inputs on board.

Data Paths and Routing in Multiplexer

The overall data routing capabilities of the multiplexer are shown in Figure 6. In particular, the HRM is capable of performing the following routing configurations:

• Multiplexed experiment data routed to one of the 3 KUSP inputs for real-time transmission

• Multiplexed experiment data recorded on one of the 2 recorders (simultaneously with real time transmission, if required)

• HDRR output routed directly to one KUSP input. Multiplexed data stream switchedoff or routed to another KUSP input

• Same as 3, but functions of HDRR and payload recorder interchanged

• Direct access channel routed to the 2-50 Mbit/s KUSP input. Multiplexed data stream switched-off or routed to another KUSP input or recorded on one of the 2 recorders

• Direct access channel routed to the HDRR and recorded. Multiplexed data transmitted in real time or recorded on payload recorder

The HRM routing modes are commanded by the subsystem computer via the subsystem data bus and the BIU interface dependent on downlink availability, Ku-band signal processor operational mode, and multiplexed data rate.

Multiplexing Concept

The HRM collects serial data from different sources, performs time-division multiplexing based on 16-bit time intervals and, finally, delivers an output of one serial data stream containing all the input data (Reference 4).

The main characteristic of the concept employed is the capability to accept serial data that are completely asynchronous with respect to the HRM internal clock. The decoupling of the input clock from the HRM internal clock is performed by means of 4 x 16-bit input buffers.

The user clocks his data into a 16-bit shift register; then—after 16 bits—the content will be loaded into the 4 x 16 buffer.

In a sequence determined by the format table, the format controller fetches one 16-bit word out of the input buffer and transfers it to the output register, where it will be serialized. In the case of an empty input buffer, a fill word is introduced, which can be identified as such by means of the fill identification as a part of the frame overhead. During demultiplexing on ground, the fill words are automatically suppressed.

High Data Rate Recorder Design

The principal function of the HDRR is to provide for intermediate recording of experimental data in the following cases:

• When the orbiter-TDRSS-ground link is not available

• When the required channel capacity is not available

• When on-board storage of raw experiment data is required

In this case, the tape change capability can be used during flight.

The HDRR and HRM form an integrated system. Both are controlled by the subsystem computer in a coordinated manner. The data display system is the control and monitoring interface with the user. The experiment interfaces with the HDRR are via the HRM only.

During record and reproduce, the HDRR is externally synchronized by the HRM clock except in the direct access mode when the experiment supplies the record clock.

Recording can be done in parallel to the real-time transmission to the ground. In the reproduce mode, the buffered data can be interleaved in the HRM with the real-time data stream.

Data recording for on-board storage without transmission to the ground is only allowed in periods when the HDRR is not used as a buffer device in the ground link.

Operational control of the HDRR is effected via discrete commands from a subsystem

RAU. However, sufficient local controls are provided on the HDRR to allow tape change and to inhibit normal control during this operation. In addition to monitoring the command status and the recorder housekeeping signals, the subsystem RAU also receives a parallel

8-bit word representing the amount of tape used. This information is interpreted by software to represent “tape used” and “remaining tape” for display on the DDU.

The design of the HDRR is based on a technology previously applied for satellite recorders requiring data rates in the 1 to 5 Mbit/s domain. In order to achieve the 1 to 32

Mbits/s record/reproduce range, with variable speed and record times, a low bit error rate, combined with an in-flight tape change capability and very low power consumption, several technological thresholds in spaceborne recorder design had to be pushed forward.

High Rate Demultiplexer Design

The decommutation of the data stream received on ground via the TDRSS link is performed by the HRDM shown in Figure 7. The HRDM input circuit receives a serial

data input from the ground station bit synchronizer. The link is composed of three lines

(data, clock, and bit synchronizer lock status). As long as the lock status is not true, input data are not considered valid. The HRDM is synchronized on the input clock and can operate at any rate up to 50 Mbits/s.

The format generator stores up to 16 formats in programmable, read-only memories plus two formats in random access memories (RAM’s). Each format consists of 768 5-bit words. Each word represents the channel address of the corresponding word in the format.

One frame of the HRDM input data consists of 96 words, so that one format repeats every eight frames. The two RAM’s can be loaded from a ground computer or paper tape reader.

In the BCH decoder and data buffer, the data are buffered line by line and the fill identification word decoded. Each line is then decommutated; fill words are removed; and the detected data are sent to the appropriate output channel buffer.

Orbiter Interface

The # 2 Mbit/s and # 4 Mbit/s links from the HRM to the KUSP are realized with differential drivers/receivers and use a 75 S twisted, shielded pair (TSP). Data are transmitted without clock. The design is state-of-the-art and has posed no technical problems.

The interface requirement for the # 50 Mbit/s link has imposed many technical problems on the deisgn of the HRM.

The link requires the following performance:

• Clock/data phase off-set # 2 ns

• Clock duty cycle (including asymmetry and jitter) in the range 50% ±5% of bit period.

(This is 10 ±1 ns at 50 Mbits/s.)

• Rise and fall times # 3.5 ns

• Signal-to-noise ratio better than 35 dB (200 MHz bandwidth)

• The bit transition density at least 64 transitions in 512 bits with a maximum separation of 64 bits between successive transitions

Payload Interface

The user interface is implemented as a “two-wire” system. The HRM input requires data and clock for each channel. The cable used between the experiment and the multiplexer is a 125 S TPS. The signal level used is 0.5 V peak to peak to 1.5 V peak to peak. The

requirements on the output drivers of the experiments are very critical at rates of

16 Mbits/s to 50 Mbits/s.

The most critical parameters are:

• Data/clock phase offset ±10% of bit period

• Clock duty cycle 50% ±10%/ of bit period

• Data/clock jitter ±10% of bit period

Ground interface

The demultiplexing on the ground is performed by the HRDM. Any user requiring realtime data on the ground will have to interface to the HRDM. The interface is realized with

ECL compatible output drivers.

The flight hardware HRM and ground HRDM, as well as the KUSP/TDRSS combination, are designed as a transparent link; that is, data that are entered into the HRM on-board appear at the output of the HRDM.

Hardware/Software Techniques

The HRM is connected with the Spacelab CDMS subsystem and experiment computers by means of the subsystem and experiment data buses as shown in Figure 8. Data from both computers to the HRM are transferred in blocks of 16 bit words. Control, monitoring, and configuring functions of the HRM are performed by the subsystem computer only.

The data transfer link from the subsystem computer has been introduced to allow any subsystem housekeeping data to be introduced into the HRM composite bit stream for assessment on ground.

The experiment-computer-to-HRM link provides the possibility to use the experiment computer as a pre-multiplexer. Data required via RAU’s are formatted within the computer and sent to the HRM. This link is constrained by the data bus performance and the computer software.

Mechanization of data transfers is realized by the (PIOL) software package. This package acquires and transmits data according to an off-line prepared input/output list. The upper limit for the actual data transfer is on the order of 25 kbits/s.

Data are organized in blocks of up to 32 words of 16 bits. Block lengths and spacing between blocks define the frequency share for this channel in the HRM. The worst case is a channel frequency of 0.8 Mbits/s for “random” data transfers.

The subsystem computer link only provides for a control and monitoring interface.


Spacelab, which has completed development and is now entering the system-level verification phase, represents Europe’s first participation in a major manned space project.

The work is being performed under contract from the European Space Agency

(ESA)/Paris, France, with the Spacelab Project Management Office at the European Space

Research and Technology Centre, ESTEC, Noordwijk, The Netherlands. A multi-national consortium of co-contractors and subcontractors, under the leadership of the prime contractor VFW-Fokker/ERNO-Bremen, Germany, operate the contract. The CDMS is being managed by MATRA Espace-Velizy, France. The various subcontractors to

MATRA for the hardware and the unit development status are given in Reference 1.

The overall contract price for this CDMS is on the order of 40 million dollars. The present contract began in 1974 and is planned to end with the completion of the second Spacelab flight.

Today, the system testing and integration at ERNO is well underway. The electronic systems integration (ESI) model, tests of which have been completed, was assembled from elegant breadboard models of the electrical subsystem hardware items. The engineering model (EM) tests are almost complete and have shown that the planned performance of the various units is being demonstrated.

The flight model (FM) tests have begun, and delivery of the first assembly (long module and pallets) to NASA is planned in 1980. Part of this assembly will be used for the first

Spacelab flight. The second assembly, consisting of the igloo and pallets, will be delivered to NASA for the second Spacelab flight.

Payload requirements have changed as regards to the CMDS since the start of the program. In the beginning, much centralized computation was required by several experiments of each payload. Now, with the advance of electronics in the micro-processor field, users are building dedicated processors within their experiments. The role of the

CDMS is that of an organizing and centralized data management function today rather than a centralized computation tool.

Development in the future is planned, but only after successful demonstration of the complete Space Transportation System performance and as the demand for more Spacelab flights materializes.

If such a user interest is forthcoming, then in the CDMS area, several improvments are envisaged. These improvements are:

• Computer — extension of the main memory

• Mass Memory — replacement of the MMU by a disk-like device

• Display System — additions to the vector/graphics capability

• RAU — simplified interface to experiment processors and equipment

• Multiplexer — improvement of the throughput of the link between the DPA and the

HRM, the buffering and reclocking of the direct channels from the HDRR and the users to reduce error rates

These will be considered under the above constraints in a follow-on development program.

The greatest boost that this program can have is a successful initial demonstration of the

Shuttle, the TDRSS, and the Spacelab programs.


1. Spacelab — Its Command and Data Management Role in the Space Transportation

System. Gordon R. Bolten, Anthony F. Emington, Redi T. Selg, G.A. Weijes. IEEE

Transducers on Communications, Vol. COM-22 No. 11, November 1978.

2. Spacelab Payload Accommodation Handbook (SLP-2104) ESA/NASA publication,

31 July 1978, issue 1, Rev. 1.

3. SPACELAB payload data path from User input (on-board) to user output (laboratory).

Gordon R. Bolton/Kai F. Clausen—Symposium on Computer Techniques for Satellite

Control and Data Processing. B.I.S. Appleton Laboratory, 11/12 Oct. 1977.

4. High rate data handling in Spacelab and on the Ground. Gordon R. Bolton/R.

Selg—Symposium on Computer Techniques for Satellite Control and Data Processing.

B.I.S. Appleton Laboratory, 11/12 Oct. 1977.

Figure 1. Overall Data Path

Figure 2. Overall Telecommunications Block Diagram

Figure 3. Block Diagram of Spacelab CDMS and Related Orbiter Avionics

Figure 4. CDMS Units

Figure 5. Data Processing Assembly

Figure 6. HRM Block Diagram

Figure 7. HRDM Block Diagram

Figure 8. HRM/Computer Links

Table 1. Units of the CDMS and Their Functions


Subsystem and Experiment Computer


Input/Output Unit (IOU)

Experiment & Subsystem Data Bus

Keyboard (kB)

Data Display Unit (DDU)

Mass Memory Unit (MMU)

High Data Rate Recorder (HDRR)

High Rate Multiplexer (HRM)

Remote Acquisition Unit





Arithmetic/Storage/Realtime Control.






2( + 2)



Databus and peripheral controller

Data acquisition and distribution of commands.

Crew commands Man/machine Interface

Display data

Storage of programmes and data.

Digital data storage during interrupted TDRSS communications.

Time division multiplexing of digital serial data.

Peripheral data bus input/output unit.

Data Pattern Sensitivity in Tracking Performance of an AC Coupled

Costas Loop with Hard-Limited In-Phase Channel


Young H. Park

California Institute of Technology

Jet Propulsion Laboratory

4800 Oak Grove Drive

Pasadena, California 91103


This paper is concerned with data pattern sensitivity in carrier tracking performance of an

AC coupled Costas loop with a suppressed BPSK signal. The signal amplitude suppression factor is derived as a function of data “asymmetry ratio” - the ratio of “1”s to the total number of bits in a period of a periodic signal. For an asymmetric pattern, the effect of AC coupling is noticeable whereas there is almost no effect for symmetric squave wave. The tracking performance with an asymmetric pattern is worse than that with a symmetric pattern. However, it is also shown that as expected, the tracking performance of a DC coupled loop with an asymmetric pattern is better than that with a symmetric pattern.


This paper is concerned with the data pattern sensitivity in the carrier tracking performance of a breadboard Costas loop designed for the Multimegabit Telemetry Demodulator/

Detector (MTDD) system at JPL [1]. The Costas loop with a hard-limited in-phase channel is AC coupled in the arm filters. The tracking performance of a DC coupled

Costas loop with a hard-limited in-phase channel has been previously studied [2]. Even though the purpose of the use of a hard-limiter and a chopper type third multiplier is to reduce the unwanted DC offset effect and not to lose the data power around zero frequency [2], the AC coupling in the arm filters is used for practical design view point.

Obviously, the AC coupling will deteriorate the carrier tracking performance for a suppressed carrier BPSK signal with an asymmetric data stream. The data with an asymmetric pattern is defined here as a data stream which has more ‘1’s than ‘0’s or vice versa.

* This paper presents the results of one phase of research carried out at the Jet Propulsion

Laboratory, California Institute of Technology, under contract NAS7-100, sponsored by NASA.

The purpose of this report is to determine the data pattern sensitivity for the AC coupled

Costas loop by calculating the signal suppression factor and the squaring loss. The result is compared to the DC coupled loop performance. The analysis shows that the phase jitter for an asymmetric pattern is slightly larger than that for a symmetric pattern. As expected, the AC coupling does not give much effect for a symmetric pattern.


The block diagram of the Costas loop under consideration is shown in Figure 1. Note that there are AC couplings with RC high-pass filters in front of the single-pole Butterworth filters (Low-pass RC filter). The loop filter is of an imperfect second order type.

The tracking performance of this type of Costas loop with DC coupling has been studied for a symmetric NRZ data case [2]. The previous results can be applied to the present problem with some modifications.

The input signal s(t) is a suppressed carrier BPSK signal:

(1) where P is the total received power, m(t) is a binary data modulation, T o carrier frequency, 2(t) is the random carrier phase, n i noise.

is the angular

(t) is the additive white gaussian

For a while, let’s assume m(t) is any binary data stream with levels ± 1. Let 2(t) = 2 o

+S o t where 2 o n i

is a random phase independent of time and S o

is the frequency offset. As usual,

(t) is assumed the bandpass additive noise process. It can be written as follows [3]:

(2) where N c

(t).and N white gaussian process of single-sided spectral density N bandwidth B i s

(t) are approximately statistically independent, stationary, low pass

/2 less than T o



w/Hz and single-sided

Figure 1. Functional Block Diagram of the Costas Loop

It can be shown easily [2] that the equation of loop operation is: where


(t) = 2(t) -

= the estimated phase of the VCO output signal

K = K


K m

K v

= total loop gain



= the VCO output rms voltage

K m

= the in-phase and quadrature-phase detector gain

K v

= the VCO gain in radians/volt-sec.

= the signal amplitude suppression factor f


(x) = a nonlinearity with period 2 B and unit slope at x = 0 n e

(t,N) = the effective noise process




(t,N) = the self noise

The effective noise process equation can be obtained: where

H(p) = the high pass RC filter transfer function in the Heaviside notation

G(p) = the low pass am filter transfer function p = the Heaviside operator


The self noise is defined by: where the overbar denotes the ensemble average and <·> denotes the time average.

Since the self noise is proportional to sin N, it will be very small in the narrow loop bandwidth.




The signal amplitude suppression factor [2] is defined by:


Now disregarding the self noise term the equation of loop operation under consideration is



We are concerned here with the tracking performance of this Costas loop for a binary data stream with an asymmetric pattern. Since it is understood that the tracking performance depends on the transition probability of the data level, let the data be periodic signal with total number of ‘1’ larger than total number of ‘0’. Then the data can be written as:

(9) m p

(t) = u(t) - 2u(t-Nt o

) + u(t-T) 0<t<T m p

(t+iT) = m p

(t) where u(t) the unit step function

œ i = 0,±1, ±2, ...

t o

the bit width

(N+L)t o

= the period of periodic data pattern / T

N = the total number of ‘1’ bits in the period

L = the total number of ‘0’ bits in the period

Thus the data m(t) is in general:


The suppression factor for the above data pattern is:


From (11) and (5) we have:

Using the relation [2]

(13) we have


The overbar in Eqn (13) is the ensemble average with respect to noise process and

Now we have to derive the filtered output data due to the AC coupling filter and the lowpass arm filter.

The transfer function of the high-pass filter is:


(16) where f



The transfer function of the low-pass filter is:

(17) where f



Figure 2. AC coupled arm filter

Generally R




<< R



1 i.e. f


>> f


With the above assumption, the total transfer function of the AC coupled arm filter becomes approximately the product of H(jT) and G(jT). I

First, let’s examine the step response of the AC coupled arm filter since the data waveform is a linear combination of step functions.

The impulse response of the high-pass filter is as follows:

(19) where u(t) = unit step function


(t) = delta function

The step response of the high-pass filter also can be obtained easily:

The output of the low-pass arm filter for the signal in Eqn. (20) is given by:



where g(t) =

= the impulse response of the low-pass filter

Carrying out the convolution we have:



Assuming the intersymbol interfence lasts for the period T, the output filtered data for the input stream of Eqn. (9) is:

(24) where T = (N+L)t o t o

= 1/symbol rate

Now we can derive the suppression factor using Equations (24) and (14)

(25) where

(26) u(x) = unit step function (27)

Basically Eqns. (25)-(27) are sufficient to define the suppression factor. To make the expression tractable, let’s change the form of equations as a function of Rd, B i

/R s

and where

R d

= the energy per symbol-to-noise density ratio

B i

= the two-sided arm filter noise bandwidth

R s

= the symbol rate = 1/t o


= the pattern asymmetry ratio

The pattern asymmetry ratio is defined by the ratio of number ‘1’ to the total symbol number in the period T:

The two-sided noise bandwidth B i filter if there is no AC coupling:

is related to the 3-dB cutoff frequency of the low-pass

B i

= Bf



Since f




, the equation (29) still is approximately valid for the AC coupled arm filter.

Substituting Eqns. (28) and (29) into (26) and (27) we have:

(30) where c = f






The suppression factor now can be expressed as a function of R d

It is summarized in Table 1.

, B i

/R s

and $ .

The SNR at the input of the arm filter is: where R d

R s

= the energy per symbol-to-noise density ratio

= symbol rate

Table 1.



For 100 kbps data, the filters are built with the parameters:



= 51 S



= 10 µf



= 510 S



= 2451 pf then


The suppression factors for these conditions are summarized in Table 2.


- 4

- 2


+ 2

AC Coupled


= 7/8

- 12.0

- 10.1

- 8.3

- 6.7

DC Coupled


= 7/8

- 9.8

- 8.0

- 6.2

- 4.6

AC Coupled


= 0.5

- 11.7

- 9.9

- 8.1

- 6.4

Data rate = 100 ksps









= 2.5 x 10 -3

* The suppression factors are obtained from Ref. [2].

DC Coupled


= 0.5

- 11.7

- 9.9

- 8.1

- 6.4

DC Coupled


- 10.5

- 8.6

- 6.7

- 5.0

The carrier tracking these jitter is given by:

(34) where



= the single-sided loop bandwidth


= the square loss

To calculate Eqn. (34), we have to derive the effective noise density N e definition of n e

(t) (4) and Eqn. (37), it can be shown

. Using the


(40) where


It is not our intention to carry out the calculation of Eqn. (40). By showing the effective noise density does not critically depend on the data pattern, we can estimate the data pattern sensitivity based on the suppression factor shown in Table 2.





In Eqn. (40), does not depend on the data at all. is the autocorrelation function of the output limiter. Thus for rough analysis, it may assume that the effect of data pattern is small.


The results in Table 2 show that AC coupling in the arm filter deteriorate the carrier tracking performance for an asymmetric pattern data but it does not affect the symmetric data.

With DC coupled Costas loop, the asymmetric pattern gives better carrier tracking performance than a symmetric pattern. An NRZ data gives better performance than a square wave.

For AC coupled Costas loop, the pattern with $ = 7/8 has slightly worse performance than the symmetric pattern ($ = 0.5).


[1] Reasoner, R. et al, “Costas Loop Demodulation of Suppressed Carrier BPSK signals in the DSN Environment”, DSN Progress Report 42-51, JPL, March & April


[2] Simon, M.K., “Tracking performance of Costas Loops with Hard-Limited In-phase

Channel”, IEEE Transactions on Communications, Vol. COM-26, No. 4, April


[3] Lindsey, W.C., Simon, M.K., “Telecommunications Systems Engineering,”

Prentice-Hall, Inc., Englewood Cliffs, N.J., 1973.

[4] Stevens G.L, “A Note on Costas Loop Design,” Jan. 1979.

[5] Park, Y.H., “Progress Report on the TDL Compatibility Test of the MTDD

Breadboard Costas Loop,” IOM 3391-79-021, JPL, April 11, 1979.




Daniel D. Carpenter

TRW Defense and Space Systems

Redondo Beach, California 90278


The acquisition of a communication satellite data signal starts with a spatial search for the incoming direction of arrival. As the antenna scans the received signal level builds up as this direction is approached. This level builds up according to the antenna gain pattern.

Once detected, this level can be used to determine when the pull-in range is reached. This allows the automatic tracking, or the autotrack system to take over and maintain the pointing control. An AGC control voltage can provide an indication of signal level, gain, and thus the pointing error. A non-coherent AGC can be used to provide both a signal presence indication and tracking error signal, if a single channel autotrack signal with AM or AM-PM modulation is used. This includes five horn and dual mode feed systems.


A non-coherent AGC loop is modeled and analyzed for transient signal level and autotrack noise effects. The AGC is designed to track the time varying power build-up during a scan, but not the AM modulation on the autotrack signal. The loop equations are obtained by simply extending the Victor-Brockman-Tausworthe AGC error approximation to the noncoherent case with a SQPSK modulated carrier and AM autotrack modulation. A Gaussian antenna pattern shape is assumed over the acquisition range, resulting in a similar shape for the normalized input SNR D(t). The ideal control voltage output is shown to be proportional to K




R n[1+D(t)] + K


. These constants are functions of loop parameters and

is a function of the input gain drift and noise power. By calibrating the constants out,

R n[1+D(t)] will follow the input D(t) closely for useful ranges of D(t). The effect of closed loop AGC filtering is obtained by using the “exponential shift” formula of operator theory to obtain an asymptotic series in inverse powers of AGC bandwidth. The first two terms of

* For general systems performance, see:

D. D. Carpenter and W. E. Lindsey, “Performance Analysis of Single Channel Autotracking

Systems for Communication Satellites”, National Radio Science Meeting, Boulder, Colorado,

November, 1978.

the series show a rapid approach to the ideal response once the AGC bandwidths exceed

2.5 to 5 Hz for many beamwidths and scanning rates of interest.

The AGC suppression of the autotrack scale factor is derived and agrees with the wellknown result of K s

= D/(1+D), usually obtained from Rice’s square law detector results.

The SQPSK data modulation produced negligible SxS contributions to scale factor variation and tracking error noise properties. Expressions for the tracking error noise power were derived in terms of K s

, AGC parameters, and input SNR.




TRW Systems

One Space Park, Redondo Beach, California 90278

(Paper not received in time for publication.)



J. K. Holmes

Holmes Associates

10636 Kinnard Avenue

Los Angeles, Calif. 90024

K. T. Woo

Aerospace Corporation

2350 E. El Segundo Blvd.

El Segundo, Calif. 90245


A theory is developed which allows one to obtain the optimum asymmetric acquisition search strategy of a PN code despreader when the a priori probability density function is given. The results developed here extend the theory of an optimum symmetric PN code search strategy [1] to the more easily implementable asymmetric search pattern. In the case when the a priori probability density function is Gaussian and for an environment such as the TDRSS (Tracking Data Relay Satellite System), the acquisition time is reduced by about 40% compared to the more standard uniform sweep.


The acquisition circuitry of a despreader (a PN code acquisition and tracking system) is commonly designed so that complete passes are made across the entire code range uncertainty, as shown in Figure 1, during the initial search for the code epoch. The actual search is commonly made in discrete steps one-half a PN code chip apart in time; however, for simplicity in the optimization, we consider “continuous steps” with negligible loss in accuracy. This search, which is commonly implemented by retarding one-half a chip at a time, then integrating and comparing to a threshold (Figure 2), continues until the signal is acquired. This scheme is efficient when the a priori location of the signal in the uncertainty region has a uniform probability density function; however, when the a priori density function is peaked, it is more likely to find the signal in the peaked region than elsewhere, so the full sweep approach may not be the best one.

This paper is concerned with a method that allows one to determine the optimum asymmetric sweep pattern to minimize the acquisition time, while achieving a required probability of signal detection, for a given a priori probability density of the signal location. The calculation is carried out for a Gaussian a priori signal location probability

* Portions of this work were performed at TRW Systems on the TDRSS project, and other portions on IR&D funds of Holmes Associates.

density function as illustrated in Figure 3. The approach is general, however, so that it can be applied to any given a priori signal location probability density function.

The basis of this method relies on the fact that any meaningful statistics (see [2], for example) of acquisition time, which is the time required to search the code until acquisition, depends directly upon the number of chips (code symbols) to be searched.

Therefore, searching where the likelihood to find the signal first reduces the number of positions and therefore time to search.


In this section, we will show how the a posteriori probability density function of the location of the true signal position changes as a function of the number of sweeps across the code phase uncertainty. In Figure 4B, the asymmetric sweep pattern is presented. This scheme, although not symmetric about the midpoint position, is easier to implement than the symmetric scheme (Figure 4A) of reference 1 ([1]) since the retraces do not have to be

“jam-set” to the next sweep’s code phase position, but just turned around.

Consider an asymmetric search centered at the mean of a symmetric, unimodal, a priori probability density function, as shown in Figure 4B (for the case of N = 4 sweeps). Let L





, L




denote the search lengths during the N sweeps (as denoted in Figure 4B for

N = 4), and assume that L


$ L


$ L shall use the notation S



...$ L


. Let p(x) denote the a priori probability density function of the location of the signal. Further, let S i

denote the event that the signal is not detected in anyone of the first i sweeps over regions L


, L




. Furthermore, we

to denote the event that the signal is not detected with zero sweeps, which is, of course, a sure event. It is clear that the conditional probability density of the signal location x, given that no sweep has yet occurred, is equal simply to the a priori density function p(x), i.e.,


This density is sketched in Figure 3 for a Gaussian distribution function, although the theory applies to all symmetric, unimodal density functions. Suppose that no signal is detected during the first sweep over L

1 line segments) and that the event S

1 equal to, by use of Bayes’ rule,







denotes the sum or union of the two

has occurred. The conditional density of p(x *S


) is


In (2), the conditional probability density function p(S


* x) is clearly given by

(3) where P

and denotes the fact that the location of the signal is within the set




is the probability of detection given that the signal is present. The notation

or not in L


, respectively, and P(S


) is the probability of the event S



(4) where P(L




) denotes the probability that the signal location x is with the set L






Substituting (3) and (4) into (2), we thus obtain, after the first sweep,


A sketch of (6) is shown in Figure 5A. Notice that the a posteriori density function is smaller inside the region L that




, but greater outside the region L




- It is easy to show

(7) as, of course, it should. For two sweeps (N = 2), it is easy to show by the same reasoning that the a posteriori density function of the location of the signal is given by


where it will be shown later that

The notation L


- L


denotes the region in L


that does not include L


. A sketch of the a posteriori density function after two sweeps is shown in Figure 5B. Extending the a posteriori density function results to the case of three sweeps leads us to the result


(10) where shortly it will be seen that

Again it can be shown that p(x*S

3 p(x*S

3 posteriori density function p(x*S i

), integrates to one. The a posteriori density function,

), is sketched in Figure 5C. We see that, as the number of sweeps increases, the a

), approaches a uniform distribution.


Probability of Detection After N Sweeps

In this section, we determine the probability of acquisition after N sweeps. Let P sweep, but not in the lst, 2 nd

,... or (i-1)th sweeps. Therefore, Q

N acquiring the signal in N sweeps is given by i

, i =

1,2,3,...N, denote, respectively, the probabilities that the signal is acquired during the ith

, the probability of


First consider the value of P probability of the signal being in the region L




. The probability of acquiring after the first sweep is the

, given that the signal is located in L







times the probability of obtaining a hit

. Hence,


The probability P


is, by definition, the joint probability of acquiring in the second sweep and not acquiring in the first sweep. So we have


For P


, we have


In the same manner, P


and P


are given by (extending Figure 4B in the obvious way)

(16) and


It therefore follows that the probability of detection in one sweep, Q

1 and (13)

, is given, from (12)


The interpretation of Q

Typically Q



is the acquisition probability accumulated after N sweeps.

would be 0.5 or 0.9 in many applications. When Q


is 0.9, it means that the probability of acquisition is 0.9 at the end of the Nth sweep. Now, to find Q



. From (12, (13) and (14), we have


, we add P




Notice that P(S


) = 1 - Q additive, we have


and, in general, P(S


) = 1 - Q


. Since the probabilities are

(20) and


Using (20) and (21) in (19) leads us to


In the same way, it can be shown that Q


and Q


are given by

(23) and


The Q i

’s will be used to obtain the optimum sweep length.


In this section, we specify the optimum lengths {L i

} so that the total search length, given by the sum of all the individaul sweep segments, is minimized.

Define as the time required to complete N sweeps with probability Q factor depends upon the false alarm probability, the dwell time, etc.


. It is assumed that is proportional to the sum of the individual sweep times. The proportionality

Hence, our problem becomes: determine the optimum search lengths L


, L




, L

N+1 our N sweep procedure so that Q



equals the desired acquisition probability and so that


is minimized. The parameter K relates acquisition time to code length search length segments { be given by


}. Our optimization procedure is to use the La Grane multiplier method. Let F

(26) where 8 is the unknown La Grange multiplier. Up to this point, the theory is quite general, the only requirement being that the a priori density be unimodal and symmetric and that

P(L i

) be differentiable.

Since this problem was initially motivated by the need to improve the acquisition time for the TDRSS multiple-access ground receiver [3] ana since the best estimates for the a priori location of the signal were Gaussian, we shall illustrate the theory by assuming that the a priori density function is Gaussian. With the Gaussian assumption, we have


For N = 1, it is easy to show that L


= L




(N = 2). From (12), we have

= L and a solution exists if L is large enough that

is equal to the acquisition probability. A more interesting case occurs for two sweeps

(28) where


From our La Grange function F (26), we have

Letting 8' = 8/F and differentiating with respect to L


gives us



Solving (31) for L




This equation can be written as

(33) where C is the constant

(34) and R i

is the normalized chip uncertainty. In the same manner, we can solve for the optimum value of R


by solving

(35) for R


. We obtain

(36) in the manner the optimum value of R




Substituting (33), (36) and (37) for R i

back into the equation for Q


(29) allows one, in principle, to solve for 8' and therefore C. Unfortunately, the resulting transcendental equation makes it nearly impossible to solve for C analytically. However, the solution can be solved simply on a digital computer by trial and error, choosing values of C so that Q

2 equals the desired value. The actual optimum may occur for values of N > 2. Hence, in general, the solutions must be obtained for all values of N, and the value of N which minimizes the value of corresponds to the true optimum under the constraint of an asymmetric search pattern.

Now consider the solution for N = 3. From (23), (25) and (26), we have

Differentiating F


in respect to L


, L


, L


and L


, respectively, we arrive at





In general, this procedure can be continued for any desired value of N.


The usual strategy for sweeping to obtain acquisition is to start at the end of the uncertainty region where the range delay is minimal, then retard the range in increments of, typically, one-half chips. By sweeping from the minimum delay to the maximum delay, the chances of acquiring a multipath signal are diminished. If the probability of detection, given that the received code and reference code are aligned, is given by P d

and, if the a priori probability density function is Gaussian with zero mean and 6 F = )T, then the cumulative probability of acquisition is as shown in Figure 6. If, for example, a probability of 0.5 is chosen as the desired probability of acquisition, the curve could be read off the

abscissa, and the associated time, denoted by T


, would be the time it takes to acquire with a probability of 0.5 (the median acquisition time)

A measure of the improvement of the optimized scheme over the uniform sweep scheme can be measured as follows. Denote as the time to acquire with a probability of Q using the uniform sweep approach. Next, let denote the time to acquire with the optimized sweep, the improvement factor of the optimized sweep over the uniform sweep is then given by


The acquisition time is then . Clearly, r


$ 1 since unity is achieved with the uniform sweep strategy and therefore the method never increases acquisition time.


In this section, we present some actual optimizations for a few cases of interest. In what follows, we let )T = 6F and neglect the end effects. In Table 1, the case of P


= 0.25 and

Q = 0.5 was specified so that the acquisition time was, in fact, the median time.

Table 1. P


= 0.25, Q = 0.5

As can be seen from Table 1 when P


= 0.25 and Q = 0.5, a reduction to 1/r

Q obtained. In Table 2, the parameters used were P


= 0.6 and Q = 0.9.

= 61.3% was

Table 2. P


= 0.6 and Q = 0.9

As can be seen from Table 2, when Q increases, the improvement factor decreases; in this case, a reduction to only 78% was obtained. A subtle point pertaining to the relationship between r



and Q give a reduction in acquisition time.



and Q is best illustrated in Figure 7 based on the theory given here [4]. A can be seen from the figure, only certain values of P


A general method has been presented that can be used to optimize (minimize) the acquisition time for a PN-type spread spectrum system when the a priori probability density function is not uniform by utilizing an asymmetric sweep.

Specifically, we have calculated for an assumed a priori Gaussian density function that the acquisition time, when the 0.5 probability acquisition time (median) was used as a measure of acquisition time, was reduced by 39% for a cell detection probability of 0.25 and when three sweeps were used. When the acquisition time probability was set to 0.9 instead of

0.5, the reduction was only 22% of the uniform sweep acquisition time.

Although the calculations were for Gaussian a priori density functions, the theory is directly applicable to unimodal, symmetric a priori density functions and P(L i

) is differentiable. Extensions to more general a priori density functions could also be made.



Holmes, J.K., and Woo, K.T., “An Optimum PN Code Search Technique for a

Given A Priori Signal Location Density,” National Telemetry Conference,

Birmingham, Alabama, December 1978.


Holmes, J.K., and Chen, C.C., “Acquisition Time Performance of PN Spread

Spectrum Systems,” IEEE Trans on Communications, Vol. COM-25, No. 8, 1977.


Holmes, M., “Private Communication,” October 1977.


Lieberman, “Optimum Asymmetric PN Code Search for a Gaussian A Priori

Distribution with 2 and 3 Sweeps,” TRW Memo #TDRSS-77-211-292, September



Gumacos, C., “Analysis of an Optimum Sync Search Procedure,” IEEE Trans of

Communications Systems, March 1963, pp 89-99.

Figure 1 Uniform Sweep Strategy.



Figure 3 Gaussian Location of The Signal.

Figure 4A Symmetric Search Pattern

Figure 4B Asymmetric Search Pattern

Figure 5. Aposteriori Density Function After One, Two, and Three Sweeps.

Figure 6. Cumultive Acquisition Time probability versus Acquisition Time, T.




Bernard Sklar

The Aerospace Corporation, El Segundo, California 9OZ45


Because it is analytically straightforward, link budget analysis often takes a back seat in engineering curricula, yet this technique represents one of the most important tools available to communications engineers and managers. This paper presents a tutorial examination of link budget development, with an emphasis on satellite communications systems, and catalogues the typical sources of loss and noise. In addition, it treats the concepts of the range equation, free space, antenna gain and effective area, system temperature, and digital versus analog parameters. This paper also illustrates a typical budget and tradeoffs using a communication satellite example.


Usefulness of the Link Budget

The communications system link budget is a balance sheet of gains and losses. It is comprised of the detailed apportionment of transmission and reception resources, noise sources, and signal attenuators measured from the modulator and transmitter, through the channel, up to and including the receiver and demodulator. The budget is mainly derived from the calculation of received useful power. Some of the budget parameters are statistical with large variances, e. g. , radio frequency (rf) propagation fades due to meteorological events. Link budget analysis is therefore an estimation technique for evaluating communications system performance and, as such, ranks as one of the most important tools available to the system engineer or program manager. Link budgets: a) Are useful for rapidly determining top level resource allocations b) Indicate hardware constraints c) Help to predict system performance, weight, size, and cost d) Allow recognition of the design ground rules and of system design flaws e) Highlight reasonable design tradeoffs f) Illustrate areas of system dependence

g) Help to predict system availability h) Highlight system nuances i) Facilitate changing configurations j) Can serve as the basis for an optimal design search

Where Is the Link?

The link can best be established as that transmission path encompassing everything from the modulator and transmitter, through the channel, up to and including the receiver and demodulator. Figure 1 illustrates a block diagram for a typical satellite system transmitterto-receiver link. The source data is first formatted and processed by a sequence of transformations within the modem (modulator/demodulator). The essential processing step here is modulation. In the case of a digital transmission system, modulation changes the bit stream to an information-bearing waveform prior to transmission. The figure illustrates the reciprocal aspect of the procedure; whatever signal processing steps take place in the transmitting chain must be reversed in the receiving chain. The modem can be thought of as the “brains” of the system


. The transmitter, receiver, and antennas comprise the radio and are considered the “muscle.”

Figure 1. Satellite Communication Xmtr-to-Rcvr Link with

Typical Loss and Noise Sources

The Channel

The propagating medium or electromagnetic path connecting the transmitter and receiver is called the channel. The concept of free space assumes a channel region free of all objects that might affect rf propagation by absorption, reflection, or refraction. It further assumes that the atmosphere in the channel is perfectly uniform and nonabsorbing and that the earth is infinitely far away or its reflection coefficient is negligible. The rf energy arriving at the receiver is assumed to be a function of distance from the transmitter (simply following the inverse square law of optics). In practice, of course, propagation in the atmosphere and near the ground results in refraction, reflection, and absorption, which modify the free space transmission.

A Catalog of Loss and Noise Sources

We shall restrict our discussion to those links distorted by the mechanism of additive white

Gaussian noise (AWGN) only. The time waveform received at the receiver can be stated as r(t) = a(t) s(t) + n(t) where r(t) = received waveform s(t) = transmitted desired waveform a(t) = attenuation (loss) n(t) = interfering waveform (noise)

Additive noise is a very useful model for the overall effects of distortion in a large class of communications systems (e. g., a satellite communication system).

Signal-to-noise ratio (SNR) is a convenient measure of performance at various points in the link. SNR is defined as

The desired waveform can be an information signal, a baseband waveform, or a modulated carrier. Such a communications system primarily degrades in one of two ways: through the attenuation of desired waveform power relative to interfering waveform power, or to the increase of interfering waveform power relative to desired waveform power. These degradations are termed “loss” and “noise” respectively. Losses occur when, by some mechanism, a portion of the signal is diverted, scattered, or reflected from its intended route. Noise occurs when unwanted signal energy is injected into the link, or thermal noise

is generated within the link. Thermal noise results from random motion of a conductor’s free electrons caused by thermal agitation - a dissipative process.

Figure 1 represents a catalog of the most prominent sources of loss and noise in a satellite communications link. The legend utilizes a dot matrix for a loss, a crosshatch for a noise source, and a criss-crosshatch for a mechanism that might involve either (or both) loss or noise. Industry usage of the terms loss and noise frequently confuses the underlying degradation mechanisms of a particular source. For instance, sometimes a loss which is also a noise source has been called a “loss” (e. g. , atmospheric loss, line loss, radome loss). Occasionally, too, a noise source which might also manifest loss has been called

“noise” (e. g. , phase noise, intermodulation noise). Since losses and noise sources are all eventually summed together, it is not essential that a precise distinction between these terms be maintained. The link budget sometimes has a single entry termed “other losses, “ which allows for a large assortment of losses and noise sources.

The space loss accounts for the largest loss of signal power in a typical communications system. In satellite systems the space loss for a C-band (6-Ghz) link to a synchronous satellite is typically 200 dB, where

The key sources of satellite system noise degradation are the noise of the receiver amplifier and feeder line in conjunction with the noise of the receiving antenna.


Link Power Calculation (the Range Equation)

Figure 2 represents an omnidirectional rf source transmitting uniformly over 4 B sterad.

The power density on a hypothetical sphere at a distance R from the source is related to the transmitter power P t


The power extracted with the receiving antenna can be written

Figure 2. Power Density on Hypothetical Sphere at Distance R from Omni Source

where A er

is the absorption cross section of the antenna and is defined by

The receiving antenna’s effective area and physical area are related by the efficiency 0 which accounts for the fact that the total power is not absorbed; some of it is lost through reradiation, scattering, or spillover. Typical values for 0 are 0.55 for a dish antenna and

0.75 for a horn.

A common antenna parameter that measures the power output (or input) relative to the isotropic power is the antenna gain G, where


Antenna gain, unlike that of an electronic amplifier, is the result of concentrating the rf flux in some restricted region less than 4 B sterad; therefore, the received power gain represents the increase in power over that calculated for an ideal omnidirectional antenna, or simply the gain referenced to isotropic. The relationship between gain and effective area is

(2) where 8 is the wavelength of the radiation. Similar expressions apply at the transmitter and receiver antennas, by the reciprocity theorem. (2) For the case in which the transmitting source manifests antenna gain over isotropic, received power is written as

Since the effect of an antenna can be expressed as gain or area, received power can also be expressed in the following ways


Thermal Noise

Noise, the basic system constraint, is assumed to be thermal (this is not always accurate), and its power spectral density is assumed to be flat up through the gigahertz range. AWGN is a good statistical model for thermal noise; it is also a very useful model for the overall effects of distortion in a large class of communications systems (including satellite communication systems). A common model for the thermal noise generator and its coupling into a receiving amplifier is shown in Figure 3.

Figure 3. Electrical Model of Maximum Available Thermal Noise at Amplifier Input

The mean squared noise value is expressed as where: T = temperature, K

B = bandwidth, Hz

R = resistance, ohms 3 k = Boltzmann’s constant (1.38 x 10 -23 Joule/K)

This represents Johnson noise, or thermal noise generated in the source resistor. replace the resistor R with an ideal noiseless resistor R eq

(3) produced by R. Let us also assume that the input impedance of the amplifier is matched to

R eq

Let us

in series with the noise source

and is also an ideal noiseless resistor. Such matching results in the maximum available noise power coupled from the generator into the front end of the amplifier


Back to the Range Equation

In evaluating system performance, the quantity of greatest interest is not the received power P r

, but is the SNR. This is because the basic system constraint is our ability to detect the signal, with an acceptable probability of error (P,), in the presence of noise.

(Since the desired signal here is a modulated carrier waveform, we often speak of the average carrier power-to-noise ratio (P r

/N) instead of the SNR.)

From Eq. 3 we can write

(5) where the effective system temperature T is a function of receiver temperature, transmission line temperature, and antenna temperature (see Section III). In the above equation for P r

/N, we have introduced a term L to represent all degradation factors due to the various losses and noise sources; this term L, which represents all sources not specifically addressed by the other terms of Eq. (5) is written as where L j

is the loss or noise due to the jth degradation factor; such factors have been enumerated in Section I and Figure 1.

A quantity of great interest is the transmitter power required. We can therefore rearrange

Eq. (5) and write it as


The term (4 BR/8) 2 in Eq. (6) is referred to as the path loss or space loss, which cannot be measured since it represents the loss due to the inverse square of the range predicated on a hypothetical isotropic receive antenna.

When Eq. (6) is expressed in decibels, the transmitted signal and noise powers are in decibel-watts (dBW), and all other terms are in decibels

P t

(dBW) = P r

/N (dB) + kTB (dBW) + space loss (dB) + L (dB) -G t

(dB) - G r

(dB) where dBW =

Digital Communication Links

The preceding equations apply to any one-way satellite rf link. In analog systems, noise bandwidth is generally greater than signal bandwidth, and P r

/N is the main parameter for measuring signal detectability and performance quality. In digital receivers, however, matched filters or correlators, where signal bandwidth is taken to be equal to noise bandwidth, are usually used. Rather than consider input noise power, a common formulation for digital links is to replace noise power with noise density, thus eliminating the transmission bandwidth term entirely


Then, combining Eqs. (6) and (7), we can write


For digital systems with AWGN, the key performance parameter is E b



instead of P r


0 where T b

is the bit duration and R b

is the bit rate. As a result

Also, received or available (E b to meet some specified P


Eq. (6) can be expressed as



) r

can be replaced with M (E b



) reqd

, the E

. The link margin M is a safety factor over (E b



) b reqd




; therefore,


Figure 4 illustrates the basic shape of the error probability P



on error probability, imposing a lower bound x

P,# P


for (E b



) reqd

$ x


. The parameter (E b



) b




on (E b

) reqd





versus (E reqd system to another; these differences might be specified error rates or different modulation and coding schemes. The parameter (E reqd

. As shown in the Figure,

reflects the differences from one

also reflects the fact that the signal and noise bandwidths may not be equal for some system configurations.




) reqd

curve for a digital communications system. Normally, the system requirement is for some upper bound

Figure 4. General Shape of the P


Versus E





The system link margin M is expressed by rearranging Eq. (9) as follows where EIRP t

is effective radiated power (with reference to isotropic)

Another way of arriving at the same expression is: M = available E b

E b



(dB), which can be written as



(dB) - required



Using Eqs. (5), (7), and (11), we again obtain Eq. (10).


Amplifier Noise and Line Losses

The noise figure F is defined as

(12) where (SNR) i




0 i i

= SNR at the amplifier input

= SNR at the amplifier output

= signal power at the amplifier input

= noise power at the amplifier input

N ai

= amplifier noise, referred to the amplifier input

G = amplifier gain

Equation (12) can be simplified as


The noise figure of an amplifier tells us how much noisier the amplifier is than its source, but it does not provide us with an absolute measure of the amplifier’s noise. For an ideal noiseless amplifier, F = 1 (or 0 dB); note that noise figure can be made to appear small by simply using a large source resistor when taking measurements. Solving Eq. (13) for N we get ai



The concept of receiver noise temperature equates internal receiver noise to an input resistive source at an effective temperature T


. That is, the receiver is replaced by an ideal noiseless receiver and an input noise source kT


B, such that the receiver output noise is equivalent to that of the original receiver. Since the receiver noise figure relates how much noisier the receiver is than its source, noise temperature can be expressed in terms of noise figure and source temperature T


. The standard IEEE value for T


is 290 K.

From Eq. (4), we can write


(16) where T


= 290 K (room temperature reference) and T


= receiver or amplifier equivalent noise temperature. Substituting Eqs. (15) and (16) into Eq. (14) yields


The same technique can be used to solve for the effective temperature of the lossy line T


(between the antenna and the receiver)

(18) where L, the attenuation of the line, is also the noise figure of the line if it is at room temperature. Note that even small line losses, between the antenna and a low-noise amplifier, can have drastic effects on system performance. (4)

Composite Noise Temperature

In general, a composite noise figure F

1+2 figures F


and F


and gains G


and G


for two tandem amplifiers with individual noise

can be written as

Similarly, a line and amplifier in tandem have a composite noise figure as follows


By analogy, we can write from Eqs. (17) and (18)


Antenna Temperature

The ideal antenna is a nondissipative device. An antenna’s noise temperature is a function of what the antenna “sees” rather than what it is. The source of antenna noise may originate from celestial or atmospheric sources, from the ground or nearby objects, and from manmade sources. For earth coverage satellite antennas, 290 K is generally used as the antenna temperature. For earth-based antennas looking out toward a cooler sky temperature, the antenna temperature is often much less than 290 K. It is important to note that the noise temperature of an antenna is comprised of the noise energy entering through the main beam as well as through the sidelobes.

Sky Temperature

Figure 5 illustrates the composite sky noise temperature due to galactic sources and atmospheric absorption. Below 1 GHz the galaxy noise predominates and above 10 GHz the noise due to oxygen and water vapor absorption predominates; hence, the spectral region from 1 to 10 GHz is associated with an optimum sky noise temperature (termed the microwave or space window). (5) Note that the sky temperature is also a function of elevation angle. An elevation angle of 0 deg corresponds to a longer atmospheric path than an elevation angle of 90 deg; hence 0 deg corresponds to the worst noise temperature within the family of curves in Figure 5.

Figure 5. Sky Noise Temperature

System Effective Temperature

Figure 6 illustrates the calculation of the system effective temperature T s antenna port (the input of the line)

, referenced to the

With the use of Eq. 20 we can write

(if LF were supplied in decibels we would then write)

Figure 6. System Effective Temperature


Link Budget Example for a 4/6-Gigahertz Communication Satellite

Figure 7 shows a plot of gains, losses, and noise over the up- and downlinks for a 4/6-GHz nonregenerative communications satellite repeater (the notation 4/6 GHz means that

4 GHz is the downlink carrier frequency and 6 GHz is the uplink). Nonregenerative means that the uplink signals are not demodulated to baseband, but are simply translated in frequency and retransmitted. The sawtooth-like plot in the figure illustrates the various increases and decreases in signal and noise levels due to amplifier and antenna gains, and also due to losses and noise sources in the link.


Figure 8 is a computer printout of the link budget shown in Figure 7. We shall “walk through” these up and down-links; the numbering of the following comments corresponds to the numbering on Figures 7 and 8:

Figure 7. Gains, Losses, and Noise Over the Up- and Downlinks of a

Communication Satellite System

1. The earth terminal has a 20-W transmitter (13.01 dBW). There are 3.2 dB of transmitter circuit losses, which effectively reduces the power output to 9.8 dBW.

2. The earth terminal 100-ft dish antenna at 6 GHz has a gain of 63.1 dB. The power output from the transmitting antenna is effective radiated power referenced to isotropic (EIRP) and is equal to 72.9 dBW (9.8 dBW + 63.1 dB).

3. The largest loss is the space loss of 200.4 dB, corresponding to a range of 22,500 nmi and a frequency of 6 Ghz.

3A (Figure 8 only) The received incident power (referenced to an isotropic receiving antenna) is attenuated down to -131.5 dBW. This signal is not measurable since its existence is predicated on a hypothetical isotropic antenna.

3B. (Figure 8 only) The satellite is configured with a dish antenna, 0.2 ft in diameter. At

6 GHz, its receiving gain is 9.1 dB.

Figure 8. 4/6-GHz Satellite (Linear) Nonregenerative-Repeater Link Calculations

4. The measurable received signal power (increased over incident power by the satellite antenna gain) is -122.5 dBW. The noise at the satellite is comprised of receiver and line noise (627 K), antenna noise (308 K), and interference from other users (zero for now). The total satellite system noise temperature is 935 K (29.7 dB), and the channel bandwidth is 500 kHz (57.0 dB); therefore, the total input noise power kTB is equal to: -228.6 dB + 29.7 dB + 57.0 dB = -141.9 dBW.

4A. (Figure 8 only) The satellite system G/T (-20.6 dB) is a receiver figure of merit. The numerator is the gain of the receiving antenna, and the denominator is the system effective temperature. Note that G and T need not be individually specified in order to perform the link calculations. G/T is adequate; it is usually specified like this to allow antenna gain versus receiver noise figure tradeoffs.

5. In this example, the satellite downlink maintains a constant EIRP of 11.54 dBW. In


UL nonregenerative repeaters, the uplink carrier power-to-noise ratio (P r

19.4 dB here) dictates how much of the downlink EIRP is signal and how much is noise.

downlink EIRP is then divided accordingly into 11.5 dBW of signal power and -

7.9 dBW of noise power.

(7) The

5A. (Figure 8 only) The satellite antenna is described in item 3B (0.2-ft-diameter dish).

This same antenna is simultaneously employed for transmitting the downlink as well as receiving the uplink. At the 4-GHz downlink frequency, the transmitting antenna gain is 5.6 dB (3.5 dB less than its receiving gain).

6. The downlink, like the uplink, suffers its worst loss as space loss. Here it is 196.9 dB

(different from the uplink because of frequency translation).

7. The carrier level at the antenna aperture (the incident signal power referenced to isotropic) is -191.4 dBW. The incident noise is -210.8 dBW. As mentioned earlier, this is not generally a measurable quantity.

8. The earth terminal antenna is described in item 2 (100-ft-diameter dish). This same antenna is simultaneously employed for receiving the downlink as well as transmitting the uplink. At the 4-GHz downlink frequency, the receiving gain is 59.5 dB; the received signal power is -131.9 dBW. The noise here is comprised of the noise transmitted on the uplink plus the usual downlink considerations of receiver, line, and antenna noise. The earth terminal antenna is looking at a cooler temperature (100 K) than is the satellite receiving antenna; the terminal system temperature is 389 K and the total noise is -144.7 dBW. The carrier-to-noise ratio on the downlink (P

12.8 dB.





8A. (Figure 8 only) Here the earth terminal G/T of 33.6 dB (as compared to -20.6 for the satellite receiver) reflects the fact that the 100-ft earth dish has a much larger gain than the 0.2-ft satellite dish and also the fact that the 389 K effective earth terminal temperature is less than the 935 K temperature for the satellite system.


(Figure 8 only) The P r to the 12.8-dB P r




of 69.8 dB is obtained by adding the 57.0-dB bandwidth

10. (Figure 8 only) The available E b

(50.0 dB) from the available P r subtracting the (E b



) reqd





(19.8 dB) is obtained by subtracting the data rate

69.8 dB). The margin (4.8 dB) is obtained by

of 15.0 dB from the available (E b



) r


Figure 8 is a typical link budget indicating expected performance at various key locations in the system. There is one important compromise in the figure, made for the sake of compactness. The item labelled “other losses” (near items No. 3A and No. 7) implies that the various losses in this category exist just prior to the point of incident power. This is not accurate. If the losses were placed in their actual locations, some of the intermediate numbers would change. The important results (e. g. , SNR and margin) would remain the same.


A Decibel is a Decibel is a Decibel

Figure 8 indicates that, assuming all losses and noise estimates are realistic, we can expect the channel capacity to exceed 100 kbps, with a link margin of 4.8 dB. Another way of describing the 4.8-dB link margin is as follows: The 15.0-dB (E b



) reqd

number comes from curves (similar to Figure 4) dependent on modulation, coding, and other detection parameters. It corresponds to a particular operating point on the curve, e. g. P


= 10


. A margin of 4 8 dB simply says that the system will handle 100 kbps with an error rate of less than 10


(the exact error rate can be found by going back into the P

, system is actually being designed to handle 100 kbps with a P


= 10 -6 .

versus (E curve with a value of 19.8 dB (15.0 dB + 4.8 dB) and reading out the actual P

, b




, e. g. , 10 reqd


For this contrived example then, the 4.8-dB link margin means that, for safety reasons, the


Assuming that this is a desirable design goal, the system engineer can next examine all the gains, losses, and noise that brought about this link margin and consider some tradeoffs.

For example (see Figure 8), he may decide that a 100-ft-diameter dish is too large or costly for the earth terminal. He may, if he has the capability of providing larger transmitter power, consider the following tradeoff: Assume that he can provide as much as 1000 W

(30 dBW) of earth terminal transmitter output power. Such a signal power increase

(17 dB) can be used to reduce the antenna size. This antenna is used twice in the link budget; hence, its gain can be reduced by 8.5 dB in each place. Using Eqs. (1) and (2) with


= 0.55, a decrease of 8.5 dB in antenna gain allows a decrease in dish diameter to 37.6

ft. As far as the overall performance is concerned, it will be the same as before. A transmitter power output decibel is just as good as an antenna gain decibel.

Trading Link Margin with Satellite Receiver Noise Temperature

The question of how much link margin should be designed into the system is frequently asked. The answer is that if all sources of gain, loss, and noise have been rigorously detailed, and if link parameters with large variances (e. g., fades due to weather) match the statistical requirements for link availability, very little margin is needed. For satellite communications at C-band, where the parameters are known and fairly well-behaved, it should be possible to design a system with only 1 dB of link margin. Receive-only television stations operating with 16-ft-diameter dishes at C-band are frequently designed with only a fraction of a decibel of margin. However, telephone communications via satellite using standards of 99.9 percent availability use considerably more margin; some of the Intelsat systems have 4 or 5 dB of margin. Higher frequency designs (e.g., 12/14

GHz) generally call for larger margins since atmospheric losses increase with frequency and are highly variable. It should be noted that a by-product of the attenuation due to

atmospheric loss is greater antenna noise. When extra margin is allowed for weather loss, additional margin should simultaneously be added to compensate for the increase in antenna temperature. With low-noise amplifiers, small weather changes can result in 40 to

50 K increases in antenna temperature.


Returning to the example used in Section IV, suppose the system engineer feels that

4.8 dB of margin is excessive and would like to trade some of it for the ability to procure a satellite receiver with noise temperature greater than 627 K. The system engineer may decide that 3 dB is an adequate link margin. Table I illustrates various link features (e.g.,

P r

/N on the up- and downlinks and margin as a function of receiver noise figure).

Examination of this table reveals that the system performs with a 3-dB margin even when the receiver noise figure (including the line) degrades to 10 dB (2610 K).

Downlink P


/N as a Function of Uplink P


/N in Nonregenerative Satellite Repeaters

For nonregenerative repeaters, the uplink carrier power-to-noise ratio (P large (P r r



dictates how the downlink EIRP will be apportioned between signal power and noise power. For



, the downlink carrier-to-noise ratio (P r



is essentially constrained by the loss and noise sources on the downlink alone (downlink-limited), For small (P

(P r



approximately equals the (P r examining Table I. For all entries in the table, the (P r as the satellite receiver noise figure. The (P r



9.5 dB, the (P r


DL rows of the table show the (P r





(uplink-limited). (7) r

/N) r







starts degrading faster than 0.3 dB per uplink decibel. The last two

following the (P


is large. For (P r r



, the



This can be seen by

degrades at about the same rate every decibel increase in noise figure when the (P

, however only degrades about 0.3 dB for

Table I. Link Features Versus Receiver Noise Figure

* Private communications (August 1979) with W. L. Pritchard, Satellite Systems Engineering Inc.,

Washington, D. C. and S. E. Levine, The Aerospace Corporation.

The link budget, then, can be used to assist the system engineer in designing an operating region to not only meet the margin requirements but also other requirements (e. g., uplink and downlink P r


Capacity as a Function of Number of Accesses

Multiple access pertains to the simultaneous utilization of a link by multiple users. There are three basic multiple access schemes for utilizing the channel resources: frequency division multiple access (FDMA), time division multiple access (TDMA), and code division multiple access (CDMA). For the first two, the multiple users’ signals can generally be considered orthogonal to one another, and capacity per user can be traded off for quantity of users. Another way of saying this is that we can apportion the timefrequency resource, either in the spectral domain or the time domain, any way we like without experiencing performance or capacity degradation.

However, with CDMA, an equitable trade of simultaneous users versus capacity cannot generally be accomplished. In CDMA each link occupies the full channel bandwidth, with signal structures chosen to minimize interference between users. However, mutual interference exists between users because the signal sets are not orthogonal to each other.

System capacity is limited by this mutual interference.

The total noise at the satellite receiver can be expressed as the sum of thermal noise and the interference due to the other users where N th

is thermal noise, It is assumed that each user’s transmission is characterized by the same incident power at the satellite (affected by user protocol); therefore, N each user is the sum of the received power from all other users interf

for where Q is the total number of accesses and P r each user.

is the power received at the satellite from

In the limit, as Q approaches infinity


In the limit, then, as Q approaches infinity, P r

/N approaches the ratio of one user’s signal to Q users’ signals. With equal signal strength per user, this is simply 1/Q.

Table II verifies the behavior of the link according to Eq. (21). This can be seen by comparing columns 2 and 3 for Q $ 16. Table II illustrates link behavior as a function of number of users for a CDM access scheme. The bit rate is 100 bps, a reduction of 30 dB from the single user example in Section IV. For other access schemes, one could expect an equitable trade of data rate versus users. Here, with 64 users, the margin is 1 dB less than our Section IV example, and the data rate is 30 dB less; therefore, the performance has degraded 31 dB. For other access schemes involving no mutual interference we would expect to serve 31 dB (1259) users, but here we are serving only 18 dB (64) users, a loss of 13 dB in system capacity.

Table II. Link Features Versus Number of Users for CDMA,

100-bps Data Rate per User

This type of loss is fundamental to CDMA, and is illustrated in Table II, column 7. The change in system capacity is obtained by summing columns 2 and 6. There is a system capacity loss of 12.5 dB that grows to 13.1 dB with increasing Q.

The advantages of CDMA are operational flexibility for low duty cycle users and potential resistance to narrowband interference and multipath.


The link budget allows the system engineer to make resource allocations, such as the choice of multiple access schemes. Is he willing to pay the price (13 dB in this example) for the improved flexibility? Where shall he attempt to compensate for the lost decibels, assuming they are not present as excess margin?

The Usefulness of a Link Budget

The multiple access example illustrates the use of link budgets in allocating resources and also illustrates areas of system dependence. The multiple access scheme must be chosen to accommodate the required number of users and retain the required capacity. Table II verifies that for a nonregenerative repeater (P r



is downlink-limited until the (P degrades, at which time it becomes uplink-limited.




By scanning the link budget one can methodically search the system design for over- or under-designed areas. Any of the budget parameters are candidates for tradeoff: antenna gain versus noise figure, margin versus data rate, quantity of simultaneous users versus power, coding gain versus atmospheric loss, etc.

The link budget is valuable for highlighting hardware constraints. If a power output stage shows up with an unrealizable wattage or an antenna with unrealistic dimensions, it can be readily seen. The budget allows the recognition of system nuances. Its tabular format provides a checklist for verifying the gains, losses, and noise sources at each important juncture. It also allows one to easily evaluate system performance for edge-of-coverage users or those with other constraints (e. g., users in volatile weather zones or users with a need to remain covert). Performance as a function of terminal location can be evaluated.

For satellite systems, this means the effects of increased range, reduced elevation angle to the satellite, or increased rain.

The link budget allows the system engineer to determine what ground rules were used in detailing the input data. He can tell whether the losses have been methodically detailed or whether they were guessed at with “broad brush” approximations. He can check that the weather loss corresponds to the users’ outage requirements and can decide whether the margin ought to be 0.5 dB, 3 dB, or 6 dB. Once the design has been configured, he can, using weather models, predict system availability as a function of location.

In conjunction with size, weight, and cost models, one can predict the mass properties and cost of a particular communications system design. The link budget facilitates making configuration changes to satisfy changing requirements; it can also serve as the basis for an optimal design search. With the help of a computer, the system engineer enumerates all the interesting and useful tradeoffs. Keeping performance fixed, a parametric search can be undertaken for some minimization (e. g., weight, size, cost, or risk).


The key sources of rf loss and noise have been catalogued. Link budget analysis has been developed, including the concepts of the range equation, free space, antenna gain and effective area, and effective system temperature. The emphasis has been on digital communication systems, particularly with the use of satellites. A typical link was described with a 4/6-GHz communications satellite example. The same example was also used to describe tradeoffs and resource allocation. In short, the paper has addressed the issue of what the system link budget tells the system engineer.


1. Sklar, B., “A Primer on Digital Communication Signal Processing,” presented at

IEEE WESCON/78, Session 15, Los Angeles, 13 September 1978.

2. Collin, R. E., and Zucker, F. J., Antenna Theory, Part 1, McGraw-Hill Book Co.,

New York, 1969, Ch. 4.

3. Nyquist, H. , “Thermal Agitation of Electric Charge in Conductors,” Phys. Rev.

Vol. 32, July 1928, pp. 110-113.

4. Panter, P. F., Communications Systems Design: Line-of-Sight and Tropo Scatter

Systems, McGraw-Hill Book Co., New York, 1972.

5. Hogg, D. C., and Mumford, W.W. , “The Effective Noise Temperature of the Sky,”

The Microwave Journal, March 1960, pp. 80-84.

6. Cuccia, C. L. , “Sensitivity of Microwave Earth Stations for Analog and Digital

Communications, Part I,” The Microwave Journal, January 1969, pp. 47-54.

7. Spilker, J. J. , Jr. , Digital Communications by Satellite, Prentice-Hall Inc. , 1977, pp. 170-177.

8. Lebow, I. L., Jordan, K. L., Jr., and Drouilhet, P.R., Jr., “Satellite Communications to

Mobile Platforms”, Proc. of the IEEE, Vol. 59, No. 2, February 1971, pp. 139-159.



Sergei Udalov


9841 Airport Blvd., Suite 912

Los Angeles, California 90045


Analog voice privacy techniques provide the advantage of being compatible with the

3 kHz audio bandwidth of the existing radio and telephone channels. The degree of privacy provided by an analog voice scrambling technique, however, is proportional to the number of time and frequency elements into which the voice signal can be divided as well as to the number of permutation patterns according to which the elements are scrambled. This implies the requirement for a high degree of signal-processing capability. Microprocessorbased implementations of an analog voice scrambling device provide a large potential for signal processing and scrambling. Furthermore, they provide this potential at a reasonable cost, small volume and moderate power consumption. In addition, a single microprocessorbased analog voice privacy device can be configured in software to yield various degrees of privacy, depending on a particular use and circumstances. Also, a variety of auxiliary functions such as timing, code generation, synchronization and analog-to-digital conversion can be time-shared within the same microprocessor chip, thus minimizing the requirement for support hardware.

The purpose of this paper is twofold: (1) to provide an overview of the existing analog voice privacy techniques, and (2) to specifically outline the capabilities of the microprocessor-based analog voice privacy system design, with a particular emphasis on achieving an analog scrambled signal compatible with the 3 kHz nominal audio bandwidth of the existing radio and telephone channels. Also, workable algorithms used for microprocessor-based analog voice scrambling in frequency as well as in time domain are described. Tape recordings of the voice scrambled and recovered with these algorithms are presented for comparison.


The requirement for voice privacy, once an exclusive prerogative of the military and diplomatic communities, is expanding rapidly into the civilian sector of the population. Not only law enforcement and public service agencies are expressing a strong interest in voice privacy, but the commercial and industrial users are as well. The reasons for this steadily increasing interest in voice privacy are twofold: (1) the ever-growing dependence of the civilian sector on radio and telephone communications, and (2) the concomitant proliferation of the inexpensive and readily available signal intercept equipment [1]. A typical list of the potential civilian sector users of voice privacy may read as follows:

• Law Enforcement Community

• City

• County

• Emergency Services

• Fire

• Ambulance

• Private Sector Users

• Public utilities

• Trucking firms

• Oil companies

• Business radiotelephones

• Fishing fleets

• Private security companies.

The degree of privacy required by a potential user depends on (1) the type of information communicated, and (2) the basic intent of the eavesdropper. Often, the monitoring of, say, emergency services, is simply a matter of idle curiosity. The equipment used is generally rather simple (typically, a standard scanner receiver) and, thus, the damage can result only if the interceptor attempts to interfere with the functioning of the emergency service. The use of a relatively simple voice scrambling technique can eliminate the majority of these eavesdroppers. Realistically, however, many cases of radio eavesdropping involve monitors whose intent is to gain a specific advantage over the communicator. Furthermore, those practicing such purposeful eavesdropping are generally better equipped than the average curiosity-driven listener and, consequently, the communicator must rely upon a rather sophisticated method of voice scrambling.

Having decided that he requires a voice scrambler for his communication, the user has to make the selection of the equipment which will meet his requirements in a most costeffective manner. Quite likely, a user has already made a considerable investment in his communication gear and, therefore, he will be looking for the scrambling equipment which is most compatible with already existing equipment. Another important factor influencing the user’s selection of the voice scrambling method is the bandwidth of the communication channel available to him. The conventional radio and telephone channels, for example,

limit the usable bandwidth to that of 3 kHz, i.e., the audio bandwidth of the human voice signal.

Fortunately for the user, there exists a rather extensive line of voice privacy devices and techniques which are compatible with the nominal 3 kHz bandwidth of a “standard” telephone-like communication channel. The devices which achieve communication privacy within the audio bandwidth are known as analog voice scramblers.

The purpose of this paper, therefore, is to present an overview of the typical analog voice scrambling techniques, with a particular emphasis on the microprocessor-based implementations. It is the microprocessor—a result of large-scale integration (LSI) development efforts of recent years—which brings new powers and capabilities to the field of the analog voice scrambler design. With a microprocessor, the signal-encoding methods hitherto impractical because of implementation difficulties are becoming available to the analog voice privacy users at an affordable cost and within small size packages.


Basically, there are two methods for achieving voice privacy. The first is an all-digital technique. The second method involves analog voice processing and, depending on the degree of privacy required, may also employ digital technology.

With an all-digital voice scrambler, the voice signal is first digitized by the conventional methods such as PCM or )-Mod [2]. The digitized voice is then combined with a digital

(normally, a binary) encryption code and applied to the transmission channel. Within the transmission channel, therefore, the composite signal remains in a digital format which requires a transmission bandwidth compatible with the original voice signal digitizing technique. This implies a transmission bandwidth requirement for nonsynthetic speech much higher than 3 kHz and more likely on the order of tens of kilohertz [3]. Figure 1 shows, qualitatively, the time and frequency domain relationships between the input and output signals of an all-digital scrambler. As can be seen from the figure, the output (i.e., encoded) digital waveform D(t) has no resemblance to input analog waveform A(t). The spectrum of the output waveform is accordingly modified and widened so that f c




is typically 3.2 kHz. The wide bandwidth of the digitized and scrambled voice where f

H signal not only precludes its utilization with conventional transmission channels but also increases the accuracy requirements for the synchronization at the receiving end. With f c being in tens of kilohertz, the synchronization aperture requirement is in tens of microseconds. Therefore, although relatively simple in its implementation and virtually free of the time delays associated with scrambling, the all-digital voice privacy technique is incompatible with the standard 3 kHz-wide audio bandwidth of the existing radio and telephone equipment. It is also incompatible with the single-sideband (SSB) operation.

In comparison, the analog voice scrambling techniques are specifically designed for compatibility with the standard, telephone-like channel. Figure 2 shows the input/output relationships of the time waveforms and the spectra of an analog voice scrambler. Because of the variety of the implementation techniques of the analog voice scramblers, the following general statements can be made with respect to information shown in Figure 2:

• Scrambled waveform is modified for all techniques, yet it remains “speechlike”

• Time delay may or may not be present

• Spectral composition (i.e., frequency ordering) may or may not be altered

• Spectrum width is essentially unaltered

• Synchronization requirement is proportional to the inverse of the audio bandwidth.

It is the latter two features of the analog voice scrambling which make it compatible with the existing telephone and radio equipment as well as with the SSB operation.

The ease of interfacing with the host communication equipment is another salient feature of an analog voice privacy device. A simple interfacing at the audio terminals of the host equipment is all that is needed in most cases to provide a voice privacy capability to the user’s communication gear. Figure 3 shows an example of how a mobile transceiver can be equipped with an analog voice scrambling device. As shown there, the “scrambled” output and input terminals of a voice scramble/descramble device (VSDD) are plugged into the microphone and speaker jacks, respectively, of the host radio transceiver. The microphone and speaker are then plugged into the “clear” input and outputs, respectively, of the

VSDD. Depending on the type of transceiver, either half-duplex or full-duplex scramble operation is possible. A voice scrambler, particularly a microprocessor-based one, can easily accommodate a full-duplex operation because of time-sharing and multiplexing capabilities inherent in its architecture. Furthermore, a “clear mode” override, which is required by many emergency-service-oriented users, can also be safely accommodated by an analog VSDD.

In summary, the advantages of the analog voice scrambling technique over that of an alldigital method, can be summarized as follows:

• Does not require bandwidth expansion (thus, provides for bandwidth conservation)

• Compatible with standard telephone and radio link channels

• Compatible with single-sideband operation

• Affected by channel distortion in the same manner as conventional analog voice

• Can be used with existing analog communication equipment without requiring equipment modification.

These intrinsic advantages of the analog voice privacy techniques, combined with a microprocessor-based implementation, provide the potential user with an immediately realizable voice privacy capability.


Thus far, the discussion has been centered around the advantages of the analog privacy techniques in general, without regard to any specific implementation. Consequently, to provide for better appreciation of how a microprocessor-based implementation can enhance the capabilities of a particular analog voice scrambling technique, a description of several fundamental analog voice privacy techniques is in order.

The techniques used for analog voice scrambling fall into two basic categories: (1) frequency (i.e., spectrum) manipulation, and (2) voice signal manipulation with time [4].

To these two basic categories can be added secondary techniques such as amplitude modification and masking. Because the latter two techniques alter the overall dynamic range requirement for the transmission of the signal, their application must be exercised with care unless the user has full control over the characteristics of the communication channel available to him. On the other hand, from the two basic techniques, i.e., frequency and time scrambling, a large number of systems can be derived with many and varied levels of privacy available to the user [5].

A. Frequency Inversion With and Without Hopping

Frequency inversion is one of the oldest methods used for achieving voice privacy. With this method, the spectrum of the prefiltered audio signal is inverted; in other words, the high frequencies become low frequencies and vice versa. Figure 4 shows the relationship between the spectra of the clear and inverted speech. The implementation of the frequency inversion is rather simple—the prefiltered clear speech signal is suppressed/carrier modulated onto a “carrier” (i.e., a tone), whose frequency is above the voice spectrum and the lower sideband is selected by filtering. This lower sideband constitutes the

“scrambled” voice signal.

For the spectrum inversion illustrated in Figure 4, the frequency of the carrier tone is

3.8 kHz. Thus, the 3.5 kHz signal of original speech becomes a 0.3 kHz signal in the inverted spectrum and, similarly, the 0.3 kHz frequency remaps into 3.5 kHz, with the intermediate frequencies being remapped according to their relationship to the 3.8 kHz carrier tone.

This method of providing voice scrambling provides privacy only against a casual, disinterested listener. The very simplicity of generating the inverted speech makes it

vulnerable to even a relatively unsophisticated “tinkerer” eavesdropper who can use a balanced modulator/filter combination to provide descrambling. In fact, simply passing the inverted spectrum through the device used to generate the inversion in the first place restores the natural frequency order to the scrambled speech. Because of this simple relationship between “encoding” and the “decoding” process of a simple frequency inversion process, this scrambling technique is generally referred to as the “single-code” method. In other words, spectrum is either inverted or not; hence, a single code which, for the scrambler described above, is also a fixed code.

To increase the degree of privacy provided by the spectrum inversion method, frequency hopping can be added to the basic inversion. With frequency hopping, the inverted spectrum shifts in frequency with time, as shown in Figure 5. The frequency displacement ranges from 50-300 Hz and the frequency hop rate is typically in the range of 10-100 Hz.

The increased privacy obtained with frequency hopping occurs because the intelligibility of a voice signal drops with an increased displacement from the normal frequency ordering of the voice spectrum [6]. Consequently, the eavesdropper equipped with a fixed-frequency recovery device will have to cope with a considerably high degree of “garbling” due to frequency hopping. The degree of such intelligibility reduction depends on the rate and degree of frequency displacement contained in the “recovered” hopped spectrum, such as shown in the right-hand portion of Figure 5. In comparison, the authorized listener has a frequency inverter (i.e., descrambler) which hops its carrier tone in synchronization with the hopping pattern of the carrier tone of the transmitting scrambler. Adding this additional degree of privacy to a simple frequency inverter introduces an element of time which, in turn, brings in the requirement for code synchronization between the transmitting and receiving stations. Because of the relatively slow hopping rates involved in the synchronization process, however, the accuracy requirement is not severe, and is on the order of tens of milliseconds.

The simplicity of a voice spectrum inverter and its frequency-hopped version still makes these devices a cost-effective alternative to some of the contemporary users. As shown later in this paper, a microprocessor-based analog scrambler can achieve frequency inversion and spectrum hopping with only a few commands in the assembly language.

B. Bandsplitting with Permutations

The degree of privacy offered by a scrambler operating on the frequency structure of the voice signal can be increased considerably by dividing the spectrum into several subbands, as shown in the left-hand portion of Figure 6. Such frequency segmentation can be accomplished by passing the clear voice through a bank of contiguous bandpass filters.

The outputs of the individual filters can then be transposed in frequency by an appropriate set of mixers and translation oscillators, followed by a bank of filters identical to the one

used ahead of the frequency transposition circuits. In the process of frequency transposition, some of the subbands can also be inverted in frequency, thus adding an additional degree of permutability to the scrambled analog signal. Such a combination of frequency transposition and inversion is shown qualitatively in the right-hand portion of

Figure 6.

As in the case of simple frequency inversion, the bandsplitting followed by transposition and accompanied by occasional subband inversion does not increase the overall bandwidth of the scrambled signal, thus placing the bandsplitting technique into a class of true analog voice privacy devices. The degree of voice privacy provided by the bandsplitting and frequency transposition and inversion is far greater than that provided by simple frequency inversion and hopping. The reason for the higher degree of security is the much higher number of combinations and permutations to which the analog signal can be subjected during the encoding process.

It is important to note, however, that, although it is natural to assume that the larger the number of filters, the better the potential scrambling capability of the bandsplitting device, practical considerations set the number of subbands at five [5] and the typical bandwidth of the individual filters at about 500 Hz. Furthermore, despite the fact that potentially there are 3,820 permutations of five subbands with and without inversion (5!x2

5 ), only 11 are useful for providing good scrambling transpositions [4]. This is due to the well established fact that the voice signal possesses extremely high redundancy in its frequency spectrum.

The logical solution for increasing the degree of security of a bandsplitter/frequency inversion scrambler is to introduce the element of time into the permutation matrix.

Specifically, the frequency transpositions and inversions are changed in time (i.e., “rolled”) according to a pseudorandom pattern determined by the key setting of a particular device.

The period between permutations may range from seconds to milliseconds, with 0.25 to

0.5 seconds being typical.

Because of the introduction of the element of time, i.e., changing the permutations of the transpositions and frequency inversions, the rolling code bandsplitters provide a higher degree of privacy as compared to the static bandsplitters. Thus, it is the added element of time-varying permutation which imparts an additional and significant dimension to the analog voice privacy techniques based on bandsplitting.

C. Time Segment Permutation (TSP)

So far, the scrambling techniques which operate on the frequency characteristics of an analog signal were discussed. The element of time was utilized by these methods solely for the purpose of changing the pattern of frequency permutation. The continuity of voice

signal in time, however, provides for another dimension in which the transposition of digitized speech elements can take place. One of the most common techniques for voice scrambling in time is known as time segment permutation, or TSP [7,8].

Figure 7 shows qualitatively the functioning of a TSP encoding. As shown there, of a clear voice is first loaded into a memory device and then, upon completion of the storage phase, is read out as a sequence of time-permuted segments of the original “clear voice” signal.

The order of the permutation and its rate of change is determined by a preselected pseudorandom sequence which, in turn, may be changing with time. The duration of the stored, unchopped voice signal is typically 250 ms and the duration of the segments is

20-30 ms, which implies that they are shorter than the duration of a syllable [8].

Because the TSP is carried out in the time domain, it is particularly suitable for microprocessor-based implementations. Of specific advantage is the microprocessor’s capability to store in and retrieve from its memory a large amount of digital data. Thus, a microprocessor-based analog voice privacy TSP system converts an analog system signal into a digital format, operates on the digital samples (i.e., permutes them) and reconverts the processed signal back to an analog format. Figure 8 shows the typical memory load and readout sequences of a TSP encoding frame.

The sequence of the frame permutation shown in Figure 8 serves as an additional explanation of the TSP encoding process shown in Figure 7. Specifically, the analog voice is digitized by an analog-to-digital converter, then stored in the random access memory

(RAM) buffer. As shown in Figure 8, the loading of the buffer proceeds in a sequential manner, i.e., the load scan address increases linearly over the memory segments 1-6.

Upon completion of the load, the buffer is then read out in the permuted sequence, whereby segments 1-6 are delivered to the digital-to-analog converter (DAC) in sequence

4, 6, 2, 1, 3 and 5. At the receiving end, the inverse procedure takes place and the segments arriving in a scrambled sequence are rearranged to provide the original clear voice signal. Similar to the limitations inherent in the bandsplitting technique, the TSP has its limitations. Specifically, the limitation of TSP relates to the shortest usable segment length and number of “good” permutations, i.e., the permutations which make the scrambled speech most unintelligible.

The solution to the permutation “quality” problem lies in the use of permutations which have been generated and “screened” by a computer prior to their application in the scrambling process. On the other hand, the limitation on the minimum usable segment length is a more fundamental one because it relates to the transmission characteristics of the voice bandwidth channel [9]. Specifically, as shown in part (a) of Figure 9, a typical audio bandwidth channel has a considerable amount of time delay variation relative to the

time delay at its center frequency. This differential delay, when of the same order as the duration of the shortest speech segment, causes distortion over a significant portion of the speech segment. This distortion results in smeared boundaries between the permuted segments which, in turn, lead to loss of intelligibility of the received signal. Part (b) of

Figure 9 shows this phenomena qualitatively. Consequently, with channel time delay variation being on the order of 3-10 ms, the minimum time segment duration is limited to the aforementioned 20-30 ms duration.

D. Reversed Time Segmentation (RTS)

The problem associated with the limit on the shortness of the time segment duration may be alleviated if the voice signal segments themselves can be made unintelligible. If the segments are unintelligible, their duration can be increased without compromising the privacy of the device. The increased length of the segments, in turn, considerably reduces the distortion associated with the time delay variation of the communication channel.

One of the most efficient methods for destroying the intelligibility of speech segments is to reverse their delivery in time. Figure 10 shows qualitatively the voice signal scrambling pattern obtained with such a reversed time segment encoding method. Part a of the figure shows a portion of a speech signal which is progressing in a normal direction, with its segments being generated in the 1, 2, 3, 4, 5... sequence. Part (b) of the same figure shows how the direction of each of the segments (i.e., time intervals) has been reversed, yet the order of their delivery, i.e., that of outputting to the channel, has not been altered. The typical duration of these reversed segments can be in the 50-400 ms range, depending on the memory capacity of the system and other implementation considerations. Some of the salient characteristics of this technique reported in the literature [10] and confirmed by the author using microprocessor-based algorithms, are as follows:

• The ability to understand the contents of the encoded speech depends on the speech delivery rate

• For very slow delivery rates, it is necessary to increase the duration of the time interval

• A time duration interval of about 150 ms is sufficient to render conventional speech rates unintelligible

• The quality of the recovered voice signal suffers little from the artifacts of the coding/decoding process.

Another important feature of the RTS is that the spectral characteristics of the scrambled signal differ very little from those of the original signal. This makes the RTS particularly attractive from the standpoint of audio channel characteristics compatibility. Furthermore, as shown later in this paper, the implementation of the RTS is relatively easy, particularly

with a microprocessor. The RTS is also compatible with other aforementioned scrambling techniques and, thus, when combined with them, may offer a high degree o privacy.


Because the degree of privacy provided by a voice scrambler device is proportional to the number of permutations of time and frequency segments of the voice signal, a good scrambler generally combines several of the techniques described so far [11]. This implies the requirement for a considerable amount of signal processing. The microprocessor is therefore an ideal device for providing the signal processing necessary for obtaining a good scrambler implementation. With a microprocessor, the signal handling power of a highspeed digital computer is available to the designer of an analog voice privacy device at a reasonable cost, within reasonable size, and an acceptable level of power consumption.

Furthermore, microprocessor-based techniques allow many support functions to be implemented in software, rather than hardware, thus reducing the overall component count.

For example, such functions as timing, code generation [12] and synchronization as well as the analog-to-digital conversion [13] can be time-shared within the same microprocessor chip, thus minimizing the requirement for support hardware. From the development and utilization standpoint, microprocessor-based implementations also offer the following advantages:

Development Phase

• Various techniques can be tried out with relative ease

• Modifications and refinements are easily implemented

Utilization Phase

• Powerful algorithms can be employed to maximize effectiveness

• Multilevel programs can be used within the same device

• Technique updates can be carried out simply by changing the ROM units.

Figure 11 shows a typical microprocessor-based microcomputer system which can form the heart of an analog voice scrambler unit. As shown in the figure, the main component of the system is the microprocessor unit (MPU) or, as it is alternatively referred to, µP. The unit shown is an eight-bit processor with an addressing capability of 16 bits. The latter provides an addressing capability of up to approximately 64 thousands of eight-bit words,i.e., a 64 kbyte capability. The architecture includes a read-only memory unit

(ROM), a random access memory (RAM), and a peripheral input/output controller chip

(PIO). A clock unit is shown external to the MPU. For most modern processors, the

“clock” is actually a frequency-determining element such as a quartz crystal or a combination of R, L and C elements. The program which controls the overall functioning

of the microcomputer resides in ROM, while the dynamically changing data is stored in

RAM. The PIO provides for the control and buffering of the digitized data flowing in and out of the microcomputer.

Figure 12 shows a functional block diagram of an analog voice scrambler which utilizes a microcomputer as its central signal processing unit. The configuration shown provides for a full-duplex operation of the scrambler. Basically, the scrambler shown converts an analog signal to a digital format, operates on the digital samples, then reconverts the processed signal back to an analog format. The duplex operation is made possible by the inherent time-multiplexing capability of the microcomputer.

For timing and code synchronization, a special synchronization receiver and generator are included in the scrambler unit. The synchronization process itself is generally a two-stage process, i.e., there is an initial phase, followed by continuous tracking and synchronization updates. The most commonly used techniques for these two phases [14] and their salient, features are as follows:


• Preamble followed by code information


• Tone supplied along with the signal (subtracted at the receiving end)

• Stable synchronization signal available during transmission (advantage)

• Takes away from the total dynamic range (disadvantage)

• Synchronization information provided during pauses

• Does not use up-channel dynamic range (advantage)

• Requires good clock stability at the receiving end (disadvantage).

With an efficient µP-based design, a major portion of these functions is performed by the microprocessor software, thus increasing the function’s efficiency and flexibility as well as providing for an optimum trade-off between the advantages and disadvantages of the various synchronization techniques.



A. Frequency Inversion Algorithm

The algorithm for frequency inversion of a voice spectrum serves as a good introductory example of a µP-based voice privacy device implementation. This algorithm is based on

the fact that polarity inversion of alternate Nyquist-rate samples taken of a lowpass-filtered signal results in a spectrum whose lower half component is the original lowpass signal with its frequency components inverted [15].

Figure 13 shows the flow chart for the spectrum inversion algorithm. For implementation of this algorithm, it is assumed that the analog signal applied to the analog-to-digital converter (ADC) is lowpass-filtered by either a hardware or software implemented filter, as indicated by the line and microphone input filters in Figure 11. It is also assumed that the output of the digital-to-analog converter (DAC) is also lowpass-filtered, as indicated by the output filters in Figure 11.

As indicated in the flow chart of Figure 13, the system, after cold startup, continues through the initialization and setup of an executive mode. For the case of no frequency hopping, i.e., simple inversion, the executive mode will be minimal and may consist primarily of supporting the lowpass filter implementation if such implementation is incorporated into the software.

The implementation of the frequency inversion algorithm per se starts with loading a

REGISTER with a value of 2. After this load, the processor executes auxiliary functions which may range from simple functions such as setting up a timing loop, if the processor is internally timed, to complex functions such as computation of the next frequency hop increment, if frequency hopping is used.

If the software lowpass filter implementation is not utilized, the next steps are then ADC conversion and data fetch, i.e., the sampling and temporary storing of the ADC output word. Following reading and storing of the ADC output, the contents of the REGISTER are decremented and tested for zero. Failure to pass the REG = 0? test results in the polarity inversion of the word obtained from the ADC. After the polarity inversion, the digital word is then delivered to the analog channel via the system output DAC. The REG

= 0? test, or its previously obtained result, is then used for directing the program to either the inner loop (i.e., REG … 0) or outer loop (i.e., REG = 0).

Consequently, if the first REG = 0? test resulted in a polarity inversion of the signal sample output by the scrambler, the second pass through the inner loop will ensure that the next sample is delivered to the output without inversion. Following this duplet of output samples, the program passes through the second REG = 0? test (or its equivalent) and returns to the SET REG = 2 step which initiates the next pass through this frequency inversion subroutine.

The presence of the “auxiliary functions” subroutines allows for addition of a frequency hopping to the aforementioned algorithms of simple frequency inversion. The exact

implementations of the hopping in software will depend on whether the system is selftimed or interrupt-driven. In either case, sufficient time is generally available for execution of the “auxiliary functions” during the intersample period of the basic spectrum inversion routine. It must also be pointed out that, if the system is indeed frequency-hopped, in addition to the inversion, provisions for the generation (at transmitter) and detecting/tracking (at receiver) of the synchronization signal must be included. In this case, the sharing of the synchronizing functions between the executive and auxiliary portions of the program must be optimized.

B. Time Segment Permutation (TSP) Algorithm

As stated previously, the manipulation of a digitized signal in time is one of the salient features offered by a µP-based implementation of an analog voice scrambler. Figure 14 shows the flow chart for implementing a generic TSP algorithm. As shown in the figure, the cold startup is followed by an initialize phase which then leaves the program in the Run

Executive mode.

Because the TSP is a highly synchronization-dependent program, one of the major functions of the EXEC mode is to generate the synchronization initiate pattern for a transmitter unit and to search for the corresponding synchronization pattern for the case of the receiver unit.

Assuming that the unit is in the RECEIVE mode, as implied by the SYNC? test, the descramble program is not initiated until the synchronization signal is detected. Upon detection and verification of the synchronization signal, the buffer addresses are selected

(assigned) and the buffer load is initiated. If the program is set up to load in a scrambled fashion and to unload linearly, the addresses of the LOAD buffers are selected to provide the loading in a permutated sequence identical to that of the transmission readout sequence shown in Figure 8.

Having selected the randomized load addresses, the program initiates the actual load. For the load on which the ADC conversion is performed, the data is fetched from ADC held in temporary storage and, if necessary, the auxiliary functions are performed. The latter may include, as stated earlier, the random code generation for the next cycle, timing cycle generation and synchronization update and tracking. Upon completion of these auxiliary functions, the stored input signal (digitized) is distributed to an appropriate address in

RAM and the next RAM storage address is computed.

Upon computation of the next RAM address, an END OF FRAME test is performed. If the test is not passed, the program returns to the CONVERSION phase and the fetch/distribute cycle is repeated until all RAM segments have been loaded according to the incoming

pattern. If the test is passed, thus indicating the end of a frame, the program tests for synchronization. If that test is also affirmative, it proceeds to modify the buffer addresses for the next load and current readout. This implies that (1) for the readout, the RAM addresses are sequence-ordered (i.e., linearized) and (2) for load, the addresses are selected to correspond to the next incoming randomized sequence.

This pattern of loads and readouts is repeated until the message is completed and the last

RAM buffer unloaded to the analog output terminal via the DAC. Again it must be emphasized that, because of the synchronization requirements, an optimum “work load” division must be chosen between the executive and auxiliary functions.

C. Reversed Time Segment (RTS) Permutation Algorithm

Figure 15 shows a block diagram which represents a generic system for the reversed time segmentation encoding. As shown in this figure, the system is comprised of an ADC, a block of RAM and a DAC. On the first pass (i.e., scan) through the memory, the RAM is loaded by the output of the ADC. During this pass, no signal readout is performed. On the second pass, however, the RAM is already loaded and the signal readout can commence.

During the second scan, the first load is read out in a time-reversed manner and, immediately upon readout of each memory byte, the read-out address is filled with a new sample of a digitized speech signal. Thus, when the readout of the inverted speech is completed, the RAM is full and ready for the next readout of the time-inverted speech. In this manner, the alternate scans of the RAM produce the required time inversion of the voice signal delivered to the output of the scrambler.

An algorithm for performing an RTS scramble is shown by the flow chart in Figure 16.

The structure of this algorithm is very similar to the one for the TSP, with the exception of the aforementioned timing differences and synchronization requirements. Thus, the executive functions, in conjunction with the auxiliary functions, perform the tasks of initial synchronization detection and subsequent update, respectively. Also, auxiliary functions can be used to provide additional degrees of signal randomization such as frequency inversion over random intervals of inverted speech readout. Other techniques mentioned earlier can also be “mixed in” via the auxiliary functions of the RTS program to provide for a higher degree of privacy.


Analog voice scrambling techniques can satisfy the immediate as well as future needs of a variety of civilian users. The chief advantage of the analog voice privacy techniques is their bandwidth compatibility with the existing radio and communication equipment.

Microprocessor-based implementation of the analog voice privacy, in particular, offers the

user the advantages of a high degree of privacy at a realtively low cost, small size and moderate levels of power consumption. Consequently, microprocessor-based implementations of several analog voice privacy techniques have been carrier out for demonstration purposes. Based on the generated and available tape recording data, the salient features of these implementations are discussed. Furthermore, the examination of a potential effect of new chip development on the existing microprocessor-based analog voice privacy device designs is examined in terms of achieving advanced configurations and improved performance.


1. Middleton, R. G., Scanner Monitor Servicing Guide, Howards W. Sams and Co., Inc.

Indianapolis, Indiana, 1975.

2. Jayant, N. S., “Digital Coding of Speech Waveforms: PCM, DPCM and DM

Quantizers,” Proceedings of IEEE, Vol. 62, No. 5., May 1974.

3. Flanagan, J. L., et al, “Speech Coding,” IEEE Transactions on Communications,

Volume COM-27, No. 4, April 1979, pp 710-737.

4. Kahn, D., The Codebreakers, The Macmillian Company, New York, 1967.

5. McCalmont, A. M., “Communications Security for Voice-Techniques, Systems and

Operations,” Telecommunications, April 1973, pp 35-42.

6. Pappenfus, E. W., Bruene, W. B., and Schoenike, E. Q., Single Sideband Principles and Circuits, McGraw Hill, New York, 1964.

7. Leitich, A., “A Survey of Available Techniques for Voice Privacy,” National

Electronics Conference, 1977, pp 180-185.

8. Hartmann, H. P., “Analog Scrambling vs. Digital Scrambling In Police

Telecommunication Network,” 1978 Carnaham Conference on Crime

Countermeasures, University of Kentucky, Lexington, Kentucky, May 1979, pp 47-51.

9. Baschlin, W., “The Integration of Time Division Speech Scrambling Into Police

Telecommunication Networks,” Proceedings, 1977 International Conference on

Crime Countermeasures, pp 141-145.

10. Belland, E. and Bryg, N., “Speech Signal Privacy System Based on Time

Manipulation,” 1978 Carnaham Conference on Crime Countermeasures, University of

Kentucky, Lexington, Kentucky, May 1978, pp 37-40.

11. McCalmont, A. M., “Achieving and Measuring High Security in Analog Speech

Communications Security Devices,” 1979 Carnaham Conference on Crime

Countermeasures, University of Kentucky, Lexington, Kentucky, May 1979.

12. Vouga, C. A., “Speech Scrambling in Radio Communication,” First International

Electronic Crime Countermeasures Conference, Edinburgh, Scotland, July 1973.

13. Lesea, A. and Zaks, R., Microprocessor Interfacing Techniques, Second Edition,

Sybex, Berkeley, California, 1978, Chapter 5.

14. Goode, G. E., “New Developments in Data and Voice Security,” 1973 IEEE

Electronics Security Systems Seminar Conference Record, pp 83-97.

15. Kak. S. C. and Jayant, N. S., “On Encryption Using Waveform Scrambling,” BSTT

Journal, Vol. 56, No. 5, May-June 1977, pp 781-808.

Figure 1. Input/Output Relationship of the Time Waveform and the Spectrum of an

All-Digital Voice Scrambler

Figure 2. Input/Output Relationship of Time Waveforms and the

Spectra of an Analog Voice Scrambler

Figure 3. Analog Voice Scrambling Device Provides for Simple Interfacing at Audio Terminals of Host Equipment

Figure 4. Frequency Inversion Spectra

Figure 5. Frequency-Hopping and Inversion Spectra

Figure 6. Bandsplitting Combined with Inversion

Figure 7. Time Segment Permutation (TSP)

Figure 8. Various Stages of the Time Segment Permutation Cycle

Figure 9. Effect of Channel Group-Delay Variation on Time Segment Boundaries

Figure 10. Reversed Time Segment Encoding

Figure 11. Typical Microcomputer System Architecture

Figure 12. Functional Block Diagram of a Full Duplex Analog Voice


Figure 13. Flow Chart for

Spectrum Inversion

Figure 14. Flow Chart for Time

Segment Permutation (TSP)

Figure 15. Generic System for Reversed Time Segment Encoding

Figure 16. Flow Chart for Reversed Time Segmentation (ITS)

Convolutional Error Detection on an

Additive White Gaussian Noise Channel

Maurice A. King, Jr.

The Aerospace Corporation

Z350 E. El Segundo Blvd.

El Segundo, California 90250


Concatenated coding schemes involving a convolutional inner code and a block outer code have occasionally been used in communication systems that are very intolerant of errors.

In these schemes the vast majority of channel errors are corrected by the convolutional decoder while the block outer code is used to detect convolutional decoder errors. Block code words containing detected errors are erased.

Soft decision Viterbi convolutional decoders operate by comparing path metrics and selecting the path with the largest metric (the maximum likelihood path). There is a substantial amount of information in the path metrics that is not used in this pick-thelargest decision. It is proposed that some of this information be used in a probabilistic decoding error detection scheme. Such a detection scheme would obviate the use of the block outer code. The result is a bandwidth savings at the cost of some additional processing of the convolutional code metrics.


Concatenated coding schemes involving a convolutional inner code and a block outer code have occasionally been used in communication systems that are very intolerant of errors.

In these schemes the vast majority of channel errors are corrected by the convolutional decoder while the block outer code is used to detect convolutional decoding errors. Block code words containing detected errors are erased.

Soft decision Viterbi convolutional decoders operate by comparing the path metrics that enter each trellis node and selecting as correct the path with the largest metric. There is more information in the value of the path metrics than is used in this pick-the-largest decision. In particular, the likelihood of a decoding error is much greater if the decision is between metrics that are nearly equal than it would be if one metric was by far the largest.

It is proposed that this information be used in a probabilistic decoding error detection

scheme. Such a detection scheme would obviate the use of block outer error detection codes.


Consider a communication system shown schematically in Figure 1.

Figure 1. Communication System

The modulation technique is 8-ary FSK. The information source produces 25 8-ary symbols per second. These symbols are encoded for error control using a rate 1/2 dual-3 convolutional encoder. The resulting 50 character per second (cps) sequence is further encoded with a four chip per character diversity code. The resulting 200 chip per second sequence is transmitted through the channel where it is distorted by additive white

Gaussian noise (AWGN). In the receiver the chips are processed independently. They are detected non-coherently and then combined and decoded by a soft decision Viterbi convolutional decoder. The decoding of the diversity code is handled in the convolutional decoder. The diversity code has the effect of increasing the convolutional code distances by a factor of four while decreasing the effective signal to noise ratio by a factor of four.

Assuming that the system is synchronized and frequency locked, a transmitted chip will have a Non-Central X 2 density at the output of the detector. This density is given by [1]

(1) where


SNR = chip signal energy to channel noise spectral density ratio




= number of chips combined

= K th order modified Bessel’s function.

The seven remaining members of the signal set will have Gamma densities at the detector output, given by

(2) where is the usual Gamma function.


Two figures of merit of a probabilistic error detection scheme are the probability of false message acceptance and the probability of correct message rejection. The first is the probability that a decoded message is in error but the errors are not detected. The second is the probability that errors are declared to be present in an error-free message. In order for an error detection scheme to be useful, both of these probabilities must be acceptably low.

The proposed detection scheme uses a threshold technique. During each trellis node decision, the values of the largest and second largest metrics are compared. If the absolute value of their difference is less than a threshold value, then a decoding error is said to have occurred with high probability. All paths leaving this node would be tagged as containing errors. A decoding error would be declared when all surviving paths contain tagged branches. A correct decoding is declared if at the end of the decoding interval the surviving path contains no tagged branches.

Consider the case of a second path merging with the correct path at any particular node.

The probability that the correct path will be falsely rejected (falsely designated as having errors) is the probability that the metric of the correct path, m c

, will fall in the region

(3) where m


is the metric of the incorrect path, and Th is the threshold value. This probability has the value

(4) where max [· , ·] represents the, largest of the two arguments. The probability of false message acceptance is the probability that an incorrect path is selected when merged with

the correct path, and that the incorrect path is accepted as error free. The probability of this occurring at any particular node is the probability that


This probability has the value


The nodal probabilities of Eqs. (4) and (6) can be used to determine upper bounds on the overall probability of first false rejection or false acceptance, respectively. These bounds are derived using the generating function technique of Viterbi [2]. The generating function for the rate 1/2 dual-3 code has been determined by Odenwalder [3J , to be


The effect of the diversity code is to alter (7) to the form


(9) where C


is the set of coefficients of the terms of the power series expansion of (8).

Following Viterbi, the first time false reject or false accept probability can be upper bounded by substituting in for D


the probability that a path deviating from the correct path in K positions causes a false reject or false accept, respectively. This is done for each



appearing in Eq. (9). The exponent K corresponds to the number of accumulated chips,

N, in Eq. (1) and (2). Thus,


(11) where P


(Th) is the first time false accept probability and P reject probability. The first seven non-zero values of C



(Th) is the first time false

are given in Table I.

Equations (10) and (11) have been evaluated for a set of effective information signal bit energy to noise spectral density ratios and normalized threshold levels. The thresholds were normalized by the channel noise spectral density for ease of computation. The results are presented in Table II.

More important than the first time probabilities as measures of the performance of the technique are the false acceptance and false rejection probabilities for a particular length message. For instance, a 12-bit message will produce 8 encoded 3-bit characters at the channel input. Including two flush characters, there are four merge times during the decoding of this message. Thus, a bound on the probability of false message rejection is


Similarly, a bound on the probability of false message acceptance is


The probabilities of Eqs. (12) and (13) have been computed and the results are presented in Figs. 2 and 3, and Table III.

A 288-bit message (48 6-bit shortened ASC II code words) will involve 96 merges at the decoder. For this case, the probabilities of Eqs. (12) and (13) would be altered to the form

(14) and

These results are presented in Figs. 4 and 5, and Table IV.


A threshold technique has been investigated with regard to its ability to serve as an error detection scheme. The quality of the technique has been measured for the case of a particular system in terms of its probability of false rejection of a correct message, and its probability of false acceptance of an erroneous message. In order for any error detection scheme to be useful, both of these probabilities must be acceptably low.

Numeric results are presented in Figs. 2 - 5 for two message lengths and a particular communication link that includes a rate 1/2 dual-3 convolutional code. As expected, the

probability of false rejection increases with the threshold level, while the probability of false acceptance declines. As was also expected, both probabilities increase with message length, and decline rapidly with increasing signal to noise ratio.


[1] Viterbi, A. J., Principles of Coherent Communication, McGraw Hill, Inc., 1966.

[2] Viterbi, A. J., “Convolutional Codes and Their Performance in Communication

Systems,” IEEE Trans. on Comm. Tech., Vol. COM- 19, No. 5, October 1971.

[3] Odenwalder, J. P. , Error Control Coding Handbook, Linkabit Corporation Report,

15 July 1976.

Table I

Generating Function Coefficients


















Table II. First Time Probabilities

Table III. 12-Bit Message Probabilities

Table IV. 288-Bit Message Probabilities

Figure 2. Signal To Noise Ratio (dB) Figure 3. Signal To Noise Ratio (dB)

Figure 4. Signal To Noise Ratio (dB) Figure 5. Signal To Noise Ratio (dB)



D. H. Martin

The Aerospace Corporation, El Segundo, California 90245


Domestic communication satellite (domsat) systems in the United States have a history of

14 years. Currently, several systems are in operation and another will be in early 1981. In recent years, many papers have been published, each describing certain details of a specific system. In contrast, this paper presents an overview and comparison of all the systems. As a background to this survey, the U. S. domsat history is briefly reviewed. The system overview then begins with a look at the satellites. Their basic designs are compared pictorially and through tabular data. Communication subsystems are also compared. The survey then goes on to the terminals, the terrestrial parts of the systems. Representative terminal characteristics are discussed. Finally, the various communication services offered by these systems are described.


In 1965, the American Broadcasting Corporation (ABC) applied to the Federal

Communications Commission (FCC) for permission to build and operate a domestic satellite system for the distribution of network TV programs. Lacking guidelines by which it could evaluate the application, the FCC returned it early the next year. At the same time, the FCC opened hearings and invited comments on the general subject of domestic satellite communications. In 1967, the legislative and executive branches of the government also began studies and hearings on this subject. Numerous comments and proposals were put forth during these hearings from a broad spectrum of sources. These included the TV networks, common carriers, aerospace manufacturers, government agencies, and others.

In December 1968, a committee appointed by President Johnson published its final report

[1]. It recommended a single pilot program because of uncertainties about the technology.

However, a new administration took office the next month, and a new study was initiated.

As a result, in early 1970, the administration recommended open competition [2]. Thus, in

March 1970, the FCC invited applications from all interested parties. One year later, eight applications had been submitted [3]. The issues involved with these applications produced more hearings. The final FCC decision was that any qualified applicant would be allowed

to develop a system, in some instances subject to specific restrictions. This occurred in

December 1972, over seven years after the original proposal by ABC.

In 1972, Western Union Telegraph Co. (WU), one of the March 1971 applicants, had contracted, at its own risk, for the development of three satellites. The WU system was approved by the FCC in 1973, as were several others. In addition, RCA received permission to operate on an interim basis by leasing satellite capacity from Telesat

Canada. Thus, in December 1973, RCA was the first company to provide domsat services in the U. S. Table 1 summarizes the main historical domsat events.


Western Union [4, 5] began domestic services with the launch of two of its Westar sattellites in 1974. American Satellite Corp. (ASC), a Fairchild subsidiary, began operations that same year [6]. RCA [7], through its RCA Americom subsidiary, transferred to its own satellites in 1976. AT&T also began operations in 1976 [8].

Table 2 gives some basic information about these four systems, plus that of Satellite

Business Systems (SBS). SBS [9] is a joint venture of Comsat General, IBM, and Aetna

Insurance. It received FCC approval in 1977, and expects to have two satellites in orbit by early 1981. Of these five system operators, only three (WU, RCA, and SBS) own their own satellites. ASC leases satellite capacity from WU. AT&T leases complete satellites and satellite control services from Comsat General. GTE participates in the AT&T system by using its own terminals and leasing satellite capacity from AT&T.


Unlike in other countries, all U. S. domsat systems are operated by commercial entities.

They are in the domsat business for the purpose of profit. This economic motivation has existed throughout the domsat history; ABC’s desire to develop a system was to reduce distribution costs. The potential for profit lies in the fact that satellite systems can do certain things better or easier than terrestrial systems. In addition, satellite systems can provide services for Alaska and Hawaii, which are not practical by terrestrial methods.

One of the technical benefits of satellite systems is, that only one repeater (the satellite) exists in any link, aside from the relatively short terrestrial tails between the terminals and users. In contrast, a long distance terrestrial link will have repeaters every twenty to forty miles, each of which contributes a small amount to signal degradation. The other primary advantage is that satellites view the entire country; whereas terrestrial links are constrained to specific point to point connections. Thus, satellite system distribution costs are

insensitive to distance, can interconnect any number of locations, and often can provide quicker and/or less costly extensions of service, especially in rural areas.

AT&T has a near monopoly on long distance terrestrial communications. Because domsat systems are competing in many cases for the same business, the FCC put a major restriction on AT&T. Specifically, for the first three years of operation, AT&T was only allowed to use satellites within its switched public and U. S. government telephone networks. The purpose of this restriction was to promote competition by giving other operators (who are allowed to provide all types of communication services) an opportunity to establish themselves.


Table 3 gives the basic characteristics of the satellites used by the domsat systems. In order of their initial launch dates, they are: Westar (W. U.), Satcom (RCA), Comstar

(AT&T), and SBS. All of them are, or will be, in synchronous equatorial orbit between

83E and 135E W. longitude. (Table 4 gives specific launch dates and current orbital locations). All of them have a view of the 48 states, but only those in the western part of the orbital arc can view Hawaii and all of Alaska.

The Westar satellite [10, 11], Figure 1, is virtually identical to the Anik A satellites of

Telesat Canada. The significant changes were modification of the antenna pattern and addition of a back-up despin subsystem. Westar is the smallest of the four satellites, and is spin stabilized with only the antenna and feed horns despun.

The RICA Satcom satellite [12, 13], Figure 2, is the only three-axis stabilized design. The antennas are fixed on the earth viewing face of the body, while the solar arrays are deployed in orbit and rotate to track the sun. The Satcom design was coupled with the

Delta launch vehicle model 3914 development to produce a weight optimized, cost effective satellite. The optimization is indicated by comparison with the Comstar satellite, which has very similar communications characteristics, yet weights 80% more and requires a more expensive launch vehicle (Atlas-Centaur).

The Comstar satellite [14, 15], Figure 3, is a spin stabilized design derived from the

Intelsat IV satellite. The despun portion includes the entire communication subsystem and the antennas. The support subsystems, such as electrical power and attitude control, are within the spinning portion of the satellite.

The SBS satellite [16, 17], Figure 4, is a spin stabilized design having much in common with the Anik C satellite which is also in development. Like Comstar, the SBS communication subsystem and antennas are despun, while the remainder of the satellite

equipment is contained in the spinning portion. Unlike Comstar, both the antenna and a portion of the solar array are deployed after the satellite arrives in synchronous orbit. The reason for this is to attain a compact launch configuration in order to minimize launch costs on NASA’s Shuttle.

Besides these satellites, Western Union is developing an Advanced Westar [18, 19] space segment, which is integrated with the TDRSS payload on the TDRSS spacecraft that W.U.

and TRW are developing for NASA. Advanced Westar should be in use beginning in


Within Table 3, the variance in weight among the satellites is due both to the size of the communications payload and the general design of the satellite. The power differences are discussed below. Attitude control, stationkeeping, and design life values are similar for all four satellites and reflect the status of satellite technology in general. Some improvement is noticeable with SBS, whose design dates 5 to 8 years later than the other satellites.


The basic characteristics of the communication subsystems of the four satellites are indicated in Table 5. Three operate in the 4 and 6 GHz bands which have been used for satellite communications since the early 1960s. The fourth, SBS, will be one of the first operational (i. e., nonexperimental) satellites to use the 12 and 14 GHz bands. Of the three,

Westar has twelve transponders and RCA and AT&T have twenty-four. The double capacity is achieved by independent transmissions on orthogonal polarizations on both uplinks and downlinks. The other satellites use one polarization for the uplinks and the other for the downlinks.

The performance of the first three satellites is about equal because the technology used in their designs is similar. The SBS performance is somewhat higher due to different technology and the fact that coverage is limited to the 48 states. The end of life satellite power (Table 3) which varies from 260 to 830 W is directly related to the TWT power. In each case, the satellite power is 4.1 to 4.4 times the total TWT power for all channels of the subsystem.

The capacity of each channel depends on the ground terminal performance. AT&T uses the largest terminals (~ 100 ft. antenna diameter) and can transmit 1500 to 1800 voice circuits through one channel. With smaller antennas (e. g., 17 to 36 ft. diameter), the channel capacity is usually a few hundred to almost 1000 voice circuits.

Figure 5 is a representative communication subsystem block diagram for all four satellites.

The actual subsystem of the Satcom and Comstar satellites is equivalent to twice what is

shown in the figure. The receiver has a 500 MHz bandwidth and amplifies and frequency translates twelve channels simultaneously. Although past receiver designs have used various types of amplifiers, future receivers are likely to use only transistor amplifiers. The

TWTs each amplify only one channel. The input and output multiplexers are combinations of bandpass filters and isolators which separate and then recombine the channels. The separation of the transmitter into odd and even halves simplifies the multiplexers’ design by increasing the spacing between adjacent channels within the multiplexers. Symbolically,

Figure 4 shows a diplexer and one antenna. In actual practice, either one or two reflector antennas are used, each with 3 to 15 feed horns. In some satellites, separate horns are used for transmitting and receiving so that no diplexing is required.


Transmission techniques vary with the type of traffic. Multiplex voice circuits are usually analog FDM/FM/FDMA, although ASC is converting to a digital TDM/QPSK/TDMA format. SBS also plans to use a digital format. TV transmissions use FM with either one signal per satellite transponder or two using FDMA, although RCA has demonstrated transmission of four simultaneous programs on one transponder. Nonmultiplexed voice or data circuits use single channel per carrier (SCPC) with FM or PSK modulation and

FDMA. Digital data uses QPSK modulation, often with error correction coding.


Characteristics of some representative terminals are listed in Table 6. Figures 6 to 7 are pictures of two terminals. Except for differences caused by the higher frequencies of SBS links, the electronics in all of the terminals are similar in nature. However, there are significant variations in the amount of equipment, which depends on the amount of traffic routed through the terminal. Also, at some important locations, there might be more than one antenna in order to maintain simultaneous communications through more than one satellite.

The other large variation among the terminals is the antenna diameter. This varies from

105 feet for the GTE terminals used with the AT&T system to 4.5 m (14.6 feet), which is the smallest size currently authorized by the FCC. The larger antennas require automated tracking of the satellite whose motion, though small, is on the order of the beamwidth. The small antennas, having larger beamwidths, are able to be pointed in a fixed direction. This simplifies the terminal by eliminating the tracking equipment - occasional pointing changes are done manually.

The receiving performance of the terminal is expressed by the gain to noise temperature ratio, G/T. The type and capacity of the links to be received, and the satellite

characteristics, determine the G/T requirement. Various combinations of antenna diameter and preamplifier noise temperature are able to provide a given G/T. The choice depends primarily on cost and also on reliability. High capacity terminals, such as the first and second in Table 6, use very low noise receivers. Smaller terminals can often use less expensive receivers with preamplifiers having a higher noise temperature. At these terminals, traffic and location influence the preamplifier choice and thus cause the G/T variations shown in Table 6. Location is important because the satellites’ ERP is lower for outlying areas (e.g., Hawaii, Puerto Rico) and some extremeties of the 48 states (e.g., southern Florida).


Terminals are either general purpose or dedicated, as illustrated in Figure 8. General purpose terminals handle the needs of many users within a certain area. The area is typically a large city plus everything within one hundred miles or more. Users within the area are connected to a domsat operator’s central office (also called an operating center) by telephone lines and/or microwave relay. The central office, usually in the heart of the large city, is connected with the terminal by microwave relay over a distance of ten to fifty miles. The terminal is located in a rural valley in order to avoid interference with terrestrial microwave systems. Figure 9 shows the locations of the Los Angeles area general purpose terminals.

SBS chose the 12 and 14 GHz bands rather than 4 and 6 GHz because they (12/14 GHz) are not used for terrestrial microwave systems. Thus, there is no interference problem and no need to locate terminals outside urban areas. As a result, all SBS terminals will be of the dedicated type.

Dedicated terminals serve one user or perhaps two neighboring users. The antenna is located on the grounds of the user or on a rooftop. Equipment may be partitioned between a special shelter next to the antenna and the interior of the building.

Approximately 1000 dedicated terminals are in use today [24]. The great majority are receive only terminals (i. e., no transmitter) used for broadcasting signal distribution. In contrast, less than forty general purpose terminals exist outside of Alaska where RCA has installed about 100 small terminals to serve scattered small communities [25]. Within the

48 states, general purpose terminals are almost all located near New York, Washington,

D.C., Chicago, Atlanta, Houston, Dallas, Los Angeles, and San Francisco.


Each domsat system includes a TT&C (telemetry, tracking, and command) network to control the satellites. Control includes monitoring of telemetry, commanding various aspects of satellite configuration, and orbit computation. In the W. U. system, the control center also monitors, via satellite, the status of the general purpose terminals. Similar monitoring of terminal status via satellite links will be implemented in the SBS system.

Basic control network parameters are provided in Table 7. The Western Union [26] and

RCA [27] ground equipment is colocated with communication terminals and partially shares equipment. The Comstar satellites are controlled by Comsat General as part of the lease arrangement with AT&T [28-29]. Details of the SBS network have not been described, but are expected to be similar to the other systems except for the new TT&C frequencies corresponding to the SBS communication frequencies.


The domsat systems described above offer a variety of services using bandwidths from a few kilohertz up to tens of megahertz. The most common service is a voice bandwidth private line circuit offered by all of the domsat systems (except AT&T). “Private line” is in contrast to the terrestrial public switched telephone network. These private line circuits handle voice traffic, or alternate voice/data service. Some customer site equipments are able to switch among voice, data, teletype, and facsimile equipment to adapt the circuit to changing customer requirements. Private line circuits of various users may be multiplexed at a general purpose earth terminal or central office and transmitted together on one carrier. Private line circuits using dedicated earth terminals typically use single channel per carrier modulation. User locations vary from urban centers to off shore oil drilling platforms [30].

AT&T and GTE user the Comstar satellites as part of the long distance switched public network. RCA provides switched voice service to many remote Alaskan communities that are not served by terrestrial facilities. This service is both intrastate and interstate.

Distribution of broadcast television is another major service, primarily provided by RCA and W. U. Users include both the commercial and educational broadcasting networks [31].

Also, over a dozen companies originate television programming which is relayed to hundreds of separate cable TV systems throughout the country. RCA has also demonstrated the distribution of audio program materials for radio networks and teletype and facsimile for press services [32].

Both ASC and RCA provide digital data links for the U. S. Government. ASC traffic includes mission data and telemetry for the Defense Meteorological Satellite Program and

T1 carriers in the Autovon network. Both companies provide links for NASA in support of the Shuttle operations. Other NASA programs using domsat links include Seasat, Tiros,

TDRSS, and Viking.

Commercial data links at 56 kbps also use domsats. ASC customers include banking and computer companies and Dow Jones, which transmits composed Wall Street Journal pages to remote printing plants.


This paper has presented a survey of various aspects of U. S. domestic communication satellite systems. These systems have grown from proposals in the early 1970s to wellestablished businesses today, with much more growth expected in the coming decade.

Table 1. U. S. Dornsat History










A.B. C. Submits First Application to FCC

FCC Hearings Begin

Congress & Administration Studies Begin

Rostow Report Recommends Pilot Program

New Administration, New Study

FCC Open Entry Policy

8 System Applications; More FCC Hearings

Final FCC Policy

FCC Begins System Approvals

RCA Begins Service with Anik Lease

1974 Two Westar Satellites Launched

1975-6 Two RCA & Two AT&T Satellites Launched

1977 SBS Approved by FCC




Weight in orbit






Design Life


Launch Vehicle


Start of Operations




Satellites in Orbit 3

Frequency Bands

Channels /sat.

Service Area

4/6 Ghz

12 x 36 MHz


Comments Advanced


Satellites from 1982

Table 2. System Comparison






4/6 Ghz



Satellite Transponders Leased from W. U.


1976 (1973 with



(3rd Dec. 79)

4/6 Ghz

24 x 34 MHz





4/6 Ghz

24 x 34 MHz



Puerto Rico


Leased from

Comsat Gen.




6 ft. diam

12 ft. tall

650 Lbs

Table 3. Satellite Comparison



Body: 4x4x5 ft.

Array: 31 ft. span

1010 Lbs



8 ft. diam

20 ft. tall

1790 Lbs

260 W after 7 yrs.




7 Yrs.


Delta 1914

490 W min.

after 8 yrs.




7 Yrs.


Delta 3914

550 W min.

after 7 yrs.


±0.2E E-W

±0.26E N-S


7 Yrs.




7 ft. diam

22 ft. tall

1220 Lbs


830 W after

7 Yrs.




7 Yrs.


Delta 3910 or




2 Launches in 1980-81

12/14 Ghz

10 x 43 MHz


Table 4. Satellite Locations



Westar 2

Satcom 1

Satcom 2

Comstar 1

Comstar 2

Comstar 3

Westar 3

Satcom. 3



Satcom 4

* To be stationed between 100E and 130E


13 April 1974

10 October 1974

13 December 1975

26 March 1976

13 May 1976

22 July 1976

29 June 1978

9 August 1979

December 1979

August 1980

March 1981

June 1981


99E W












Table 5. Communication Subsystem Comparison


Channel Bandwidth

Transmit Band

TWT Power*

TWT Redundancy


Receive Band

Receiver Redundancy

Western Union

12 x 36 MHz

3.7 - 4.2 GHz

5 W none

33 dBW (spec)

34-36 dBW (typical)

5.925 - 6.425 GHz

2 for 1

CONUS G/T -6 dB/EK (typical)



24 x 34 MHz

3.7 - 4.2 GHz

5 W

0/4 spares

32 dBW (spec)

34 dBW (typical)

5.925 - 6.425 GHz

2 for 1

(each polarization)

-6 dB/EK


24 x 34 MHz

3.7 - 4.2 GHz

5/5.5 W none

33 dBW (spec)

34-36 dBW (typical)

5.925 - 6.425 GHz

4 for 2

33 dB

-8.8 dB/EK (spec)

-4.5 dB/EK (typical)

33 dB Dual Polarization


* Per channel


10 x 43 MHz

11.7 - 12.2 GHz

20 W

6 spares

40 - 44 dBW

14.0 - 14.5 GHz

4 for 1

-3 to +1 dB/EK


Antenna diarn. ft.

Frequency bands,


Transmit gain, dB

Transmit power, W


Receive gain, dB

G/T, dB/EK

Transmit beamwidth, deg



Table 6. Example Terminal Characteristics





33 - 36

(10-11 m)



(4.5-5 m)



















W. U.













1.1 (receive)

Step or

Manual many (some are TV receive only)

Manual many (usually

TV receive only)


(5.5 m)











Table 7. Satellite Control


Control Center Site

TT&C Sites

Command Frequency

Command Format

Telemetry Frequencies

Telemetry Format

TT&C site features

Western Union

Glenwood N. J.

Glenwood, Dallas,


6420 MHz

RCA at TT&C sites

Vernon Valley NJ

Moorpark CA

6423.5 MHz


Wash. D.C.

Southbury CN

Santa Paula CA

5930 MHz




14.49 GHz

(6175 MHz**)





3700.5 & 4199.5



3705 & 4195


12.19 GHz

(3950 MHz**)

PAM/FM/PM or analog/FM/PM

PAM/FM/PM or analog/FM/PM

PCM/PSK & FM subcarriers PM


Manned 24 hr/day, Glenwood manned

24 hr/day, UPS***, one extra antenna

Manned 24 hr/day,

* Part of Comsat General’s integrated Cornstar/Marisat control network

** Transfer orbit and initial testing only

***UPS = uninterrupted power supply digital PCM analog FM

Expected to be similar to others

Fig. 1. The Westar Satellite

Fig. 2. The Comstar Satellite

Fig. 3. The RCA Satellite


Figure 5. Generic Block Diagram

Fig. 6. A Typical 10 m (33 Ft.) Terminal

Fig. 7. A Roof-Top Terminal with a 15-Ft. Antenna

Figure 8. Two Types of Terminals

Figure 9. Terminals in the Los Angeles Area



E. V. Rostow, “Final Report of the Presidential Task Force on Communication

Policy”, December 17, 1968.


W. R. Hinchman, “Public Policy and the Domestic Satellite Industry”, International

Conference on Communications Record, ICC ’72, June 1972.


R. E. Greenquist, “First Generation Domestic Satellite Systems”, AIAA Paper

71-842, ALAA Space Systems Meeting, July 1971.


S. N. Verma, “U. S. Domestic Communication System Using Westar Satellites”,

World Telecommunications Forum Proceedings, Geneva, October 1975.


C. L. Washburn, “Westar Operations as Part of the Western Union Integrated

Transmission System”, AIAA Paper 78-540, AIAA 7th Communications Satellite

Systems Conference, April 1978.


L. Kilty, “American Satellite Builds a Major Network for Government Users,

EASCON ’75 Conference Proceedings, September 1975.


J. Christopher and D. Greenspan, “RCA Satcom Communication System”,

EASCON ’77 Conference Proceedings, September 1977.


R. D. Briskman, “The Comstar Program”, Comsat Technical Review, Vol. 6, No. 1,

Spring 1977 (Special issue on the Comstar Program).


J. D. Barnla and F. R. Zitzmann, “Digital Communications Satellite System of

SBS”, EASCON ’77 Conference Record, September 1977.


D. J. Lee, “System Performance of America’ s First Domestic Communications

Satellite-Westar”, EASCON ’74 Conference Record, September 1974.


S. N. Verma,. “Westar Communication Characteristics”, National

Telecommunications Conference Record, NTC ’74, December 1974.


J. E. Keigler, “RCA Satcom: An Example of Weight Optimized Satellite Design for

Maximum Communications Capacity”, Acta Astronautica, Vol. 5, Nos. 3-4, March-

April 1978.


J. E. Keigler and C. R. Hume, “The RCA Satcom. Satellite”, Journal of the British

Interplanetary Society, Vol. 29, No. 9, September 1976.


M. C. Kim, “Comsat General’s Domestic Satellite System (Comstar)”, Paper 9/4,

Wescon Technical Papers, September 1966.


G. E. A. Abutaleb, et al., “The Comstar Satellite System”, Comsat Technical

Review, Vol. 7, No. 1, Spring 1977.


H. A. Rosen, “The SBS Communication Satellite - An Integrated Design”,

EASCON ’78 Conference Record, September 1978.


M. T. Lyons and P. C. Dougherty, “Spacecraft Design for the SBS System”, AIAA

Paper 78-545, AIAA 7th Communications Satellite Systems Conference, April



J. Ramasastry, et al., “Western Union’s Satellite-Switched TDMA Advanced

Westar System”, AIAA Paper 78-602, AIAA 7th Communications Satellite Systems

Conference, April 1978.


J. Ramasastry, et al., “Advanced Westar SS/TDMA System”, Proceedings of the

Fourth International Conference on Digital Satellite Communications, October 1978.


J. Cuddihy and J. M. Walsh, “RCA Satcom Earth Station Facilities”, RCA

Engineer, Vol. 22, No. 1, June/July, 1976.


L. Zahalka, et al., “Frequency Re-Use in GTE Earth Stations with Beam Waveguide

Feed”, Paper 8.4, International Conference on Communications Record, ICC ’77,

June 1977.


D. H. Westwood, “Customer Premises RF Terminals for the SBS System”, Paper

6.3, International Conference on Communications Record, ICC ’79, June 1979.


Papers of Session 32 (Small Earth Terminals for Satellite Communications),

National Telecommunications Conference Record, NTC ’78, December 1978.


S. Topol, “Satellite Communications - History and Future”, Microwave Journal,

Vol. 21, No. 11, November 1978.


J. L. Rivard, “The Domestic Satellite Program in Alaska”, National

Telecommunications Conference Record, NTC ’74, December 1974.


J. W. Van Cleve, “Operation and Control of an Integrated Satellite/Terrestrial

Transmission Network”, International Conference on Communication Record , ICC

’77, June 1977.


J. Lewin, “Ground-Control System for Satcom Satellites”, AIAA Paper 78-539,

AIAA 7th Communications Satellite Systems Conference, April 1978.


W. J. Gribbin and D. J. Lee, “Technological Development in Spacecraft Command and Control Systems”, National Telecommunications Conference Record, NTC ’77,

December 1977.


W. J. Gribbin and R. S. Cooperman, “Comsat General Satellite Technical Control

Network”, Comsat Technical Review, Vol. 7, No. 1, Spring 1977.


G. L. Sarver, “Satellite Communications for Off-Shore Oil Operations Using

Westar”, National Telecommunications Conference Record, NTC ’75, December



J. E. D. Ball and P. A. Rubin, “Communication Satellites for Public Television”,

IEEE Transactions on Broadcasting, Vol. 24, No. 2, June 1978.


R. M. Lansey and M. R. Freeling, “RCA’s Satellite Distribution System for Small-

Dish Earth Terminals”, EASCON ’78 Conference Record, September 1978.







We investigate a time-varying trellis coded multiple-access scheme using noncoherent mFSK signals. Techniques similar to this were originally proposed by Cohen, Heller, and

Viterbi [1] and more recently in a mFSK form by Viterbi.

[2] In these multiple-access systems van der Muelen






, Gaarder and Wolf


, Kasami and Lin


, Weldon

[8] and Wolf


have shown that the decoded symbols of one user can be used to reduce the

“multiple-access noise” to other users and thus allow for a larger achievable rate region than one would expect with conventional time division multiple-access techniques. In some cases, specific codes were investigated. Peterson and Costello [11] have extended these earlier works to convolutional and trellis codes. In this case the decoder is designed as a “super” Viterbi decoder that regards all transmitter trellis codes combined to form a single “super” trellis encoder.

[10] and Chevillat

In this paper we investigate the noncoherent mFSK scheme discussed by Viterbi [2] and generalize to single level and multi-level energy detectors with a single “super” Viterbi decoder at the receiver. The main results are random coding bounds for the general case where L users each have remotely located time-varying trellis encoders of constraint length

K. We assume throughout that the channel is noiseless, and symbol timing synchronization is maintained among the L users. These assumptions are being relaxed in the thesis research of Sorace [12] .


We assume that during T seconds each of L users will transmit an mFSK tone where the m frequencies are {f




,...,f m

}. By choosing

we have orthogonal tones of total bandwidth B m

given approximately by

Conventional frequency division multiple-access (FDMA) divides the m frequencies into L disjoint subsets and assign each user one such subset of frequencies. Time division multiple-access (TDMA) allows all m frequencies to be used by each of the L users on a time-shared basis. In this paper we allow all L users to simultaneously use all m frequencies in a mFSK modulation scheme where each user sends one of m frequency tones every T seconds.

Throughout this paper we assume the channel is noiseless and the receiver’s front end consists of m energy detectors that measure the incoming signal energy at each of the frequencies {f




,...,f m

}. All L users are synchronized so that each user transmits a tone at the same fixed T second intervals. Also we assume each of the L tones have the same energy and whenever several tones are transmitted at one frequency the resulting received signal energy during the T second interval is the sum of the individual signal energies

(noncoherent combining of signals). The m frequency energy detectors can be implemented by “chirp Z-transform” devices and have been suggested for use on-board satellites.


Figure 1 illustrates the system discussed above. The energy detectors output the vector at each T second interval where e e i i

is the measured energy at the frequency f i

during the T second interval. Since the tones are orthogonal we assume there is no energy component in f i

due to tones of frequency f

, we have e i j

where j … i. Thus, when R tones are transmitted at frequency

= R> where > is the energy of each tone. As shown in Figure 1, the energy detectors are followed by quantizers with outputs denoted by {n

1 quantization.



,...,n m

}. This completes the model of the mFSK multiple-access channel. We shall consider here two types of


For some threshold 0 < * < > we define the hard quantizer output of the ith detector as


This results in energy detectors that measure the presences of one or more tones at each frequency without regard for the number of tones at each frequency. Every T second interval results in the channel output

If L $ m, the number of possible distinct channel outputs is since (0,0,0,...,0) is not allowed. If L < m then there can be at most L nonzero components in n. Hence the number of distinct outputs is we define the output alphabet size as

(2) and denote the possible outputs as b




,...,b j


With L users each sending one of m tones, there are possible channel inputs which we denote as a








. Hence we have a K input, J =

(L,m) output discrete memoryless channel. We illustrate this with some examples.

Example: L = 2, m = 2



The soft decision channel is one where the ith quantizer output is given by

That is, each quantizer output indicates the number of tones received at that particular frequency. The channel output is thus

(4) where and

Let J

SD n

1 n



(L,m) be the number of distanct channel outputs. The number of sequences m-1 where is J


(i, m-1) so that we have the recursion formula,


It is easy to see that end conditions are


We denote the J = J


(L,m) outputs as b






and the K = m


inputs as a




,..., a



Example: L = 2, m = 2

This is identical to the hard decision channel with L = 2, m = 2.


Let us now return to the L = 2, m = 2 multiple-access channel where now the two users have time-varying trellis codes


of constant length L is the rate r = b/n bits per channel symbol. Here b and n are integres and L is the encoder memory consisting of L 2 b

-ary sumbols or bL data bits. Each time b data bits enter the encoder, n FSK signals are transmitted (m = 2 here). Hence data bits enter the trellis encoder at a rate of one bit every nT/b seconds.

If the Rth trellis code output sequence is where then we can define a “super” trellis code output sequence where belong to the multiple-access channel input alphabet.

The “super” trellis encoder illustrated in Figure 2 consists of the L user trellis encoders and therefore has constraint length L and rate Lb/n bits per multiple-access channel input symbol. That is, it inputs Lb bits (b bits from each of L sources) and outputs n symbols from, , the multiple-access channel input alphabet. The maximum likelihood receiver can be realized with the Viterbi algorithm. In practice, since we have a zero-one metric, the sequential decoding algorithms may be more practical. Also sequential decoding can handle the large number of states 2 LbL . In the following, however, we shall assume the Viterbi decoding algorithm to examine achievable performance.


[1] A. R. Cohen, J. A. Heller, and A. J. Viterbi, “A New Coding Technique for

Asynchronous Multiple-Access Communication, “ IEEE Trans. Commun, Tech.,

Vol. COM-19, October 1971, pp. 849-856.

[2] A. J. Viterbi, “A Processing Satellite Transponder for Multiple-Access by Low-Rate

Mobile Users,” National Radio Science Meeting,” Boulder, Colo., November 6-9,


[3] E. C. Van der Meulen, “A Survey of Multi-Way Channels in Information Theory

1961-1976,” IEEE Trans. Inform. Theory, Vol. IT-23, pp. 1-37, January 1977.

[4] R. Ahlswede, “Multi-way Communication Channels,” Proc. 2nd Int. Symp. Inform.

Transmission, Hungarian Press, Tshkadsor, Armenia, U.S.S.R., 1971.

[5] Henry H. J. Liao, “Multiple-Access Channels,” Ph.D. dissertation, Dept. Electrical

Engineering, University of Hawaii, Honolulu, HI, 1972.

[6] N. T. Gaarder and J. K. Wolf, “The Capacity Region of a Multiple-Access Discrete

Memoryless Channel can Increase with Feedback,” IEEE Trans. Inform. Theory,

Vol. IT-21, pp. 100-102, January 1975.

[7] T. Kasami and S. Lin, “Bounds on the Achievable Rates of Block Coding for a

Memoryless Multiple-Access Channel,” IEEE Trans. Inform. Theory, Vol. IT-24, pp. 187-197, March 1978.

[8] E. J. Weldon, “Coding for a Multiple-Access Channel,” Inform. Contr., Vol. 36, pp. 256-274, March 1978.

[9] J. K. Wolf, “Error Control for Channels with Crosstalk,” National Radio Science

Meeting, Boulder, Colo., November 6-9, 1978.

[10] R. L. Peterson and D. J. Costello, “Binary Convolutional Codes for a Multiple-

Access Channel,” IEEE Trans. Inform. Theory, Vol. IT-25, pp. 101-105, January


[11] P. R. Chevillat, “N-User Trellis Coding for a Class of Multiple-Access Channels,

IBM Research Report, Zurich, Switzerland, November 1978.

[12] R. Sorace, “Analysis of Noncoherent Multiple-Access Channels,” Ph.D. thesis research, University of California, Los Angeles, (in progress).

[13] R. G. Gallager, “A Simple Derivation of the Coding Theorem and Some

Applications,” IEEE Trans. Inform. Theory, Vol. IT-11, pp. 3-18, January 1965.

[14] A. J. Viterbi and J. K. Omura, Principles of Digital Communication and Coding,

McGraw-Hill, New York, 1979.





Wai-Hung Ng

The Aerospace Corporation

Los Angeles, California


The use of convolutional codes with feedback decoding is the most common errorcorrection technique in simple communication systems. A drawback of conventional feedback decoding is the limitation to a class of self-orthogonal codes which, in general, are non-optimum. Based on distance properties of the utilized code and test-error pattern analysis, we propose generalized feedback decoding that does not have the above mentioned limitation. This is minimum distance decoding and can be applied to any convolutional code while still maintaining its simplicity. Therefore, it has the advantage of being easily adopted in the existing systems. We can use complicated Viterbi or sequential decoder in large terminals and, with the same code, use the proposed decoder in small terminals; otherwise, both large and small terminals must utilize the same type decoder.

Also, we may use the proposed decoding scheme to simplify and accelerate sequential type decoding. In addition, by means of the special recovery property of convolutional codes, advanced ARQ retransmission systems could be much improved; several practical applications are suggested and discussed in the last section of this paper.


Various decoding approaches to convolutional codes have resulted in significant improvements in practical digital communication applications. In general, feedback decoding is utilized in simple systems, Viterbi or sequential decoding in relatively complicated systems, and stop-and-wait or continous retransmission (ARQ for detection only or hybrid forward error-correction/ARQ) in computer communication systems over noisy channel. Thus, it is desirable that, with appropriate decoders, all these different systems could use a common code to minimize their interface problem. However, conventional feedback decoding requires self-orthogonal codes [1] , which are not suitable for other decoders, and therefore obstructs such an approach.

In this paper, we propose generalized feedback decoding which can be applied to any convolutional code, but still maintains its simplicity. In order to give a clear illustration, the

discussion will be concentrated on binary rate 1/2 single generator convolutional codes.

The approach presented, however, can be extended to other code rates.

This algorithm is minimum distance decoding, which guarantees better performance than conventional feedback decoding. Examples are given for both theoretical analysis and system design to show its simplicity in hardware implementation.

Although the main purpose of this study is to introduce the generalized feedback decoding, ways of utilizing properties of convolutional codes to further simplify and speed up sequential type decoding and suggestions for improving ARQ retransmission system will also be discussed.


Without loss of generality, we shall state that two code branches stemming from the same node are always binary complements of each other, either 00 and 11 or 01 and 10. Thus, independent of the received code branch, two branches of test-error patterns stemming from the same node are either 00 and 11 or 01 and 10. Since choosing between 01 or 10 would make no difference in terms of distance, we could impose a rule that under BBO, the newly accepted test-error pattern must be either 00 or 01, and eliminate the possibilities of being equal to 10 and 11.


In this section, we shall begin with the analysis of coding properties and their utilization before introducing the implementation.

3.1 Test-Error Pattern Tree

3.1.2 Development of Minimum Test-Error Pattern Tree

3.2 Permissible Path Decoding


It is believed that the only class of common codes could be utilized for different decoders is convolutional codes. Since the probability of decoding error decreases exponentially with constraint length K, it is preferable to utilize a long code, especially when extremely low probability of data transmission error is one of the system requirements.

The performance of a system using a long code depends on computational and buffer capabilities of the decoder (i.e., the performance is a function of the average and maximum number of decoding operations required per search).Here, we shall discuss several

different approaches to minimize the decoding effort and thus to achieve a much higher decoding speed, based on both the proposed generalized feedback decoding approach and the special recovery property of convolutional codes [9].

4.1 Utilization of Generalized Feedback Decoding

It has been suggested that multiple stack algorithm (MSA) can achieve lower error probability and higher decoding speed than soft decision Viterbi decoding [7] . One could see that the decoding procedure described above would be highly desirable.

[6] would be simpler and require less decoding effort than MSA, effort in utilizing this decoding approach to simplify MSA

4.2 Suggestions for Improving Advanced ARQ Retransmission Systems

Recently, study on computer communication systems with ARQ retransmission techniques has been increased [8] . Most of these studies are limited to either error detection only or detection with simple forward error correction, and because of the high retransmission rate, these systems are quite inefficient


. Therefore, it is necessary to utilize powerful error correction code to minimize the required retransmission process.

In general, sequential decoding would be a very powerful error-correction technique.

However, the difficulty of employing the usual sequential type decoding in ARQ systems lies in the fact that its Paretian computational distribution problem often causes buffer overflow.

It is fully understood that the Paretian computational distribution problem arises from sequential decoding algorithm, not from the basic properties of convolutional codes, and the recovery property of convolutional codes (the ability to recover to correct path after decoder accepting errors [9] ) could be utilized in the following way to overcome the existing difficulties of buffer overflow and to achieve maximum likelihood decoding of using long codes. By carefully selecting the coding parameters (such as employing

systematic code, soft-decision with longer search length), the length of recovery decoding error bursts could be minimized. The recovery decoding error bursts are then detected from the forward decoded sequence, and redecoded by a bidirectional search procedure to achieve a maximum likelihood decoding without the hazard of buffer overflow


. Study in this direction to improve advanced ARQ retransmission has been initiated, and results are anticipated in the near future.



Massey, J. L. : Threshold Decoding, The MIT Press, 196Z.


Wozencraft, J. M. and Reiffen, B. : Sequential Decoding , The MIT Press, 1961.


Ng, W. H., and Pfeiffer, P. E. : “Minirnum-Hamming-distance decoding of single generator binary convolutional codes”, Information & Control, October, 1968, pp. 295-315.


Wozencraft, J. M. and Jacobs, I. M. : Principles of Communication Engineering,

John Wiley & Sons, 1965.


Ng, W. H. : “An upper bound on the back-up depth for maximum likelihood decoding of convolutional codes”, IEEE Trans. on Information Theory, May 1976, pp. 354-357.


Ng, W. H. and Goodman, R. M. F. : “An efficient minimum-distance decoding algorithm for convolutional error-correcting codes”, Proc. IEE, February 1978, pp. 97-103.


Chevillat, P. R. and Costello, D. J., Jr. : “A multiple stack algorithm for erasurefree decoding of convolutional codes”, IEEE Trans. on Communications, December

1977, pp. 1460-1470.


Towsley, D. and Wolf, J. K. : “On the statistical analysis of Queue lengths and waiting times for statistical multiplexers with ARQ retransmission schemes”, IEEE

Trans. on Communications, April 1979, ppl 693-702.


Ng, W. H. : “Study on decoding recovery behavior for convolutional codes”, IEEE

Trans. on Information Theory, Nov. 1970, pp. 795-797.


Ng, W. H. : “Bidirectional search for convolutional codes”, Proc. IEE, June 1978, pp. 495-500.



Frank Torres

Supervisor, STS Payload Integration

Electronic Systems, Space Systems Group

Rockwell International

12214 Lakewood Blvd, Downey, California


The NASA/Rockwell Space Shuttle with its highly versatile avionics and electrical provisions for use by the Shuttle payloads will provide an efficient system for future national space program activities and space program activities from foreign countries. This paper summarizes the avionics and electrical payload capabilities and interface characteristics. It includes a description of the command and data systems interface, the caution and warning system interface, and the aft flight deck accommodations; the electrical power distribution system; and the standard mixed cargo harness.


The primary mission for the Space Shuttle is the delivery of payloads to earth orbit. In orbit, the Space Shuttle has the capability to carry out missions unique to the space program: to deploy payloads whose destination is high-altitude orbits; to retrieve payloads from orbit for reuse; to service or refurbish satellites in space; and to operate space laboratories in orbit. To accomplish these objectives, the Shuttle orbiter provides an avionic system that interfaces with payloads through the payload and mission specialist stations by means of hardwired controls and displays when the payloads are attached to the orbiter and through a radio link when the payloads are detached. The capability and interface characteristics of the avionics system that has been designed in support of payload operations and described in detail in NASA/JSC 07700, Volume XIV, are summarized in this paper.

The standard avionics provisions for payloads include communication, command, and data management interfaces with attached and detached payloads and electrical power and caution/warning interfaces for attached payloads. A functional block diagram of these standard avionics provisions is shown in Figure 1.

Orbiter transmitters, receivers, and signal structure are compatible with the established characteristics of NASA’s space tracking and data network (STDN) and tracking and data relay satellite (TDRS), and the Air Force space-ground link subsystem (SGLS). The orbiter is equipped to communicate with ground stations, TDR satellites, and detached payloads at S-band frequencies and with the TDRS at Ku-band frequencies.


Commands to attached or detached payloads can be forwarded from the ground or generated on-board. A block diagram of the command system interface is shown in

Figure 2.

Payload Interrogator

The commands to detached payloads are via the payload interrogator, whose transmitter is capable of operating in three prime modes: STDN, deep space netork (DSN), and DOD.

The transmit frequency channels available are as follows:



STDN compatible payload - 808 channels from 2025.833333 to 2118.722328 MHz in

115.104 kHz increments (221/240 transmit/receive ratio).

DSN compatible payloads - 29 channels from 2110.243056 to 2119.792438 MHz in

341.049 kHz increments (221/240 transmit/receive ratio).


SGLS compatible payloads - 20 channels from 1763.721 to 1839.795 MRz in 4.004

MHz increments (205/256 transmit/receive ratio).

In the STDN and DSN modes, the interrogator accepts a 16-kHz subcarrier (PSKmodulated by command signals ranging from 125/16 to 2000 bps rate) and phase modulates the selected S-band carrier for RF transmission to the detached payload. In the

DOD mode, the interrogator accepts a 65 (“S”), or 76 kHz (“0”), or 95 kHz (“1”) subcarrier, amplitude-modulated by a 1-kHz or 2-kHz triangular wave from the userprovided signal processor in the payload station. The accepted signal phase modulates the selected RF carrier.

Payload Signal Processor (PSP)

In the STDN and DSN modes, the baseband signal is available to attached payloads via the PSP, which offers five differential outputs. Each output provides a 4 volt peak-to-peak

(line-to-line), 16-kHz sinewave, PSK-modulated by the command data at data rates

ranging from 125/12 to 2000 bps. The data waveform can be any one of the following:

Bi-N-L, -M, or -S; or NRZ-L, -M, or -S.

Multiplexer/Demultiplexer (MDM)

Serial digital and discrete commands are provided by the MDM. These commands can also be forwarded from the ground stations or generated on-board initiated via the keyboard. In the serial mode, the MDM command signal consists of a Manchester II biphase signal at a 1-mps burst rate. The true logic level is a nominal +4.5 Vdc, and the false logic level is a nominal -4.5 Vdc. The command signal channel includes word, message-in, message-out discretes, whose logic levels are nominally +3.5 Vdc and -

3.5 Vdc for the true and false logic levels. The command message can include up to 32

16-bit data words. In the discrete output mode the MDM provides low-level and high-level discretes whose true/false logic levels are nominally +5/0 Vdc and +28/0 Vdc. Thirty-six high-level discretes (DOH) and thirty-two low-level discretes (DOL) are available.

Data Bus

For payloads requiring a greater number of commands, a data bus stub is provided to accommodate the installation of a payload-provided MDM.

Standard Switch Panel

Hardwired commands are provided by the standard switch panel (Figure 3) and the C3A5 panel located in the cockpit center console.

Ku-Band Signal Processor

The Ku-band signal processor provides the capability to forward 128 kbps of NRZ-L data plus clock. The true and false logic levels for this interface are +5 Vdc and -5 Vdc.


The Shuttle orbiter avionics provides the capability to process payload data on-board, transmit data to the ground stations in real time, or record the data for later data dump to ground stations. A block diagram of the payload data system interfaces is shown in

Figure 4.

Payload Interrogator

Data from detached payloads are via the payload interrogator, whose receiver is capable of operating in three prime modes: STDN, DSN, and DOD. The receive frequency channels available are as follows:



STDN compatible payloads - 808 channels from 2200.000 to 2300.875 MHz in

125-kHz increments.

DSN compatible payloads - 27 channels from 2290.185 to 2299.814 MHz in

370.37-MHz increments.


SGLS compatible payloads - 20 channels from 2202.500 to 2297.500 MHz in 5-MHz increments.

In the STDN or DSN modes, the payload interrogator receives the selected RF carrier and detects a PCM/PSK-modulated 1.024-MHz subcarrier. The detected data are routed to other orbiter avionics line replaceable units for additional on-board processing and to be transmitted to ground stations.

In the DOD mode, the payload interrogator receives the selected phase-modulated carrier and detects the PSK-modulated 1.024-MHz subcarrier or the FM/FM-modulated 1.7-MHz subcarrier. The detected data are made available at the payload station where they can be accepted by the user-provided data processor.

The payload interrogator has the capability to select three predetermined sensitivity levels:

Sensitivity (dbm)

Carrier acquisition


16 kbps, PSK, BER 10


16 kbps, FM/FM, BER 10














Payload Data Interleaver (PDI)

Attached payload telemetry interface is via the PDI. The PDI also receives data from the payload signal processor (PSP), which processes data from detached payloads received via the payload interrogator. The programmable PDI can interface with up to five payloads and accept data simultaneously from four different payloads, then select and individually decommutate the data for storage in a buffer memory. The memory is accessible to the

pulse code modulator master unit (PCMMU), which, after accepting the data from the

PDI, formats the data into a serial digital stream for telemetry to the ground. A block diagram of the PDI is shown in Figure 5. The PDI provides the following capabilities to accommodate attached payload telemetry requirements: (1) seven input channel ports (five payloads, 1 each of two PSP) and capability to switch any of the seven inputs to any of four PDI channels; (2) capability to decommutate and process up to four asynchronous pulse code modulated data input channels simultaneously; (3) 64 kbps maximum input data rate for complete throughput of data; and (4) in-flight programmable PDI decommutators so as to be able to accommodate a change in payload data rate and format in flight. The

PDI input telemetry format constraints are as follows:


Bits per word - 8 or multiples of 8




Words per frame - 8 to 1024

Minor frame rate - 200 per sec maximum

Minor frames per master frame - 1 to 256



Minor frame synchronization - 8, 16, 24, or 32 bits

Master frame synchronization - 8-bit unique pattern in first minor frame or 8-bit minor frame counter


Format sample rate - 5 maximum

The PDI has a balanced differential input circuit. The minimum/maximum input voltage requirements measured line-to-line are 3/9 volts peak-to-peak. The PDI will accept

NRZ-L, NRZ-M, NRZ-S, Bi-N-L, Bi-N-M, and Bi-N-S data. The bit rate clock is required with NRZ codes.

S-Band FM Link

Because of limited ground coverage, utilization of this link is restricted to ascent and landing phases after the tracking and data relay satellite system (Ku-band) is operational.

Attached payload data can be patched to the FM signal processor at the payload station distribution panel. Payload data can time-share the FM link with the orbiter operational recorder dump, TV video, and main engine data. The FM signal processor will select one of the following payload inputs for transmission to the STDN at any time the FM link is not being used by the orbiter: payload recorder dump; wideband digital data (NASA

payloads), 200 bps to 5 mbps; wideband analog data (NASA payloads), 300 Hz to 4 MHz; and digital data (DOD payloads), 250 bps to 250 kbps.

The FM signal processor has balanced differential input circuits. The digital data (DOD) channel will accept either biphase NRZ data. The input voltage requirement is 1 volt peakto-peak. The digital data wideband channel receives Bi-N-L or NRZ-L data (Bi-N-L is limited to 2 mbps). The input voltage requirement is 5 volts peak-to-peak. The analog data channel input voltage requirement is 1 volt peak-to-peak.

Ku-Band Link

On orbit, the Ku-band system provides near continuous ground coverage. Attached payload data can be patched to the Ku-band signal processor at the payload station distribution panel. For downlink, the Ku-band system operates in two modes, quadrature phase shift key (QRSK, Mode 1) and FM (Mode 2), with three channels of input data in each mode. The Ku-band communication system is a combined system with the rendezvous radar. The system cannot be used in both modes at the same time. Ku-band service is provided only when the payload bay doors are open.

Two out of three channels in the Ku-band signal processor receive payload data. The interface characteristics are given in Table I.

Table I. Ku-Band Digital Processor Channel Interfaces








Mode 1

16 to 1024 kbps Bi-N-L or 16 to 2000 kbps NRZ-L, -M, or -S

Balanced differential

5 volts peak-to-peak

TSP, 75 ±5 ohms

Digital (analog input not used)


16 to 4000 kbps NRZ-L, -M, or -S



Balanced differential

5 volts peak-to-peak

Analog (digital input not used)


Dc to 4.5 MHz




Balanced differential

1 volt peak-to-peak

TSP, 75 ±5 ohms

Mode 2

Same as Mode 1

Digital only





2 to 50 mbps NRZ-L,

-M, or -S

Single ended

5 volts peak-to-peak

Coaxial, 50 ± 5 ohms

In Channel 2, the payload timeshares the Ku-band system with the operational recorder dump, and the payload recorder dump. Mission planning is required to ensure that use of this interface is properly coordinated. Likewise in Channel 3, the Ku-band system is timeshared by the payload that is sending digital data, the payload that is sending analog data, and the orbiter TV downlink.

Payload Recorder

Payload experiment data recording is provided via the payload station. Predetermined payload station wiring permits digital data recording in parallel (up to 14 tracks) or in a combination of parallel and serial track sequences. Data rates from 25.5 to 1024 kbps can be selected from four tape speeds provided by premission wiring of recorder program plugs.

Analog data can be recorded on up to 14 tracks in parallel with frequency from 1.9 kHz to

1.6 MHz by premission program wiring. The payload recorder has the following record/playback capabilities:

Data Rate


64 to 128

128 to 256

256 to 512

512 to 1024

Freq Range


1.9 to 250

3.8 to 500

7.5 to 1000

15 to 1600

Selectable Tape Speed






The payload recorder interface characteristics are as follows:

Signal level

Coupling type

Cable impedance

Rise/fall time

Bit jitter

Digital Input

(Bi- phase -L)

3 to 9 volts peak-to-peak, line-to-line

Direct coupled balanced differential

75 ohms TSP

10 percent of bit duration

< 2 percent of bit duration


Time Per Tape Pass






Analog Input

1 volt rms

Transformer coupled

75 ohms TSP


Multiplexer/Demultiplexer (MDM) Interface

The MDM is capable of receiving payload data in the form of 5-volt discretes (DIL), 0 to

5-volt differential analog inputs (AID), and serial data (up to 32 16-bit words) at a 1-mbps burst rate. Payload downlink telemetry originating at this interface may be substituted for

PDI downlink data on orbit. Thirty-two DIL’s, eight AID’s, and four serial channels are available to the user as a standard service. The interface characteristics for these inputs are as follows:



- Logic levels: 5 Vdc, True; 0 Vdc, False

- TSP, 75 ohms



- Signal amplitude: differential +5.11 to -5.12 Vdc

- TSP, 75 ohms


Serial digital channel (4 lines)

- Manchester II Bi-N at 1 mps (burst)

- Data line true/false logic levels (nominal): +4.5 Vdc/-4.5 Vdc

- Word, message-in, message-out true/false logic levels (nominal): +3.5 Vdc/-3.5 Vdc

Master Timing Unit (MTU)

The MTU provides Greenwich Mean Time (GMT) and Mission Elapsed Time (MET) time code outputs in IRIG-B/modified code formats to payloads as a standard service. The voltage level of these signals is nominally 5 volts peak-to-peak. A 75-ohm TSP cable is used.

Payload Caution and Warning System Interface

Orbiter caution and warning is defined to include emergency, warning, and caution parameters. In the case of manned payloads, two identified orbiter emergency parameters require instinctive, immediate crew corrective action. They are fire/smoke and loss of cabin pressure. The caution and warning electronics unit will receive up to five hardwiredsensor dedicated payload warning parameters. The MDM will receive a total of 50 software-controlled payload caution inputs. They consist of 25 5-volt discretes and 25 0 to

5-volt differential analog inputs.

Safing commands can be generated by use of the five redundant switches on the C3A5 panel located at the commander/pilot center station. Use of these switches requires wire

patching at the payload station distribution panel. The payload MDM’s provide up to 36 safing commands consisting of 28-volt discretes. The commands can be initiated on board via the keyboard or on the ground via the uplink/forward link. Eighteen of the 36 safing commands require wire patching at the mission station distribution panel.

Figure 6 shows a block diagram of the payload caution and warning system interface. The interfaces are keyed to the caution and warning signal characteristics in the following list:

1. Dedicated caution and warning electronics unit


5 signal paths


Any combination of 5-Vdc discretes, 28-Vdc discretes, 0 to 5-V analog

2. MDM caution and warning inputs - direct to MDM


12 analog differential input (AID) signal paths




Voltage range: +5.12 Vdc to -5.12 Vdc

13 discrete input: low level (DIL) signal paths

Logic levels: True, +5 ±1.0 Vdc; False, 0 ±0.5 Vdc

3. MDM caution and warning inputs: via mission station


13 analog differential input (AID) signal paths




Voltage range: +5.12 Vdc to -5.12 Vdc

12 discrete input: low level (DIL) signal paths

Logic levels: True, +5 ±1.0 Vdc; False, 0 ±0.5 Vdc

4. MDM safing commands: direct from MDM


20 discrete outputs: high-level (DOH) signal paths


Logic levels: True, +19.5 to +32 Vdc; False, 0 to 3 Vdc

5. MDM safing commands: via mission station


18 discrete outputs: high-level (DOH) signal paths


Logic levels: True, +19.5 to +32 Vdc; False, 0 to 3 Vdc

6. C3A5 panel. switches


5 redundant safing command signals consisting of switch closures

7. )P/)T to Panel R13 A3


One differential analog signal


Voltage range: 0 to 5 Vdc

8. Master alarm light to payload


One witch closure to activate the master alarm light in the payload



No alarm: 20-megohm isolation

Alarm: #0.1-ohm contact resistance

9. Master alarm tone to payload


One tone: analog 1 Hz


Voltage: 2.2 V ±10 percent, dc isolated and balanced

Fire/smoke: siren - )P/)T: klaxon - C&W: C&W tone

Aft Flight Deck Accommodations

The aft portion of the flight deck is an integrated crew station arranged for flight control, rendezvous and docking, payload manipulating, and payload operations. The design provides the flexibility and capability for the check out, monitor, command, and control of attached and detached payload operations. The aft flight deck crew stations (mission, payload, and orbit stations) are designed in a modular fashion to facilitate ground changeout and installation of Shuttle and payload-supplied display and control panels and associated equipment.

Figure 7 illustrates the aft flight deck stations. The panels that are cross-hatched are dedicated to payloads. Panels R-11, L-10, L-11, and L-12 are standard 19-inch wide panel spaces with required standardized electrical power connectors for accommodating user unique modules.

Two standard switch panels are provided as a service to users. They are located in L-12.

One-half of a standard switch panel (SSP) has been allocated to each of four payloads sharing a Shuttle flight. One-half of an SSP provides twelve switches and twelve talkbacks.

Payload Electrical Power Distribution System Interface

DC power is available to payloads in the payload bay and to payload equipment in the aft flight deck during the ascent/entry and on-orbit mission phases. The on-orbit payload bay power level is 7 kW steady state, with a peak power of 12 kW available for up to 15 minutes once each three hours when the heat rejection kit is installed and all eight radiators are deployed. Without the additional heat rejection kit, total steady-state power drawn by all payloads in the payload bay is limited to 6 kW. During ascent and entry, the total power available to the payload is limited to a maximum of 1.0 kW. Auxiliary power is available to a maximum of 400 watts. The auxiliary power remains on in the event that the main power is interrupted. Use of the auxiliary power reduces the amount of power that

may be used from the primary feeder by an equivalent amount. During emergency conditions, a maximum of 200 watts auxiliary power will be available for payload safing.

In the aft flight deck, the total on-orbit power level is 750 watts continuous, 1000 watts peak. Three-Phase ac power up to 0.69 kVA is also available and can be substituted for an equivalent amount of dc power. The ac power can be made available to the payload bay by doing some wire patching at the mission station. Power levels during ascent and entry are

350 watts continuous, 420 watts peak for two minutes.

The payload electrical power distribution system interfaces are shown in Figure 8. The maximum power available to payloads at the different interfaces is as follows:


Mid-payload bay

- Main power: 7 kW avg; 12 kW peak from dedicated fuel cell (with radiator kit)

- Main power: 5 kW avg; 8 kW peak (orbiter shared) fuel cell

- Auxiliary power*: 0.4 kW avg (continuous)


Aft payload bay

- Power: 1.5 kW avg; 2 kW peak each


Aft flight deck** PS and MS distribution panels

- DC power: 0.75 kW avg; 1 kW peak

- AC power: 0.69 kVA (3-phase); 1 kVA (3-phase) from crbiter inverter

- Auxiliary power*: 0.4 kW avg (continuous) (PS only)

*0.4 kW total for combined interfaces

**AFD total, all sources, 0.75 kW; 1 kW peak

Standard Mixed Cargo Harness

The standard mixed cargo harness (SMCH) in the cargo bay provides the following to each of four payload sections: main dc power cable, control and signal wiring to AFD,

GPC data bus (2), direct connection to an orbiter MDM, and control and signal wiring to

GSE via T-0 umbilical.

Distribution to four payload sections in the cargo bay is effected by four functionally identical cable sets (Figure 9). The standard AFD harness permits connection of the control and signal wiring from the cargo bay to the orbiter’s avionics services and to display and control panels on the AFD. This harness provides access to all standard avionics services, as described in the preceding sections of this paper, for each user, with some provisions for additional avionics services. It also provides both powered and

unpowered switch functions and electromechanical status indicators from the standard switch panel. Figure 10 depicts the four SMCH sets that provide access to the signal and control avionics services. Figures 11 and 12 depict the nonstandard avionics services at the payload and mission stations that can be wire-patched to the payload interface using userprovided patch cables.


The Shuttle avionics system provides a wide range of services to accommodate user payloads, including commands, data management, displays and controls, data recording,

RF communications, electrical power, aft flight deck displays and controls panel space, and standard wire harnesses. The standard avionics and electrical provisions have enough flexibility to accommodate four major payloads per flight. The orbiter communication subsystem supports the transfer of payload telemetry and uplink data commands to and from the space networks. A payload recorder is provided for recording analog and digital data, which can be transmitted to ground stations via the orbiter communications system.

A caution and warning subsystem is provided to monitor payload signals announcing a failure that could result in a hazard to the orbiter or crew. A standard switch panel is provided for generating hardwired commands to attached payloads. The payload stations in the orbiter cabin provide panel space for installing displays and controls unique to a specific payload. Standardized electrical interfaces are provided for payload power, monitoring, command, and control. Standard wire harnesses are provided for use in the payload specialist station and in the payload bay for connecting the standard avionics and electrical services to four payloads. Payloads can avail themselves of the nonstandard avionics services by use of nonstandard harnesses, which require user funding.

The design of the orbiter-to-payload avionics interfaces and accommodations has been optimized to provide an efficient and flexible system that will accommodate mixed cargo flight requirements and will require minimum changes between flights. This highly versatile avionics/electrical system for payload support, together with the flexibility of the

Shuttle system, will offer foreign as well as national space programs an effective, viable system for future space activities.

Figure 1. Standard Avionics Provisions for Payloads - Functional Diagram

Figure 2. Payload Command System Interface Block Diagram

Figure 3. Standard Switch Panel Front Panel Layout

Figure 4. Payload Data System Interface Block Diagram

Figure 5. Payload Data Interleaver Block Diagram

Figure 6. Payload Caution and Warning System Interface

Figure 7. Aft Flight Deck

Figure 8. Payload Electrical Power Distribution System Interface

Figure 9. Standard Payload Cables for Shared Cargo Showing Services for One of Four Users

Figure 10. Standard Mixed Cargo Harnesses

Figure 11. Nonstandard Avionics Services, Payload System

Figure 12. Nonstandard Avionics Services, Mission Station



G. K. Huth and S. Udalov


9841 Airport Blvd., Suite 912

Los Angeles, California 90045


The Inertial Upper Stage (IUS) system is intended for DOD and NASA space vehicle

(S/V) launches from the Shuttle Orbiter. Prior to and during the IUS launch, commands must be transmitted to the S/V and IUS and the telemetry data must be communicated to the Orbiter. This paper describes the communication links between the IUS and the

Orbiter, with particular emphasis on the RF link equipment and performance. The transponder equipment carried on board the IUS system is also described.


The Inertial Upper Stage (IUS) system is a part of the Space Transportation System

(STS). The IUS is intended for carrying out both the DOD missions and the NASA missions. The STS missions involving the IUS include (1) launching payloads into earth orbits and planetary trajectories, (2) recovering payloads from low-earth orbits, and (3) performing other space missions as required.

Because of the versatility of the IUS system, the telemetry, tracking and command

(TT&C) subsystems carried by the IUS must be compatible with both DOD and NASA formats. This applies to both the IUS/Orbiter links and to IUS/ground links.

Consequently, the avonics equipment carried by the IUS and on board the Orbiter must be compatible with both DOD SGLS and NASA STDN/TDRS systems.

It is the purpose of this paper, therefore, to provide the description of the IUS/Orbiter radio link with specific emphasis on the TT&C equipment carried by the Orbiter for the purpose of communicating with the IUS during both the DOD and NASA missions. The corresponding equipment carried by the IUS is also described in this paper.

* This work was performed by Axiomatix for NASA Johnson Space Center, Houston, Texas, under

Contract NAS9-15409C.


The block diagram representation of the IUS/Shuttle Orbiter avionics subsystems is shown in Figure 1. This figure applies to both the hardline and RF links. The Orbiter avionics equipment shown in Figure 1 can be logically subdivided into four categories.

The first category includes the equipment which performs payload RF and baseband signal processing functions. This category includes the Payload Interrogator (PI), Payload

Signal Processor (PSP), Communication Interface Unit (CIU), and Ku-Band Signal

Processor (KuSP). The second category consists of the equipment which handles and processes the IUS commands and telemetry on the actual data format level. Included in this category are the Payload Data Interleaver (PDI), PCM Master Unit (PCMMU),

Network Signal Processor (NSP), and General-Purpose Computer (GPC). The third category includes the radio equipment which is used for communicating the IUS data to and from the ground. The Ku-Band Communication Transmitter/Receiver, the S-Band

Network Transponder and the FM S-Band Transmitter belong in this category. Into the fourth category we have arbitrarily included the support equipment such as payload recorder, Mutiplexer/ Demultiplexer units (MDM’s) and the associated interface equipment.

The IUS avionics equipment consists of either an SGLS transponder for the DOD missions or an STDN/TDRS transponder for the NASA missions. The hardline signal processing equipment on board the IUS provides for the communication with the Orbiter during the attached phase.

The major emphasis of this paper is on the RF communications between the Orbiter and the IUS when the latter is in the detached configuration. The attached mode is discussed only briefly for the purpose of completeness.

Therefore, from the standpoint of the avionics system description, we will concentrate on the Shuttle avionics equipment in the first category and the SGSL as well as STDN/TDRS transponders on board the IUS. The performance of the RF link will be described, therefore, in terms of this equipment.

A. Attached IUS Communication

In the attached mode, a hard line (umbilical) provides two-way communication between the IUS and the Orbiter. Scientific data, engineering data, guidance, navigation and attitude control data (GN&C) are received by the Orbiter from the IUS. Alternately, command data, GN&C, and uplink data are transmitted to the IUS from the Orbiter.

For the scientific data, only limited processing (i.e., as required to throughput data to a ground terminal is provided for IUS medium-band and wideband data inputs (inputs in the range of 16-256 kbps). For data rates below 64 kbps, the scientific data can be routed through the PDT to the PCMMU, where it is made available to the general-purpose computers (GPC) for processing and on-board display. A payload specialist crew member may then interface directly with a specific experiment, as required. Medium-band scientific data is routed to the receiving ground terminal via either the S-band FM link or the Ku-band system, as follows:

(1) S-band FM

Analog: 300 Hz - 4 MHz or

Digital: 200 bps - 5 Mbps NRZ-L, or

200 bps - 2 Mbps biphase-L

(2) Ku-band

Analog: DC - 4.5 MHz BW plus or

Digital: 16 kbps - 1024 Mbps biphase-L

16 kbps - 2 Mbps NRZ-L, M or S.

The Ku-band wideband analog channel input (DC - 4.5 MHz) can be used by the IUS or

CIU for analog telemetry as a transparent throughput channel, which provides flexibility and minimum Orbiter processing. Capability is constrained only by the KuSP bandwidth.

The PDT provides the capability to receive engineering data from up to five attached payloads simultaneously. The PDT then decommutates up to four of these inputs and provides time-tagged, time-homogenous data from these four payloads simultaneously to the Orbiter data processing subsystem (DPS) for on-board display and/or transmission to the ground via OI downlink. The throughput data rate (composite PDI output to the

PCMMU) is limited to 64 kbps maximum on-orbit and 5 kbps for ascent.

A capability for direct recording of certain types of payload data is also provided on board the Orbiter. The payload recorder has 14 tracks capable of serial or parallel recording of digital and analog data. Data rates from 25.5 kbps to 1.024 Mbps and analog data of 1.9 kHz to 2 MHz may be recorded. A minimum record time of 56 minutes is provided at the maximum data rate. Simultaneous analog/digital parallel recording is limited to the first record pass. Subsequent passes are restricted to sequential singlechannel digital records. A total of 14 tape speeds (four per mission) are available and selectable by on-board or ground control.

Guidance, navigation and attitude control services are provided for the IUS by the CIU or

PSP using an MDM/GPC/MDM interface. Over this interface, the Orbiter provides state vector update data words to the IUS. The CIU transmits the Orbiter state vector data to the IUS using the SGLS command format of ternary frequency-shift-keying (FSK) with

“S” tones of 65 kHz, “0” tones of 76 kHz, and “1” tones of 95 kHz. The PSP transmits the Orbiter state vector data to the NASA IUS on a 16-kHz sine wave subcarrier at a binary command data rate of 2 kbps.

B. Detached IUS Communication

The basic low rate data-processing/display services provided for the attached IUS are also provided for detached or deployed IUS via an S-band RF communications link between the IUS and the Orbiter. Figure 1 shows the interfacing hardware that supports this link. Note that, when a spacecraft is launched by the IUS, the spacecraft communicates only in the attached mode through the IUS, as indicated in Figure 1. Also note that the PI cannot communicate with the IUS and the spacecraft simultaneously. The

Orbiter S-band transceiver (PI) supporting RF communications with detached payloads is compatible frequency-wise with STDN, SGLS, and DSN-compatible payloads. The PI is capable of operating at approximately 850 selectable frequencies in the 2200-2300 MHz range.

Telemetry signals in the Orbiter standard mode of operation are routed from the PI, after carrier demodulation, to the PSP or CIU, where the data is demodulated off of a

1.024 MHz subcarrier (and a 1.7 MHz subcarrier by the CIU). The data is then routed to the PDI/PCMMU/GPC for decommutation processing, display and downlinking in the same manner as the attached IUS or payload.

Data rates that can be accommodated by the PSP in the standard mode are 16,8 4,2 and 1 kbps. Processing of 16 kbps may be provided by the CIU. When the CIU is used for

DOD missions, the PSP is bypassed.


A. Payload Interrogator (PI)

The function of the PI is to provide the RF communication link between the Orbiter and detached payloads. For communication with the NASA payloads and the NASA IUS, the

PI operates in conjunction with the PSP. During DOD missions, the PI is interfaced with the CIU. Nonstandard (bent-pipe) data received by the PI from either NASA or DOD payloads is delivered to the KuSP, where it is processed for transmission to the ground via the Shuttle/TDRSS link.

Simultaneous RF transmission and reception is the primary mode of PI operation with

NASA IUS, DOD IUS, and payloads. The Orbiter-to-payload link carries the commands while the payload-to-Orbiter link communicates the telemetry data. In addition to this duplex operation, the PI provides the “transmit only” and “receive only” modes of communication with some payloads.

Figure 2 shows the functional block diagram for the PI. The antenna connects to an input/ output RF port which is common to the receiver and transmitter of the PI unit. Because of a requirement to operate the PI simultaneously with the Shuttle/ground S-band network transponder which radiates and receives on the same frequency bands, a dual triplexer is employed.

The receiver frequency and phase-tracking loop begins at the second mixer. As shown in

Figure 2, the output of the first IF amplifier is down-converted to the second IF as a result of mixing with a variable second LO frequency, f


. Demodulation to baseband of the second IF signal is accomplished by mixing with a reference frequency, f


. The output of the tracking phase detector, after proper filtering, is applied to the control terminals of a

VCO which provides the second local oscillator signal, thereby closing the tracking loop.

For the purpose of frequency acquisition, the local oscillator frequency may be swept over a ±80 kHz (minimum) uncertainty region. Sweep is terminated when the output of a coherent amplitude detector (CAD) exceeds a preset threshold, indicating that the carrier tracking loop has attained lock. The output of the CAD also provides the AGC to the first

IF amplifier.

A wideband phase detector is used to demodulate the telemetry signals from the carrier.

The output of this detector is filtered, envelope level controlled, and buffered for delivery to the PSP, CIU, and KuSP.

The PI receiver frequency synthesizer provides the tunable first LO frequency and the corresponding exciter frequency to the transmitter synthesizer. It also delivers a reference signal to the transmitter phase modulator. Baseband NASA or DOD commmand signals modulate the phase of this reference signal which, in turn, is supplied to the transmitter synthesizer. There it is upconverted to either the NASA or DOD transmit frequency and applied to the power amplifier.

B. Payload Signal Processor (PSP)

The PSP performs the following functions: (1) it modulates NASA IUS and payload commands onto a 16 kHz sinusoidal subcarrier and delivers the resultant signal to the PI and attached payload umbilical, (2) it demodulates the NASA IUS and payload telemetry

data from the 1.024 MHz subcarrier signal provided by the PI, and (3) it performs bit and frame synchronization of demodulated telemetry data and delivers this data and its clock to the PDI. The PSP also transmits status messages to the Orbiter GPC; the status messages allow the GPC to control and configure the PSP and validate command messages prior to transmission.

The functional block diagram for the PSP is shown in Figure 3. The PSP configuration and payload command data are input to the PSP via a bidirectional serial interface.

Transfer of data in either direction is initiated by discrete control signals. Data words 20 bits in length (16 information, 1 parity, 3 synchronization) are transferred across the bidirectional interface at a burst rate of 1 Mbps, and the serial words received by the PSP are applied to word validation logic which examines their structure. Failure of the incoming message to pass a validation test results in a request for a repeat of the message from the GPC.

Command data is further processed and validated as to content and the number of command words. The function of the command buffers is to perform data rate conversion from the 1 Mbps bursts to one of the selected standard command rates (see

Table 1). Command rate and format are specified through the configuration message control subunit.

From the message buffers, the command bits are fed via the idle pattern selector and generator to the subcarrier biphase modulator. The idle pattern (which, in many cases, consists of alternating “ones” and “zeros”) precedes the actual command word, and is usually also transmitted in lieu of command messages. Subcarrier modulation is PSK

NRZ-L only.

The 1.024 MHz telemetry subcarrier from the PI is applied to the PSK subcarrier demodulator. The demodulated bit stream is input to the bit synchronizer subunit, where a

DTTL bit synchronization loop provides timing to an integrate-and-dump matched filter.

This filter optimally detects and reclocks the telemetry data. From the frame synchronizer, the telemetry data with corrected frame synchronization words and clock are fed to the


C. Communication Interface Unit (CIU)

The primary function of the CIU is to provide command and telemetry data conditioning between the Orbiter and the IUS transponder. The CIU consists physically of four boxes and two control panels mounted in an Orbiter standard console. The CIU accepts command data from one of five sources, as follows:

1. S-band MDM

2. KuSP


4. Crew-generated data from control panel

5. T-0 umbilical.

Input command data is validated, formatted, modulated on an SGLS baseband carrier

(see Figure 4A) at 1 k baud, and forwarded to one of six destinations. In the attached mode, the CIU forwards the conditional command data directly over hard line to one of two redundant IUS transponders on one of two IUS’s in the Orbiter payload bay. In the detached mode, the CIU forwards the conditioned command data to one of two redundant PI’s for RF transmission to the IUS transponder.

The CIU receives IUS telemetry over hard line (attached) and from the PI (detached). In the attached or hard-line mode, the CIU receives data from one of two IUS’s and provides selected telemetry data (NRZ-L) to the COMSEC and the PDI. The CIU provides the same telemetry data after NRZ-L to biphase-L conversion for selection to the

Payload Recorder (PR), FMSP or KuSP. The CIU also receives NRZ-L data from the

Wideband Data Interleaver (WBDI) on the CIU and performs NRZ-L to biphase-L conversion. The WBDI data is selected to be supplied to the PR, FMSP or KuSP. The

IUS EMU analog environmental data is received by the CIU for selection to the PR. In the detached or RF mode, the CIU receives telemetry data from one of the two PI’s as a

PSK subcarrier (1.024 MHz) frequency multiplexed with FM/FM environmental data on a

1.7 MHz subcarrier. The CIU performs PSK demodulation and bit synchronization to generate NRZ-L telemetry data and clock to be supplied to the PDI. The same telemetry data is NRZ-L to biphase-L converted to selection to the PR, FMSP or KuSP. The CIU performs FM demodulation on the 1.7 MHz subcarrier to generate three-channel FM (16,

24 and 32 kHz). The CIU provides the three-channel FM plus a 100kHz reference for selection to the PR.

Figure 4B shows a simplified block diagram of the CIU. Microprocessor technology is fundamental to the CIU operation. The microprocessor performs the bit synchronization function on the telemetry data for processing by a COMSEC and receives telemetry data

(NRZ-L) and clock from the COMSEC. The microprocessor performs frame synchronization, VCC extraction (required for DOD commands), command authentication, and determines command rejection. The microprocessor also accepts

GN&C data and provides the command generator function to send GN&C or crewgenerated command data to the FM/AM modulator. The required binary-to-ternary conversion on the command data is also performed by the microprocessor. Additional functions performed by the microprocessor are CIU mode control and status display.

D. Ku-Band Signal Processor (KuSP)

The KuSP receives IUS and payload data from the PI, PSP, CIU, PR, operational recorder (OR), and attached payload interface (API). Similarly, the KuSP transmits data to the IUS and payload via the CIU or NSP/GPC/PSP (or CIU). Table 2 presents the characteristics of the data that are handled by the KuSP. The 216 kbps data shown for the forward link originates at the TDRSS ground station and can be 72 kbps command data to the NSP, 128 kbps DOD command data to the CIU, 128 kbps text and graphics data and 216 kbps data containing 72 kbps command data plus digital voice data that is sent directly to the NSP.

Figure 5 illustrates the functional processing of the KuSP for data to be transmitted to the

IUS and payload (i.e., the forward link). When the forward link contains the normal

S-band 216 kbps operational data of the 72 kbps command data plus digital voice data, the data mode select is set to transfer the data directly to NSP1 and NSP2 without any processing in the KuSP. Note that, in this data select position, the possible data rates are

32, 72, 96 and 216 kbps. When the 216 kbps forward link data contains either text and graphics data or DOD command data, then data mode select is set to transfer the 72 kbps command data to NSP1 and NSP2. The 128 kbps DOD command data is actually 2 kbps which has been coded to use the available 128 kbps data rate without having to modify the KuSP bit synchronizer or frame synchronizer design.

The characteristics of the data that must be processed by the KuSP on the return link are quite varied, as shown in Table 2. The return link is transmitted in one or two modes which are identified by the type of carrier modulation utilized. Mode 1 implements unbalanced quadriphase-shift-keying (UQPSK) while Mode 2 implements FM. In both modes of operation, two of the channels (1 and 2) UQPSK modulate a subcarrier. Mode

1 utilizes this modulated subcarrier along with the third channel to UQPSK the carrier, as shown in Figure 6. Mode 2 linearly sums the modulated subcarrier with the third channel and frequency modulates the carrier with the resultant summed signal, as shown in

Figure 7.

Channel 1 always (Modes 1 and 2) carries the operations data of 192 kbps consisting of

128 kbps telemetry data and two 32 kbps delta-modulated voice channels. Similarly, the data on Channel 2 does not change from Mode 1 to Mode 2. Channel 2 carries the output from the PR, the OR, and the PSP as well as low rate data for the API and narrowband bent-pipe data from the PI. The range of data rates handled by the KuSP Channel 2 is shown in Table 2 to be 16-1024 kbps Manchester-coded data, 16-2000 kbps NRZ-coded data or DC-2 MHz analog bent-pipe data.

The data carried on Channel 3 in Mode 1 is digital data of 2-50 Mbps (NRZ) which is rate

1/2-constraint length 7, convolutionally encoded by the KuSP to maintain adequate performance margin at bit error probability of 10


. Because the output data rate of the convolutional encoder is twice the input, the input data clock must be doubled by the



Two Orbiter/IUS communication configurations will be used for the DOD and NASA

IUS missions. Operational constraints, however, may require the use of a DOD IUS for

NASA payload missions such as the TDRS launch. The DOD IUS uses the SGLS transponder for communications with the Orbiter. Alternately, the NASA IUS uses the

STDN/TRDRS transponder in the STDN mode for communications with the Orbiter.

A. IUS SGLS Transponder

The telemetry, tracking and command (TT&C) SGLS transponder acquires and tracks, with a phase-locked loop, an incoming S-band signal and provides demodulated spacecraft commands to the decoder. The transponder also receives data and telemetry from the spacecraft and phase modulates this information and the internally demodulated ranging tones onto an S-band 3W carrier which is provided to the antenna for downlink transmission.

The transponder shown in a functional block diagram, Figure 8, is a single unit consisting of an S-band receiver, transmitter and the auxiliary circuitry. This transponder configuration performs the following functions:

Searches and acquires an SGLS-compatible S-band signal with modulation

Provides a coherent link, when in the VCXO mode, with a fixed 256/205 transmit-toreceive frequency ratio

Provides a noncoherent stable return link signal when in the auxiliary oscillator mode

Receives, demodulates command signals, outputting commands and clock signals

Receives, demodulates to baseband, and remodulates ranging signals on the return link carrier to provide coherent turn-around ranging

Accepts, modulates and transmits various analog and digital telemetry data on the return link

Provides telemetry outputs of key transponder parameters and operational status of the transponder

• Operates in the receive and transmit modes independently by way of having separate dc-to-dc converters.

The receiver utilizes a dual downconversion, fully phase-coherent design, incorporating a second-order phase-lock loop. S-band input signals in the frequency range of 1763-1840

MHz are amplified in a low noise preamplifier before downconversion to a first IF frequency of approximately 44 MHz. Amplification, gain control and bandwidth limiting are accomplished in the first IF circutis before further downconversion to 12.515 MHz.

Then the signal is further amplified and sent to the demodulator module circuits. Here four functions are performed:

Acquisition. Operates in conjunction with the discriminator module to acquire an

SGLS signal (including modulation).

• Phase detection. A predetection filter (30 kHz crystal filter) reduces the noise spectrum before phase detection takes place in the carrier tracking phase lock loop.

Loop bandwidth (B


) is 2 kHz.

• Coherent amplitude detection. Another detector, using a 90° phase-shifted reference, produces an output proportional to the RF carrier amplitude. This output forms the correction signal in the automatic gain control (AGC) loop and signal strength information for telemetry.

• Wideband detection. A wideband phase detector/frequency discriminator which, unlike the above two detectors, is not preceded by a narrowband filter, is used to demodulate the phase modulation from the uplink carrier. The output of this demodulator provides the wideband data output to the baseband circuits where filtering separates ranging and command data. The discriminator function of this module does not allow the receiver to acquire to a sideband and, upon carrier acquisition, commands the sweep off in the demodulator. The module supplies the command tones to the FSK tone demodulator and the ranging information to the baseband conditioner.

The S-band transmitter operates from an internal auxiliary oscillator in the noncoherent mode, or from a VCXO output provided by the receiver (coherent mode). Selection of the source can be determined by command or will be automatically set by the phase lock status of the receiver. Both sources are at a frequency of 2f

Frequency multipliers utilizing SAW filters increase the output frequency to S-band at

256 f o o

, approximately 17.5 MHz.

. Phase modulation is performed at 1/4 the output frequency, approximately

560 MHz. The modulator utilizes a quadrature hybrid terminated in voltage variable reactances to achieve linear phase modulation.

Digital telemetry and data are biphase modulated on 1.024 MHz subcarrier while the analog data is FM modulated onto a 1.7 MHz subcarrier. These signals are summed with the turnaround ranging tones before they are provided to the linear-phase modulator. An option is available to replace the 1.7 MHz FM subcarrier with a 1.7 MHz biphase modulated subcarrier. The transmitter also provides variable modulation indices of the subcarrier automatically when either of the subcarriers are commanded off.

The S-band power amplifiers are wideband circuits culminating in a circulator protecting the 2.5 to 3.5W output from shorted or open loads. A separate high efficiency dc-dc power converter provides operating power to the transmitter upon command.

B. IUS STDN/TDRS Transponder

The STDN/TDRS transponder isa multimode device capable of receiving and transmitting signals compatible with both the STDN and TDRS operational modes and signal formats.

An abbreviated block diagram of the transponder is shown in Figure 9. A summary of transponder functions is given below:

• Provides two-way coherent communications with the Orbiter, STDN ground station or TDRS satellites a the appropriate S-band frequency

• Transponds with a coherency ratio of 240/221

• Acquires and demodulates STDN signals in 1/2 second ($ -117 dBm) when commanded to the STDN ONLY fast acquisition mode

Recognizes, acquires and demodulates either STDN or TDRS signals when commanded to the DUAL mode; transponder recognizes signals based on the signal structure rather than on the signal level

Incorporates a command decoder unit which demodulates the STDN subcarrier and recovers the 2 kbps clock and data signals in either mode; a data squelch circuit based on the measurement of E


/N o

is also included

Removes the 3 Mbps spread spectrum code from the TDRS command channel signal and recovers the range synchronization from the TDRS spread spectrum range channel signal

• QPSK or PM modulates the coherent S-band transmit carrier with telemetry and ranging data; in TDRS, the carrier is also coherently spread at about 3 Mbps

Provides noncoherent telemetry transmission in the absence of received signals or when commanded.

The basic transponder configuration is the same for both the STDN and TDRS modes; that is, both are configured to utilize the same frequency plan, receiver and transmitter RF and IF modules. When communicating with the Orbiter, the transponder is configured to the STDN mode.

The received signal is amplified by a low noise preamplifier prior to a first downconversion to approximately 47 MHz. A second coherent downconversion brings the signal to the second IF (F


= 13.8 MHz); due to the design of the frequency plan, this second IF operates at a fixed frequency regardless of the input frequency. The second IF signal is divided three ways and is simultaneously sent to the discriminator, the carrier

(coherent) demodulator and the spread spectrum processor. The discriminator is employed as an aid to fast acquisition of modulated STDN signals; its sole purpose is to prevent the receiver from locking onto the STDN 16 kHz subcarrier.

The coherent demodulator is employed to recover the data signals which are modulated onto the received carrier. It can be configured as either a linear demodulator (STDN signals) or a Costas PSK demodulator (TDRS signals). Demodulation also generates the phase error signal used to lock up the receiver VCXO; it also generates the coherent amplitude detector (CAD) signal which indicates receiver lock. The TDRS signals are demodulated using a conventional linear phase lock loop.

In the normal configuration, the demodulated data (baseband) signals are sent to the command detector unit. This unit contains a subcarrier demodulator and a bit synchronizer. The subcarrier demodulator recovers the 2 kbps data from the STDN 16 kHz PSK subcarrier; it employs a frequency-doubling subcarrier recovery loop. The bit synchronizer is an advanced design capable of recovering and reclocking data at three different selectable data rates (2 kbps, 2 kbps/N and 2 kbps/M), where N and M may be an integer from one to 16. Only 2 kbps is used for the IUS program.

The bit synchronizer also includes a combined squelch and bit synchronization lock circuit which is insensitive to bit transition density. The outputs of the CDU are the data and clock and an ACTIVATE (desquelch) signal. Upon command, the CDU is also capable of demodulating an auxiliary 16 kHz PSK subcarrier signal.

The return link signal is provided by the transmitter, shown in the lower portion of

Figure 9. The STDN mode service consists of a 1.024 MHz PSK subcarrier for digital telemetry, a 1.7 MHz subcarrier for analog telemetry, and the ranging signal. These signals are assembled at baseband in the baseband conditioner unit and are used to modulate a

linear phase modulator operating at 1/4 the output frequency. This signal is frequency multiplied X4; in the STDN mode, the inputs to the digital modulator are held to a logical one. The signal is amplified to 3.0 watts nominal and passed through an isolator to the output.

One of the salient features of the STDN/TDRS transponder is the capability to automatically identify the format of the received signal (i.e., STDN versus TDRS) and to configure the transponder to the appropriate mode. For example, if an STDN (or Orbiter) signal appears and is recognized, the receiver VCXO select switch is switched directly to allow the carrier loop to commence tracking. In either event, the receiver control algorithm configures the command detector unit (CDU) to match the received signal (TDRS or

STDN), and also sends the lock-up information to the transmitter control algorithm circuit. The transmitter is then reconfigured and switched to coherent, assuming that the external TXR command is COHERENT. In the TDRS mode, this does not occur until long (range) code lockup is verified.

In the STDN (and Orbiter) mode only, the requirement is to correctly acquire an STDN signal modulated with a 16 kHz command subcarrier (but no range tones) and to achieve this within 1/2 second at signal levels $ -117 dBm. The STDN signal may be swept or it may be stationary. In this mode, the receiver VCXO select switch is set to permit the receiver to sweep ±150 kHz every 300 ms. The demodulator is configured for linear. A true lock indication is declared when both of the following conditions are achieved:

• The demodulator CAD indicates the presence of a CW signal within bandwidth

(±4 kHz)

The discrminator indicates that this CW signal is not a sideband

The discriminator is thus used to prevent false lock on the STDN (or Orbiter) subcarrier.

It is equipped with a dual (±) threshold which indicates positive if the incoming carrier is

±16 kHz from the true receiver center frequency, as it would be if the receiver was attempting to lock on to a subcarrier sideband.

If both of the above conditions are satisifed, the receiver select switch is positioned so that coherent tracking commences. When tracking is verified, the carrier loop bandwidth is narrowed to 800 Hz to prevent tracking out the 4 kHz range tone when it is turned on.


One of the requirements for the RF link between the IUS and the Orbiter is that the link be operational from a minimum range equivalent to IUS being in the payload bay of the

Orbiter to a range of 10 nmi. The resulting wide range in the RF signal levels requires special considerations for protecting the STDN/TDRS and the SGLS transponder when they are in the payload bay. Conversely, the PI receiver must also be protected. The considerations pertaining to the short-range operation are discussed below.

A. Payload Interrogator (PI) Transmitter to the IUS Receiver Link

The transmitter for the PI has the capability of being switched to a power output compatible with the distance to the IUS. The power steps are 5W, 0.5W, and 0.0025W.

Since the cable loss is of the order of 9.8 dB, the payload antenna EIRP is +29.7 dBm,

+19.7 dBm, and -3.3 dBm. For the case of the IUS in the payload bay, the output power should be set at the lowest power level (-3.3 dBm) to avoid the possibility of saturating or damaging any of the IUS SGLS or STDN/TDRS transponders. At this level, the received power at the SGLS receiver is -56.8 dBm, which is within the operating dynamic range of threshold (approximately -120 dBm) to -40 dBm maximum. However, the other two output power levels are -23.8 and -33.8 dBm, which exceed the upper limit of the SGLS transponder specified dynamic range. Similar examination of the received signal power for the NASA IUS with the STDN/TDRS transponder shows that the lowest PI output power level must be used so that the IUS received power (-52.5 dBm) is less than -40 dBm (i.e., the upper limit of the specified dynamic range for the STDN/TDRS transponder).

B. IUS Transmitter to Payload Interrogator Receiver Link

No adjustments are possible for controlling the output power of the IUS transmitter so that the 20W (13 dBW) from the TWT results in 13.2W (11.2 dBW) being radiated into the bay. Above -20 dBm, however, the receiver IF amplifier circuits begin to saturate.

Although this would not adversely affect demodulation of constant envelope signals, it may cause receiver false-lock under certain conditions. Therefore, receiver operation with signals higher than above -20 dBm is not recommended. At input signal levels of +10 dBm and higher, a preamplifier protective diode breakdown limiter becomes operative.

Purposeful receiver operation above +10 dBm is not recommended.

The PI received power from the IUS SGLS transponder at 3 m is estimated at -19.4 dBm.

Thus, the receiver IF amplifier will begin to saturate. Furthermore, this estimate of the received power was calculated based on a -4.0 dB IUS pointing loss (worst case). If there is no point pointing loss, the received power is -15.4 dBm and the PI IF amplifier could be in saturation but there will be no damage to the PI receiver. Similarly, the PI received power from the IUS STDN/TDRS transponder at 3 m is estimated to be at -13.6 dBm including a -3.4 dB IUS pointing loss. Therefore, when at close range, the antenna selection on IUS will be such as to keep the PI receiver out of saturation. Thus, there will be no damage to the PI receiver at close ranges.

C. IUS and Orbiter Received Power Versus Range

Figures 10 and 11 present, respectively, the IUS and Orbiter received power as a function of range. Figure 10 presents the received power at the IUS SGLS and STDN/TDRS transponders versus range for the three PI output power levels. The IUS transponder are specified to achieve acquisition at a received power of -117 dBm in 0.5 sec with a probability of 0.9 for a modulated signal. In terms of the Orbiter acquisition procedure, a modulated signal would be present only during reacquisition. Therefore, the reacquisition threshold for the SGLS transponder is reached at 5 nmi in the low-power mode, while the

STDN/TDRS transponder reaches the reacquisition threshold at 8.5 nmi in the low-power mode. In the medium-power mode, the SGLS and STDN/TDRS transponders reach the reacquisition threshold at 24 nmi and 36 nmi, respectively. Considering the range limitations due to performance, one determines the received power at which the system operates with zero margin. We call such points ZPMO in Figures 10 and 11. Thus, as seen from Figure 10, the ZPMO for the STDN transponder is outside the range for acquisition. Thus, the STDN range is reacquisition limited rather than performance limited.

For the SGLS, we have a ZPMO range of 14 nmi with medium-power PI transmission.

Therefore, for either transponder, the 10 nmi operating range requirement can be Mt with medium-power PI transmissions.

The power received by the PI versus the range to CIU is presented in Figure 11. As can be seen from the figure, the reacquisition threshold is far above the zero margin performance (ZPMO) for both the SGLS and STDN/TDRS transponder transmissions. It can also be seen from Figure 11 that the 10 nmi operating range requirement is met for both transponders in the digital telemetry mode.


The Inertial Upper Stage (IUS) system will be used by DOD and NASA for space vehicle

(S/V) launches from the Shuttle Orbiter. The communication equipment to support these launch missions consists of (1) Shuttle Orbiter on-board equipment, and (2) IUS onboard transponders compatible with either SGLS (DOD) or STDN (NASA) signal formats. This equipment has, as described in our paper, sufficient capability to provide two-way transponded communication for an IUS/Orbiter range of 10 mni (minimum) for commands and digital telemetry modes.


1. Carrier, L. M., and Pope, W. S., “An Overview of the Space Shuttle Orbiter

Communication and Tracking System,” IEEE Trans on Communications, Vol.

COM-26, No. 11, Special Issue on Space Shuttle Communication and Tracking,

November 1978, pp 1494-1506.

2. Teasdale, W. E., “Space Shuttle Payloads and Data Handling Accommodations,”

IEEE Trans on Communication, Special Issue on Space Shuttle Communication and

Tracking, Nov. 1978, pp 1557-1567.

3. Springett, J. S., and Udalov, S., “Communication with Shuttle Payloads,” IEEE

Trans on Communication, Special Issue on Shuttle Communication, November

1978, pp 1584-1594.

4. Batson, B. H., Teasdale, W. E., and Huth, G. K., “Payload Data Processing for the

Space Shuttle Program,” NTC ’77 Conference Record, Vol. 2, Los Angeles,

December 1977.

5. Batson, B. H., and Moorehead, R. W., “The Space Shuttle Orbiter

Telecommunication System,” IEEE Communication Society Digest of News, Vol.

14, No. 3, May 1976.

6. Batson, B. H., and Johnson, J. H., “Space Shuttle Communications and Tracking

System,” ITC Conference Record, Los Angeles, October 1974.

7. “Space Shuttle System Payload Accommodations,” Level II Program Definition and

Requirements, NASA Document JSC 07700, Vol. XIV, Ref. D, Houston, Texas,

November 1975.

8. “Space Transportation System User Handbook,” NASA JSC, July 1977.

9. Huth, G. K., “Orbiter CIU/IUS Communications Hardware Evaluation,” Axiomatix

Report No. R7906-8, June 29, 1979.

Table 1. NASA Command System Parameters

Subcarrier Frequency

Bit Rates

E b



for P b e

= 1 x 10


16 kHz, sinewave

2000 ÷ 2 N bps, N= 0,1,2,...,8

10.5 dB

Table 2. Ku-Band Signal Processor Data Characteristics

Processor Interface


Operations Data - NSP(1,2)

Command/Text & Graphics - NSP

(1,2) and Text & Graphics

Command/DOD Payload Command

Data - NSP(1,2)/CIU





Rate or Bandwidth

32,72,96,219 kbps (Manchester)

72 kbps Command

128 kbps Text & Graphics

16 kbps Frame Sync (Manchester)

72 kbps Command

128 kbps DOD Payload

16 kbps Frame Sync (Manchester)



Operations Data - NSP(1,2)


Payload Recorder (PR)

Operations Recorder (OR)

Payload low data rate - PSP

(1,2) or Attached Payload

Interface (API)

PI(1,2) low data rate

Digital 129 kbps (Manchester)





25.5-1024 kbps (Manchester)

25.5-1024 kbps (Manchester)

16-2000 kbps (NRZ)

16-1024 kbps (Manchester)

16-2000 kbps (NRZ)

16-1024 kbps (Manchester)

0-2 MHz


Attached Payload Interface



PI(1,2) high data rate

Attached Payload Interface

Video Interface Unit

Digital 2-50 Mbps (NRZ)




16-4000 kbps (NRZ)

0-4.5 MHz

16-4000 kbps (NRZ)

0-4.5 MHz

0-4.5 MHz

Figure 1. Inertial Upper Stage/Shuttle Orbiter Subsystems and Interfaces

Figure 2. Payload Interrogator Functional Block Diagram

Figure 3. NASA Payload Signal Processor Functional Block Diagram

Figure 4A. Command Tone Modulation Envelope

Figure 4B. Communication Interface Unit Functional Block Diagram

Figure 5. Ku-Band Signal Processor Forward Link Functional Block Diagram

Figure 6. Ku-Band MODE 1 Three-Channel Downlink Modulation

Figure 7. Ku-Band MODE 2 Three-Channel Downlink Modulation

Figure 8. SGLS Transponder Functional Block Diagram

Figure 9. IUS STDN/TDRS Transponder Functional Block Diagram

Figure 10. IUS Received Power Versus Range for Each PI Output Power Setting

Figure 11. PI Received Power Versus Range from IUS



Larry H. Kasulka

Unit Chief

Darrell D. Wilkinson

Engineer Scientist/Specialist

Data Handling Subsystems Data Handling Subsystems

McDonnell Douglas Astronautics Company McDonnell Douglas Astronautics Company

5301 Bolsa Avenue

Huntington Beach, California 92647

5301 Bolsa Avenue

Huntington Beach, California 92647


A new equipment command and data system (CDS), being developed by NASA for

Shuttle/Spacelab-payload users, takes advantage of the more liberal constraints being offered a Shuttle payload user (i.e., size, weight, and power) to save cost. This Spacelab

Payload Standard Modular Electronics (SPSME)


system provides an extremely flexible approach derived from the modular design of the CAMAC system. SPSME promises to provide a real cost savings to the space community due to its universal application to most

Spacelab experiments. NASA projections based on the mission model for the first 19

Spacelabs using SPSME versus the historical hardware results in approximately $125.4

million savings.


Before the Space Transportation System (STS) the risks and costs associated with delivering a payload into orbit justified unique design and development to extract the ultimate benefit from each ounce of payload put into orbit. With the STS a payload user

(1,2) no longer has this problem and can take advantage of designs which are less expensive and reusable. The Spacelab being built by European Space Agency (ESA) in cooperation with

NASA is a prime example of this; this STS payload provides a shirt-sleeve environment in space for experimentors to utilize like their labs on the ground. The Spacelab provides the experimentor standard racks (core module) or cold plates (pallet) for mounting his hardware (see Figure 1). Other host services include power distribution and control, thermal conditioning, and certain command and data handling functions. Figure 2 illustrates the experiment and its interface with the Spacelab CDMS. The command and data capabilities of the experiment must still be provided within its own subsystem and will

* SPSME is being build by McDonnell Douglas Astronautics Company for NASA Marshall Space

Flight Center on Contracts NAS8-33287 and NAS8-33468.

Figure 1. Space Shuttle

Figure 2. CDMS Functional Block Diagram

vary depending on the complexity of the experiment, but here are certain common elements for all — formatting data for the HRM and interfacing the experiment with the

RAU for transfer of data to and from the experiment computer. NASA lab


and determined that by taking existing CAMAC Ground Equipment Standards


and applying them to requirements for Spacelab hardware you could substantially reduce the cost of developing this hardware plus for ground applications the existing ground CAMAC hardware would be functionally and mechanically interchangeable (7) . In accordance with studies, implementing the above recommendations would reduce the development cost anywhere from 0.1 to 0.5 and recurring cost approximately 0.5 to 0.75 of equivalent space

hardware. NASA Headquarters realizing the potential savings this offered, made this approach a NASA Standard Hardware Item (8) and appointed MSFC to manage the definition and development. Space CAMAC hardware was named Spacelab Payload

Standard Modular Electronics (SPSME), and as well as being space CAMAC, would provide the necessary interfaces between the Spacelab CDMS and the experiment dedicated experiment processor (when required).


Since SPSME is based on CAMAC, a basic description of CAMAC is presented followed by a description of the major SPSME differences.

Basic CAMAC Architecture for Ground-Based Applications

Computer Automated Measurement and Control (CAMAC) is an internationally accepted measurement and interface system in use for ground applications in the United States and

Europe. In the United States the CAMAC standards have been adopted by the Institute of

Electrical and Electronic Engineers (IEEE) (6) .

Use of the CAMAC standards produces a system that offers significant advantages in system flexibility, reduces the need for special software, and minimizes system obsolescence. It has enjoyed widespread use in industrial, medical, and laboratory measurement and control. It has also been used extensively in the atomic energy field.

Since the CAMAC system deals with physical and electrical standards, it can assure physical and electrical compatibility between units with vastly different characteristics. It serves as a basis for interfacing a wide variety of transducers and other measurement and control devices.

Two typical configurations of ground-based systems using CAMAC are illustrated in

Figures 3 and 4. Figure 3 represents a fairly complex minicomputer-based instrumentation and control system, while Figure 4 illustrates a simpler system where the computational and control capability of a microcomputer is sufficient.

The basic building block of the CAMAC system is the module (Figure 5). The module is the interface between the CAMAC system and external elements which are sources or sinks for data and commands. The physical dimensions and the connections between the

CAMAC dataway and the modules are specified in detail in the IEEE Standards connector type, pin assignments, signal characteristics, and power supply voltages are also specified for the dataway connector. In addition, standard CAMAC commands to be decoded by the modules have been specified.

(6) . The

Figure 3. Typical CAMAC Application in Ground-Based Systems

Figure 4. Simple CAMAC Application in Ground-Based Systems

The size of a single-width module is 305 mm deep, 221.5 mm high, and 17 mm wide.

Other modules may be integer multiplex of this width. A module of twice this width is referred to as a double-width module. As long as the designer adheres to the physical and electrical specifications and properly decodes the standard CAMAC commands, he may place whatever circuitry he desired on the remainder of the module.

Figure 5. SPSME Module

The functional CAMAC modules interface with the CAMAC dataway, which is essentially the back plane or interconnection system of the container or CAMAC crate (Figure 6) for mounting the modules. The dataway distributes data, control and timing signals, and power between the modules, power supply, and module known as the crate controller. The

CAMAC crate designer may use any connection method that he wishes as long as he distributes the signals and power supply voltages and provides the specified connectors for the nodules as specified in the IEEE Standards.

The crate controller serves as the interface between the crate dataway and the CAMAC branch highway, a connection system external to the crate. The crate controller monitors the branch highway for commands addressed to it, decodes the crate and module portion of the command, and routes the command to the addressed module or modules. If required by the command, the crate controller will also transfer data between the dataway and the branch highway. The crate controller may be either a passive device that merely decodes signals or it may have considerable intelligence and contain a microprocessor or a special sequencer to do specific functions. Some commercially available crate controllers do not interface with the branch highway, but interface with a computer bus such as the PDP-11

Unibus. The crate controller usually utilizes module locations 24 and 25 with the other locations being available for other modules.

Figure 6. SPSM E Crate

The CAMAC branch highway serves as the control and data path between the CAMAC system and the system controller or digital computer. The branch highway driver serves as the computer I/O unit that drives the branch highway. In the parallel highway system, up to seven different crates may be addressed and the data and address are transmitted in parallel over the branch highway.

The signal level and signal sequencing for the branch highway are specified in detail in the

CAMAC specifications. The connector at the crate controller end of the branch highway, together with the number of wires in the branch highway cable, are also specified.

The format of the CAMAC command that is transmitted over the branch highway is given in the IEEE Standard. The command consists of an n-of-7 crate address, a 5-bit module address, a 4-bit module subaddress, and a 5-bit function code. The actions to be performed are defined in the specifications for each of the 32 function codes. The data word length is specified as 24 bits although words shorter may be used.

Provisions are also made for interrupts by means of the look-at-me (LAM) signals from the modules. Since only one line in the branch highway cable is allocated to interrupts or branch demands, some means must be provided in the software or hardware or both to locate the source of the interrupt. Provisions are available to test each module for the presence of an active LAM signal and to mask the LAM when desired

The crate itself mounts in a standard 19-inch rack and has up to 25 stations for plug-in modules on a pitch of 17.2 mm. The minimum height of a crate is 222.25 mm. The

ventilation unit or power supply may increase the height in integral increments of

44.45 mm. The width of the crate is 483 mm and the minimum depth is 360 mm with a recommended maximum depth of 525 mm.

The standard voltages called out in the CAMAC specifications are +6 VDC, -6 VDC,

+24 VDC, and -24 VDC. Since TTL logic devices require +5 VDC and ±15 VDC, these voltages are derived from the available voltages by inserting some series element, such as a resistor or voltage regulator on each module requiring these voltages. This wastes considerable power, but the additional power dissipation is generally not significant in ground-based equipment.

The following discussion will be concerned with the interface between the module and the

CAMAC dataway that is described in IEEE Standard 583-1975. This standard permits a maximum data word length of 24 bits. Data word lengths of less than this are permitted by the specification and are often used where 16 bits or other word lengths are more compatible with the computer or controller.

The module data output logic drivers that drive the dataway read lines must be the open collector type without a pull-up resistor and must be capable of sinking 16 mA at the down level. The CAMAC specification requires that the read line drivers be capable of performing the intrinsic OR function for data gated onto the dataway. The pull-up resistors for these signals are located in the crate controller.

The data receivers that accept data from the dataway write data lines may be a standard

TTL circuit. This circuit must not present more than one standard TTL load to the write data line. The current load must be no more than 1.6 mA when the signal is at the low state.

The dataway interface logic must be capable of accepting a unique N line that indicates that this module has been selected. The module does not gate data onto the dataway unless this line is active. The module also fully decodes the subaddress (A lines) and the function

(F lines). Fully decoded means that all of the F lines have been used in the function decode logic to determine the function. There are four signals to be decoded for the subaddresses and five signals for the function codes.

The four A lines allow partitioning the module into 16 functional areas, each of which has a group 1 and a group 2 register. Any one of the 32 functions shown in Table 1 called for by the five coded F lines may be performed in any of the 16 functional areas. The group 1 or 2 registers are selected by the functional code. Some of the functions make use of the dataway data lines. Those that do not may be used as control and test functions for the module.

Table 1. Definition of Standard CAMAC Dataway Function Codes

The dataway timing is shown in Figures 7 and 8. The busy signal serves to notify the module that a dataway operation is in progress. Data are clocked off the dataway at strobe

S1 time and data output to the dataway must also be ready by strobe S1 time. Data are permitted to change at strobe S2 time. Any registers that are cleared must be cleared at this time.

Figure 7. Timing for a CAMAC Dataway Command Operation

The signals Q and X indicate module responses to the commands received. The X signal, if a one, indicates acceptance of the command. An X signal of zero usually means that the module was not able to execute the command for such reasons as a module failure, the module is absent, the external equipment is not connected, or the module is not designed to execute the command. A Q signal of logic zero may mean that the register addressed does not exist or some other minor error. It may also indicate that the module is busy and cannot reply at the moment.

The LAM signals are available from modules to signal the system controller that a module needs attention or service. The LAM serves the function of an interrupt and may have several sources in one module. The LAM signal may be capable of being disabled and must have the capability of being tested by one of the CAMAC functions. The LAM signal must have the capability of being cleared.

Figure 8. Timing of a CAMAC Data Unaddressed Operation

The initialize signal (Z) is used to initialize the module during power up. Any LAMs that can be disabled must be disabled by this signal and all LAM status registers must be reset.

Any registers to which it is connected must be set to a predefined state. A similar signal is the clear signal (C). The designer is free to choose which registers and bistables are affected by this signal as he is with the inhibit signal (I). This signal will inhibit any feature to which it is connected.

Since there is no hand-shaking sequence between the module and the crate controller, any data transfer must be handled immediately by the module. Data to be gated onto the dataway read lines must be held in readiness at all times. Data accepted from the write lines into a register must be processed prior to the receipt of the next data word for that register. The time available depends upon the dataway cycle time (about 1 microsecond) and the speed to the system controller.

Comparison of SPSME Requirements and Basic CAMAC Systems

The variations between the SPSME configurations to be used for Spacelab payloads and the basic CAMAC configurations stem from four SPSME requirements:

1. The requirements to interface with the RAU for two way transfer of commands and data.

2. The requirements that the SPSME accepts and use timing information provided by the

SL/CDMS via the RAU.

3. The requirement to generate a set of data appropriately formatted to be compatible with ground data handling capabilities.

4. The requirement that the SPSME be readily adaptable to various dedicated experiment processors.

These four requirements will be discussed further below.

SPSME-TO-RAU Interface Considerations

The selected SPSME-RAU interface uses serial digital transfers over the RAU PCM command output and serial input channels.

The transfer of signals over the RAU is a command-response data bus operation under the direction of the Spacelab experiment computer (EC). This means that a data bus transaction transferring information to or from the experiment system through the RAU occurs only when requested by a specific command originating in the EC software or a related EC I/O microprogram. The data transfers between experiment system and RAU are not hand-shaking type operations; rather, the experiment hardware must, contain the appropriate buffers and logic to accept information from the RAU and/or provide the requested data within a time period in the range of a few microseconds.

To satisfy these requirements, the SPSME architectural structure contains a modular element called the remote acquisition unit interface (RAUI).

SPSME Timing Considerations

The RAU provides periodic 1,024 kHz and 4 Hz outputs to the experiment system plus

GMT time updates transmitted in a serial format via a RAU PCM command output channel. The experiment hardware must use this information to construct time tags of

appropriate resolution and accuracy to identify the time of occurrence of experiment events. It must then append this time of occurrence information to the scientific and engineering data which is to be used for post mission analysis and evaluation. In some cases this time information must provide the basis during postmission analysis to correlate the time associated with events or processes in other elements of Shuttle or Spacelab to the experiment data.

To satisfy these requirements the SPSME architectural structure includes a modular element called the time interface (TMI).

SPSME to HRM Interface Considerations

The primary route by which experiment data needed for postmission record flows to the ground is via the high rate multiplexer (HRM). This requires that the experiment electronics assemble the scientific and engineering data into a pulse code modulation

(PCM) format based on very specific PCM format rules. These standard format rules are necessary to facilitate the processing of various experiment data streams at the POCC and the GSFC data reduction center.

The SPSME modular element which supplies the output PCM data stream to the HRM is called the high rate multiplexer interface (HRMI). Selection of data samples, sync codes, etc., in the proper sequence for the PCM format is accomplished by programming of the system element which controls the dataway.

SPSNE to DEP Interface Considerations

The basic CAMAC architecture contains an element called the branch driver (BD). The

BD contains the logic, buffers, and drive circuitry which are necessary to translate the input/output word structure, timing, and protocol of a specific computer or controller to the word structure, timing, protocol, and line drive requirements of the branch highway. The

IEEE CAMAC specifications do not specify the BD design characteristics except that it must meet the operational and interface requirements of the branch highway.

As shown in Figure 9, one could describe a given BD design in terms of three functional areas as follows:

1. Computer interface logic

2. A control section

3. Branch highway interface logic

Figure 9. Functional Representation of CAMAC Branch Driver

The computer interface logic includes those circuits which interface with the data, address, and control lines of the computer input/output channel. It also includes “hand-shaking” logic or whatever is required to exchange data between the computer and a set of registers through which information is available to the BD control section. Clearly the design of the computer interface logic must be a unique design to match the I/O characteristics of the computer used in a specific CAMAC application.

The BD control section performs the sequencing and timing of functions required by the

CAMAC branch highway. To improve the efficiency of computer utilization, in a specific design it may also assist the computer by performing certain overhead functions such as identifying the source of LAM signals (interrupts) and maintaining word counts for multiple word transfers. The BD control section can be designed to be essentially independent, of computer selection. Some computers may have differences in word structure or block transfer protocol which could best be handled by changes in the control section. However, if one implements the BD control section with microprogrammed logic, these differences can be easily accommodated with firmware changes.

The branch highway interface logic includes circuits which provide the buffers, hand shaking logic, drivers, and receivers for data being transferred to and from the branch highway; clearly, this portion of the BD can be designed to be independent of computer selection.

SPSME Systems Architectural Configurations

A comparison of Figures 9 and 10 illustrates how the SPSME systems architecture functionally relates to the basic CAMAC branch driver configuration. The BD control section is functionally combined with the crate no. 1 controller to form the programmable crate controller (PCC) . This arrangement results in the basic, SPSME systems architecture shown in Figure 11. Note that crate No. 1 contains the PCC, while crate No. 2 and other crates use type A crate controllers. Mechanically, the DEP interface (DEPI) and the branch highway interface (BHI) are plug-in modules similar to the other SPSME modules; therefore, the physical arrangement is that shown in Figure 12.

Figure 10. Functional Representation of SPSME Interface with DEP

The PCC is a microprocessor-based design which, in the SPSME configuration shown in

Figure 1 performs the following functions:

1. In response lo instructions from the DEP, sequences command and data transfers between the DEP and SPSME modules.

2. Identifies the source of LAM signals within the crates and initiates the appropriate response. (A module called a LAM-grader may be optionally used to assist in this function.)

3. Implements a stored sequence of dataway commands for transfer of data to the


4. Implements a stored sequence of dataway commands for the update of data buffers in the HAUT.

Figure 11. Basic SPSME Systems Architecture

Figure 12. Physical Arrangement of SPSME Modules

Figure 13 shows a SPSME configuration which does not use a DEP. For this configuration, the DEPI is removed and the BHI is used only if more than one crate is needed. In the non-DEP configuration the PCC performs the following functions:

1. Implements the required sequence of dataway commands for transfer of data to the


2. Implements the required sequence of dataway commands for update of RAUI data buffers.

3. Accepts command from the EC (via the RAUI) and implements the proper response.

4. Identifies the source of LAM signals and initiates the appropriate response.

5. Performs other unique functions related to, the specific experimental process, such as experiment timeline sequencing.

Figure 13. Physical Arrangement of SPSME Modules for Non-DEP Application

A choice of the non-DEP configuration is feasible only if the experiment unique functions represented by (5) are simple enough to be implemented by the PCC capacity not utilized by (1) through (4).

Some SPSME applications may require functional modules similar to NIM modules which perform purely analog functions such as signal amplification. In the SPSME architecture, where it is feasible to package such functions in the standard SPSME modules, these functions may be operated in a SPSME crate alongside other SPSME modules. Normally, this type module connects only to power inputs via the dataway connector and both input and output connections are made via the front panel. In cases where it is not feasible to package NIM type functions in the SPSME modules because of noise characteristics or some other consideration, such functions will be defined as non-SPSDE electronic equipment and packaged as appropriate for the specific application.


NASA Marshall Space Flight Center has contracted for the following SPSME modules:

Common Use Modules

1. Programmable Crate Controller (PCC) - * JA008/C001 — Microprocessor-based crate controller used for sequencing dataway transactions and controlling the periodic transfer of data to the RAUI and the HRMI.

2. Auxiliary Memory Module - * JA008/C001 — The baseline design contains 12,288 words of 16-bit memory, allocated as 4096 words of RAM and 8195 words of

PROM. This unit is used to augment the memory capability of the PCC.

3. Remote Acquisition Unit Interface (RAUI) Module -


JA0o8/C002 — Provides interface between Spacelab CDMS and SPSME through the RAU serial PCM data/command channels.

4. High Rate Multiplexer Interface (MRMI) Module - * JA008/C003 — Interfaces the

SPSME to the HRM for downlink of science and engineering data.

5. Master Timing Unit Interface (TMI) Module -


JA008/C004 — Provides accurate data time-tagging (± 10 ms) and synchronous interrupts to the PCC for data formatting.

* NASA Specification Number

6. 13 and 25 Module Crates - * JA008 — 13 slot and 25 slot crates for cold plate (pallet) mounting. A conversion kit for rack mounting the 25 slot crate is also supplies.

Functional Modules

1. Peak Sensing Analog to Digital Converter - * JA008/F011 — 12 channel, 11 bit converter for use with pulse type measurements.

2. Analog to Digital Converter -


JA008/F001 — 32 channels with 10-bit resolution.

3. Customizing Module -


JA0o8/F012 — Blank module for use in customized development.

Additional modules specified by NASA-MSFC but not presently contracted for are as follows:

Functional Modules

1. Digital Input Register - levels.

* JA008/F002 — Dual, 24-bit input register with strobe/ handshake data transfer. Basic input signal level is TTL with options for other input

2. Digital Output Register -


JA008/F003 — Dual, 24-bit output register with strobe/ handshake data transfer. Basic output signal level is TTL with options for other input levels.

3. Digital to Analog Converter - * time< 5µsec.

JA008/F004 — 8-channel, 12-bit DAC with settling

4. Motor Controller -


JA008/F005 — Dual channel, programmable pulse train generator to control high power stepper motors and drive low power stepper motors.

5. Scanning Analog to Digital Converter - * JA008/F006 — 32-channel, 12-bit ADC system with medium conversion speec (< 1 ms/channel). Continuous scanning of channels with buffer memory containing the last conversion result for each channel.

6. Serial Input Register (Scaler) - *

16-bits capacity per channel.

JA008/F007 — 12-channel, 50 MHz scaler with

7. Serial Input/Output Register - * JA008/F008 — Sends and receives 16-bit serial data words at 1 MHz clock rate with handshake control of data transfer.

8. Relay Contact Output Register - * JA008/F009 — 12-channel output register with relay contacts (50 V at 0.5 A rating).

9. Isolate Input Gate -


JA008/F010 — 24-bit, individually isolated, input gate.

Common Use Modules

1. Branch Highway Interface — Standard CAMAC interface module to be used in multiple crate configuration allowing tip to seven crates to be tied together in parallel.

2. SPSME Crate Controller — In multiple crate configurations only one PCC is required in the master crate; additional crates operate under the control of standard CAMAC type A crate controller.

3. Look At Me (LAM) - Grader — The LAM Grader serves as a priority interrupt grader between various modules and the crate controller.

4. Dedicated Experiment Processor Interface (DEPI) — When interfacing a Dedicated

Experiment Processor with SPSME the DEPI provides the interface logic required to gairi acec?.ss to the SPSME bus in a compatible form.

NASA Goddard Space Flight Center has also developed requirements for SPSME modules which have not been contracted for; however, some have actually been built by GSFC

(9,10) and flown on balloon flights. Following is a preliminary list, of GSFC modules.

Common Use Modules

1. High Rate Multiplexer Interface Buffer Module - SSPP/C020 — Provides for up to

16-Mbps input from external source with dataway access for annotation and control with output to the high-rate multiplexer.

2. High Rate Multiplexer Switch Matrix Module - SSPP/C021 — Permits multiple highrate multiplexer input sources to be switched to a single high-rate multiplexer input channel.

3. Type U Crate Controller Module - SSPP/C022 — Directs multiple data word transfers between modules in the crate and memory interfaced to the system bus.

4. 9900 Microprocessor Dedicated Experiment Processor (DEP) Module - SSPP/C023

— Contains a 9900 microprocessor to provide autonomous local control and data processing for the experimenter.

5. Memory Module for the 9900 Microprocessor (DEP) - SSPP/C024 — Supporting

RAM/ROM for the 9900 dedicated experiment processor.

Functional Modules

• Communications Module - SSPP/F030 — Standard RS-232 interface module.

• Alpha/Numeric Video Scan Converter Module - SSPP/F031 — Alphanumeric

(ASCII) input to CCTV output video scan converter.

In addition to the previous described modules, SPSME offers the experiment user a low risk/ cost development of any of the remaining 100 plus functional modules now available as CAMAC hardware.

Experiment Applications

The various experiments to be carried on Spacelab missions represent a wide spectrum of data handling and control requirements. The items below depict typical data system functions required by a Spacelab experiment and describe how they may be accomplished:

Loading of microcomputer volatile memory with appropriate software in preparation for an experiment operation. The programs may be stored in MMU and loaded via the data bus/RAU.

Initiation of an experiment operation by means of a DDU/keyboard signal generated by the payload specialist. This signal is delivered to the experiment via the RAU.

Time line scheduling of steps in a sequence of events during an experiment operation.

Although this can be accomplished by means of application software in the EC, in most experiments it will be done by means of a microcomputer or minicomputer located in the experiment electronics.

Display of selected experiment outputs by the SL/DDU upon request by the payload specialist. This requires that the values be available to the CDMS through the data bus/RAU.

Interaction with the experiment operation by the payload specialist using the keyboard. Based on observation of experiment outputs shown by the DDU, he may wish to stop or modify the sequence of events or update parameters used in the test.

This may also involve observation of experiment outputs at POCC with instructions given to the payload specialist via voice communications.

Status monitoring of the experiment operation. This may require periodic limit checking of analog signals and/or change detection for discrete signals. It can be accomplished either by the local microcomputer (minicomputer) or by means of application software in the EC (using signals input to the RAU). This may also involve “redline” monitoring with provisions for automatic shutdown of experiment operations.

Acceptance and processing of clock and timing information from the CDMS. The experiment electronics must provide a means for the reconstruction of GMT to the appropriate resolution for time tagging of experiment data.

Acceptance and processing of other information received via the RAU such as shuttle orbiter attitude where this is required by experiment operation.

Acquire, digitize, and format data properly for the HRM. This includes data required in near real time at the POCC, as well as data for post mission analysis.

The items above are not, of course, an exhaustive listing of data and command related functions required by SL experiments. However, they are representative of data system functions required by a typical experiment and are illustrative of the guidelines and constraints applicable to design and operation of an experiment.

Figure 14 represents a typical SPSME configuration for a rather complex Spacelab experiment (pallets only); crate No. 1 resides in the orbiter aft flight deck and provides not only Spacelab CDMS interfaces but also a functional interface with an experiment dedicated control panel. Connected to crate No. 1 via a branch highway is crate No. 2 with a SPSME crate controller for control and monitoring of the pallet-mounted subsystems.

Table 2 represents quantity, weight, and power associated with the Figure 14. In this configuration the experiment is under control of the dedicated experiment processor.

Figure 15 represents a much simpler Spacelab experiment utilizing a single crate located in the Core Module; this system depends on the Spacelab experiment computer and PCC for command and control.

Because of the versatility of SPSME almost any spacelab experiment CDMS can be configured from the SPSME module inventory; in addition to those modules already mentioned, over 100 different CAMAC modules can be easily configured to SPSME module with minimum cost and risk giving extreme versatility and ease of configuration for experiment users.

Figure 14. SPSME Experiment Application with DEP


Standardization has shown the way to real savings in space hardware; this is even more significant with the advent of the Space Transportation System where standard hardware that can be easily reconfigured from mission to mission can be reused. Taking advantage of this opportunity, SPSME is so flexible that by simple front panel reconfiguration of modules and simple PCC microcode programming a complete experiment CDMS can be created from mission to mission for different experiments with a minimum of cost. This is just another way that the NASA Standard Hardware Office has through advanced long range planning implemented a system that can truely reduce the cost of performing experiments in space.



The authors wish to express their appreciation to W. Walt Frost and his people in the Data

Labs at NASA-Marshall Space Flight Center who generated some of the original data utilized



Figure 15. Animal Holding Experiment


1. “Space Shuttle System Payload Accommodation,” JSC-7700, Vol. 14. NASA, LBJ

Space Center, Houston, Texas.

2. “Spacelab Payload Accommodations Handbook,” NASA/ESA SLP2104 Issue 1,

Rev. 1, 31 July 1978.

3. “A Cost and Utility Analysis of NIM/CAMAC Standards and Equipment for Shuttle

Payload Data Acquisition and Equipment for Shuttle Payload Data Acquisition and

Control Systems,” NASA-CR147852. 30 June 1976.

4. “Feasibility Study of Common Electronic Equipment for Shuttle Sortie Payloads,”

NAS9-13784, NASA, LBJ Space Center, Houston, Texas, 1975.

5. “Study of SPAMAC/CAMAC Interface for the Spacelab Programme,” Z508/75JS.


6. a. “Modular Instrumentation and Digital Interface System (CAMAC),” IEEE

Standard 583-1975. IEEE, New York, N.Y., 1975.

b. “IEEE Recommended Practice For Block Transfers in CAMAC Systems,” IEEE

Standard 683-1976.

c. “Serial Highway Interface System (CAMAC),” IEEE Standard 595-1976. IEEE,

New York, N.Y. 1976.

d. “Parallel Highway Interface System (CAMAC),” IEEE Standard 596-1976.

IEEE, New York, N.Y. 1976.

7. “SPSME Task II Report,” NAS8-32642. NASA-MSFC 1977.

8. “SPSME Action Plan.” NASA Headquarters, 17 January 1977.

9. Ehrman, C. H., Baker, R. G., Smith, R. L., Kaminski, T. J., “A Flexible CAMAC

Based Data System for Space Shuttle Scientific Instruments.” 1978 Nuclear Science

Symposium, Washington, D.C. 1978.

10. Baker, R. G., Ehrman, C. H., Smith, R. L., Kaminski, T. J., Zipse, J. “Type U

CAMAC Crate Controller for the Spacelab Multi Crate, Multi Processor System,”

1978 Nuclear Science Symposium, Washington, D.C. 1978.

11. “Spacelab Payload Standard Modular Electronics (SPSME) Task I Report,” Data

Systems Lab EF-11. NASA-MSFC, Alabama April 1978.


W. R. Hedeman, Jr.


Annapolis, MD

E. L. Law

Telemetry Engineer

Pacific Missile Test Center

Point Mugu, CA


An instrumentation magnetic tape recorder, free of tape drop-outs, wow and flutter, is simulated by filters and a chromatic noise source. At a fixed bit error probability the capacity of the link is measured as a function of rms signal-to-noise ratio for NRZ-L,

Manchester and Miller codes. Two operating regions are observed: (1) noise limited at low values of SNR and (2) band limited at high values of SNR. In the noise limited region doubling the data rate requires a 6 dB increase in SNR; in the band limited region an increase of approximately 12 dB is required to produce the same result. The conclusion is that, for baseband recording of digital data, operation should be in the noise limited region slightly below the transition to the band limited region. If SNR margin is available at this operating point more data per square of tape can be stored by increasing the number of tracks rather than increasing the storage per track.

The theoretical penalty of 3.5 dB for the Miller code bit detector should, and does, result in a data rate decrease to .67 of the data rate with the NRZ-L code at the same SNR in the noise limited region. For the Manchester code the transition to the band limited region occurs at a lower SNR than for either NRZ-L or Miller codes. It is concluded that the

Manchester code would result in approximately the same data storage per square of tape as NRZ-L, and more than Miller, if the number of tracks is doubled.







This standard includes recommendations for pre-detection PCM recording, post-detection

PCM recording, and serial high density digital recording using IRIG analog wideband recorder/ reproducers. The serial high density digital standard, based on studies conducted at the Pacific Missile Test Center, recommends the BI0/-L code for packing densities up to

15 kilobits per inch and the randomized NRZ-L code (with 15-stage register length for packing densities up to 25 kilobits per inch. The signals are recorded using bias and the standard IRIG record levels. The minimum recommended reproduce rate (without special electronics) for randomized NRZ-L is 200 kilobits per second.

* Presented by E. L. Law, Pacific Missile Test Center, Point Mugu, Ca. 93042 and W. R.

Hedeman, Jr., Consultant, Annapolis, Md. 21403




Andrew R. Chi

Goddard Space Flight Center

Greenbelt, Maryland


A computer oriented time code designed for users with various timing requirements is proposed. Its format meets with the packet data format requirement of the new data handling and management concept, known as the NASA End-to-End Data System

(NEEDS). It is equally applicable to spacecraft and ground users. The time code is arranged in parallel groups of binary numbers containing day, second, millisecond, microsecond, and nanosecond resolutions. The day count system is a four digit number truncated from Julian day numbers known as Truncated Julian Day (TJD). It has a repetition period of 27.379 years.

Four options of resolutions in seconds, milliseconds, microseconds, and nanoseconds are offered. They are formatted in 4, 6, 7, and 8, eight-bit Byte words, respectively. To identify each resolution option of the time code, a variable prefix code consisting of 1 to 3 bits is used.

This paper will present in detail the time code and its applications.




B. L. Smith

Manager, Data Processing Engineering and Maintenance

Vandenberg Air Force Base, California


An interface was required for transmitting telemetered inertial guidance data from any one of twelve SEL 32/55 computers to an IBM 360-65 and a CDC 3300 computer. A flexible interface was designed to meet this requirement. The interface is comprised of several nested microcomputer systems and a fiber optic data transmission system for accomplishing the control and data transmission functions. A unique approach was used to transmit the data asynchronously and make it appear to be synchronous at the response of the computers receiving the data. A description of the approach and system operation will be discussed.



Ernie A. Dahl and L. Bates

Test Instrumentation Department

Naval Ship Weapon Systems Engineering Station

Port Hueneme, CA 93043


The paper presented by NSWSES at the 1975 ITC Conference in Washington, DC, described the Portable High Frequency Telemetry System (PUTTS) being assembled for

NATO. This system used the best of the then standard state of the arts commercial components and was used for shipboard missile data receiving/recording and for quick look missile performance evaluation. In 1977 the Naval Sea Systems Command made funds available to update the AN/SKQ-3 system by utilizing the RF assembly similar to that used in PUTTS. This new RF assembly provided dual antennas for (1) a wide angle for verticle launch and initial acquisition; (2) narrow beam high gain for long range tracking. The RF unit also included frequency scan with automatic lock when a signal was received, and sector search. In 1978 funding was received from Australia and Iran to procure additional PUTTS. These units (PUTTS III) were updated to handle faster intercept rates as well as improved range tracking and adapt the new RF features from the

SKQ-3 Mod. Added to this were the capability to handle both PAM and PCM data with light weight hardware. These systems were completed and the Australian unit system was delivered after acceptance tests with U.S. fleet operation in the Gulf of Mexico. As a result of these successful improvements a new portable system has been built to (1) adapt microprocessors to the set-up of data format; (2) Provide automatic selection remote control of the RF head within the antenna frame; (3) provide the basic data to make automatic processing possible when and if desired; (4) Add the new low noise GASFET preamp to the system to increase the range; (5) Add capability for four receivers in the space presently occupied by the dual receivers to permit the handling of the new missiles with dual RF outputs and (6) provide the capability of system checking of all modules from the antenna through the system to the paper read-out device. This paper now presents the new updated system combining the state of the art development in programming, remote control, low noise preamps, miniature RF assembly, matrix control programming as well as automatic data set up and selection for data processing.


NSWSES, the developer of the Portable UHF Telemetry Test System (PUTTS), has combined advantages of the PUTTS III and updated SKQ-3 into a new light weight, portable assembly. This system will be capable of processing real time, PAM, PCM, or

FM data from missiles and providing automatic quick-look evaluation of data between preset limits in ships. This new system is now known as the Weapon Instrumentation

Telemetering System (WITS).

The WITS is based on modular unit construction in which each sub-unit is complete within itself and can be interchanged to meet the NAVY missile telemetry monitoring needs as dictated by changing program requirements. Figure 1 shows the evolution of the WITS antenna system. The PUTTS III dual antenna with large receiver (Microdyne), the same antenna system re-packaged into a operation/shiping container with the new small receiver, and the WITS configuration with dual antenna, test antenna, dual receivers and programmable by two wire.


A programmer is built onto a common switching matrix which itself may be manually or electronically programmed, and provided in each sub-unit. This is based on the use of either manual switching or a multi-frequency tone code which permits microprocessors and memories to retain program requirements and set them up automatically from either a telephone dial, a printed card or magnetic tape.

This new type of matrix or program control was developed by NSWSES in 1978 and it was demonstrated at the SMS TLM Program Review at ECI in Tampa, Florida. It contained the matrix control and remote control assembly. The first model used a two-tone code system. The unit was updated in 1979 to use a five tone code which has the advantages of multi-toned, multifrequency receiver code utilization and the ability to actually discriminate between any frequencies. A rejection is achieved of all noise and any type of interference that is not compatible with the mixed tones. The actual algorithm used was developed imperically in actual telecommunication environments by using the statistical differences between noise, tone and speech. This then means that a two wire control using either (1) circuits combined on videocircuits; (2) circuits on two wire systems; (3) circuits on radiolinks or (4) on soundpowered lines on ships; can be used for giving detailed performance requirements to the units without jeopardizing the transmission with noise or jamming.

Figure 1. WITS Antenna System

This newer system was again demonstrated to an SMS TLM program review in 1979 at the Applied Physics Laboratory of John Hopkins University. An engineering development model is now undergoing evaluation in a PUTTS junior system used in USS

GOLDSBOROUGH (DDG-20) during Ship Qualification Acceptance Tests.

This remote programmer unit uses a two wire signal distribution to all switchable elements of the TLM system. The source of the signal is (1) a 20 section system switching memory which is programmed by a normal telephone dial or (2) a telephone-type dial assembly may be used to set up all of the elements on a real time basis. Figure 2 shows the prototype programmer used in WITS. When final operations are started, the preprogramming material is fed into a common line to set up all the units in the configuration required.

A one digit action code is utilized to start all the active elements within the system. Thus tape recorders, power recorders, timing clocks and other dynamic devices are automatically started by dialing a one digit code. The recording on cassette tape and the instrumentation tape of the complete set up and the starting of these functions means that

Figure 2. Protype Programmer with Receiver Control Elements

when the original tape is received for data processing the format is automatically defined on the main tape and that the cassette recorder tape can be utilized to set up automatic data processing if desired.


The original PUTTS System contained a dynamic system tester. This included a RF test frequency transmitter set up for external or internal antenna use with modulated so as to test the system. PUTTS III and utilized the same transmitter mounted in a plastic assembly

(flashlight) in which the reflector had been replaced by a circularly polarized plated antenna. The case contained the RF transmitter and a coded signal generator microprocessor controlled for 64 channels of PAM binary code. This binary code would identify each of the 64 channels. That is channel one would have a binary code, one channel two a binary two, etc. Figure 3 shows the flashlight tester with battery charger.

During test operation it would check thru the antenna, the preamp, the receiver the switchbox or switching matrix, and last but not least, it would write out on the paper recorder the selection out of the 64 channels that had been set up in the DECOM unit. This gives a thorough full operating system check, since this code is also placed on the tape recorder.

This RF dynamic tester (flashlight) also provided information from signal from the missile by indicating the relative transmitted power so that both a check of the missile and the receiving system could be made with one unit. The size of the flashlight is 4" x 6" x 4" without batteries and the weight is 2 pounds. In the new proposed WITS the flashlight is repackaged to physically mount in the RF assembly and under remote control would initiate a signal which will start the tester radiating. The tester will be coupled externally into the WITS receiving antenna to check antenna the full WITS without the need for an external test set.

Figure 3. Flashlight tester with Battery Charger


In order to provide improved portability over PUTTS the size of the sub systems comprizing WITS was reduced. These systems are packaged into containers which are set up with the tops and bottoms removable. When the equipment is set up for operation the bottom of the case is removed exposing all the plugs and wiring connections to the chassis of the modularized instrumentation. It was decided not to use slides and cord refractors since the increased cost and weight of these very desirable devices multiplies the physical size and shipping costs. Thus, WITS used a standard section case to provide the lightest weight and minimum size. Figure 4 shows a typical WITS sub assembly container, the RF assembly.


The RF assembly of WITS contains two separate cross dipoles elements set for right hand circular polarization. The use of this dual antenna system with two receivers permits a antenna gain of three from each antenna to each receiver. If a single antenna was used to feed both receivers a loss of 3db results due to the power splitter. Provisions are made for two types of antenna and two types of operation with this RF assembly in the surface missile ship configuration.

Figure 4. Typical WITS shipping container


The RF assembly with two receivers, antenna preamps, filters and test transmitter are installed on the CCTV bracket normally a part of the missile fire control Radar Director.

This unit fits on the CCTV mounting base of the director after the removal of one of the counterweights. The RF assembly that’s added weighs approx 35 pounds. The unit receives its prime power (117 volts 400 Hz) from the director. The two coaxial outputs containing receiver video and receiver performance data are fed through the existing coaxial sliprings and through the mount into the data processing equipment located two decks below. There is no additional modifications to the director. The cover is then placed back on the RF assembly and the RF assembly is ready to use. Since the WITS TLM RF assembly is mounted on the target tracking radar, a autotrack assembly is not required for this mode of operation. Figure 5 shows the WITS RF assembly mounted on a AN/SPG-51 radar.


The WITS systems utilizes a new generation receiver, specifically developed for the

NSWSES high performance protability requirement. The receiver utilizes a frequency synthesized local oscillator with 100 channel spacing at 1 MHZ intervals. Remote

Figure 5. WITS RF assembly mounted on AN/SPG-51 Radar

frequency tuning is accomplished using the 4 x 4 multi frequency touch tone signal on a two-wire system. The receiver is housed in a aluminum housing 5 x 7 x 4 inches weighing

6 pounds and is manufactured by TRAK Microwave. A summary of the peformance specifications are:

FREQ: 2200-2300MHz

INUPUT NOISE: 2 db maximum

SENSITIVITY: 104 dBm. @ 6 db signal and noise to noise


OUT OF BAND RESPONSE: $ 40 dB 1980-2140 MHZ


60 db DC- 1099, 2600-10,000 MHZ



The receiver design allows for mounting four receivers in a standard 19 inch rack size. The

TRAK model 1500-9303 is shown in Figure 6.

Figure 6. TRACK model 1500-9303 TLM receiver


In keeping with the portable, minimum size/weight requirement for WITS the Bell Howell

Model M-1400 (Modified) was selected. This recorder has been modified to provide the following necessary features for shipboard telementary and reproduced applications:

(1) 2 MHz direct frequency response @ 120 IPS

(2) 14 Track record/reproduce head assembly from the M14G system for increased performance during reproduce

(3) 14 reproduce preamplifiers as part of M-14G head assemblies

(4) Special power supply with built in 3 channel reproduce amplifier, switch selectable with input/output level monitor

(5) Voice playback amplifier.

The M-1400 (Modified) tape recorder maintains all of the standard desirable IRIG tape recorder features such as: 14 track direct or FM record capability, 6 speed 3-3/4 to

120 ips, 14 inch NAB reel capacity, ruggedized configuration with RF1 shielding and pressure sealing to insure reliable operation even during condition of water run - off. The recorder used with PUTTS III with full 14 track six speed record/reproduce capability is shown in Figure 7.


The strip chart recorder chosen for the WITS system is the new all solid state CEC Model

HR 2000 datagraph. This recorder was chosen to allow display of up to 28 data channels on any size direct print papers up to 12". The employment of the programmable light gate

Figure 7. MARS 1400 Tape Recorder

array by CEC allows sinewave frequency responses to 5KHz and DC-10KHz square wave with signal amplitude up to 12". The standard HRZ000 has slight modification of a BNC input panel with trace positioning capability through the input comparator plug-in to allow system operation directly from the DAC in the PAM/PCM data DECOM system without additional pre-amplifiers.


The WITS is designed to handle both PAM, PCM, or FM data from Navy Missiles. The system selected to accomplish this is a compromise between large laboratory type systems and small portable field equipment. The system block diagram for the EMR 600 series equipment used in WITS is shown in Figure 8. As shown the unit consists of the EMR 691

PCM BIT sync., a EMR model 699 frame/subframe synchronizer, a EMR 686-03 digital/analog converter and a model 515M-23 PAM sychronizer. The PCM portion of the unit operate with selectable bit rates from 100 bps to 1 Mbps with NRZL or Biphase-L.

The system will-handle variable word lengths from 4 to 16 bits. The bit synchronizer has four selectable bit rates, set by switch selection with rates of 100Kb, 244.K Kb, 614.4Kb

or 755.75Kb. Others may be chosen as future requirements dictate. The frame/subframe synchronizer will be initially configured to meet special navy surface missile requirements.

Up to 14 dual DAC will be mounted in the 600 series card cage to allow 28 continuous


analog outputs. When PAM data is received from missiles the PCM serial output from the

Model 5115 PAM sychronizer will be patched into the PCM bit synchronizer for processing through to the DAC. The whole assembly fits into a portable carrying case 24 x

10½, x 20 and weighs 50 pounds. The unit is designed for 0-55EC operating temperature,

20g shock and 5% to 90% humidity.


It is now possible to combine a programmer, microprocessor controlled evaluation circuits, and a printer into a combined unit which will take fixed format data from the PAM/PCM

DECOM, re format it for evaluation against pre set limits and prints. The data will be automatically annotated in engineering units with those outside program limits flagged.

The unit in WITS selected for this function is the Digistrip II which allows for evaluation of 16 channels at a 8 channeled/sec. scan rate. This unit meets the WITS concept of modularity, stand alone capability with system integration by simple cable hook-ups. The unit weighs 45 lbs in a 17 x 12 x 19 inch container.


Programming is entirely by the front panel keyboard. Front panel graphics guide the operation and built-in diagnostics test all entries. A valid entry is recognized with a short, audible beep-tone. An invalid entry generates a long beep-tone, and the invalid entry is

ignored automatically. The programmed instructions may be reviewed on the display or may be printed by pressing the LIST PROGRAM key.

The programming sequence includes: configuration, header programming, function definition and individual channel programming.


In addition to the missile TLM program the functions program provides functions which the user can define for linear scaling and accumulation of data from Ships Fire Control

System. These data can be recorded as analog voltages via a VCO multiplex signal and processed through FM discriminators. These FM discriminators outputs are then fed into the Digistrip II for post reduction processing.


The WITS concept is to provide a TLM ground station that achieves state of the art capability by utilizing as much as possible industry standard modules and instruments integrated into a small ruggedized portable system for shipboard use. This technique has highly desirable features for (1) achieving high performance at affordable prices (2) procurement and delivery without long development time; (3) modularity allowing reconfiguration as state of the art hardware is produced by industry; (4) ease in system reconfiguration to meet new program requirements; (5) size and weight reduction to allow portability and transportation ease; (6) data evaluation and matrix program control without complex computer hardware or software. These advantages make WITS a viarable option for shipboard telemetry system procurement.




Arthur Sullivan

Vice President

Electra Magnetic Processes, Inc.

9612 Owensmouth Avenue

Chatsworth, CA 91311


A dual-beam, single-axis, tracking antenna capable of receiving and tracking telemetry data from an instrumented airborne vehicle is described. The dual-beam, single-axis, tracking antenna system consists of both a high-gain and a low-gain antenna positioned by the same antenna pedestal/servo electronics. Automatic switching between the high-gain and the low-gain antennas, based on received signal strength, permits tracking from maximum range (using the high-gain antenna) to close-in nearly overhead passes (using the low-gain antenna) by exploiting the wide-beam characteristics of the low-gain antenna when space losses are at a minimum. The wide beamwidth of the low-gain antenna permits its use as an acquisition aid for the high-gain antenna during initial acquisition, while its wide beamwidth also precludes locking on to a side lobe. The use of only one tracking axis rather than two reduces cost and improves reliability.


For most flight testing of telemetry instrumented vehicles it is desirable to be able to track and receive data from the vehicle from take-off or launch point (which may be close to the tracker) to maximum range, and also to track and receive data from the vehicle at maximum altitude during overhead passes. Conventional single antenna, single-axis, tracking systems are not capable of doing this because nulls in the tracking antenna pattern can cause loss of track and loss of data.

The majority of mission requirements can be satisfied with a six-foot diameter tracking antenna at L-band. A six-foot pencil beam antenna, however, has a half-power beamwidth of approximately 7 degrees and the first null is about 9 degrees from the peak of the beam.

If the 1 dB point of the antenna beam were on the horizon then the vehicle would be in a null at an elevation angle of about 11 degrees; indeed, because of the steep slope of the

radiation pattern an unacceptable carrier-to-noise ratio (C/N) would be experienced at an elevation angle of about 9 degrees.

Close-in operation using a single-axis system can be improved by the use of a cosecant squared elevation pattern which shapes the beam to increase gain at high elevation angles.

This apparent advantage of a cosecant squared elevation pattern is largely negated by a reduction in peak antenna gain of 3 to 4 dB due to pattern shaping. In addition cosecant squared coverage is only realistically obtainable to an elevation of about 40 degrees so that close-in operation and overhead passes cannot be accommodated by the use of a cosecant squared antenna.

This paper describes an azimuth only telemetry tracking antenna with a narrow-beam and a broad-beam mode of operation. The narrow-beam antenna is used for tracking the vehicle at long ranges and the broad-beam antenna is used fro tracking the vehicle in close including near overhead passes, The broad-beam antenna, with an acquisition angle of nearly 70 degrees (acqusition within ±35 degrees of boresight axis) also serves as an acquisition aid for the high-gain antenna. The system automatically switches between the low-gain and high-gain antenna based on received signal strength.


Figure 1 is a photograph of the EMP Model LTS-06C Telemetry Tracking Antenna in the dual-beam configuration. The high-gain antenna is a six-foot reflector focal-point fed by a radome-enclosed stripline single-channel monopulse feed. In back of this feed is an equipment enclosure which houses polarization hybrids, comparators, scan converter, bandpass filter, preamplifier, and switches for switching between the high-gain and lowgain antennas. At the rear of the enclosure is the low-gain 4-element stripline singlechannel monopulse dipole array also enclosed in a radome. The high-gain antenna has a minimum gain of 25 dBic and a minimum beamwidth of 7 degrees from 1435 to 1535

MHz. The low-gain antenna has a minimum gain of 12 dBic, a minimum azimuth beamwidth of 20 degrees, and a minimum elevation beamwidth of 80 degrees in the same frequency range. The high-gain antenna is tilted upward approximately 2 degrees so that the 1 dB point of the pencil beam is on the horizon. When the control unit is turned on the system automatically selects the low-gain antenna since acquisition is usually made at close-in ranges. The acquisition angle of the low-gain antenna is approximately 70 degrees so that side lobe lock-on is precluded. When the vehicle is going away from the tracker the system will automatically switch to the high-gain antenna at a carrier-to-noise ratio of

50 dB. When the vehicle is coming in towards the target in the high-gain mode the system will switch back to the low-gain antenna at a carrier-to-noise ratio of 30 dB. This 20 dB hysteresis precludes unnecessary switching between the high and low-gain antennas.

The pedestal has a velocity capability of 20 degrees per second and an acceleration capability of 50 degrees per second squared. On a near overhead pass while the azimuthal velocity and acceleration of the vehicle may greatly exceed that of the pedestal, the antenna can lag the vehicle by as much as 35 degrees without losing track. Therefore the probability of loss of track on near overhead passes is greatly minimized.

The LTS-06C system can also be configured to operate from 1435 to 2300 MHz. With this configuration the system can track either aircraft or missiles or could initially track an aircraft and then switch over to a launched missile.


The maximum range of a tracking antenna is dependent on many parameters such as transmitter power, transmitter antenna gain, receiver bandwidth etc. Figure 2 is a block diagram of an airborne tracking mission with the applicable definitions and equations.

Figure 3 is a computer plot of range versus effective isotropic radiated power (EIRP) as a function of receiver predetection bandwidth. Although over-the-horizon tracking is generally possible, this plot is based on line of sight considerations. The receiving antenna gain used is 24 dBic to correspond to the high-gain antenna being elevated so that the 1 dB point is on the horizon. The ranges calculated allow for a multipath fade margin of 10 dB which is conservative for tracking over land but may be marginal for tracking over water.

This plot also assumes that the airborne vehicle has a linearly polarized antenna while the receiving antenna is circularly polarized. The preamplifier noise figure used is 2.5 dB

(noise temperature = 226 degrees K) which allows for the use of a low cost bi-polar amplifier. Should more range be required a GaAsFET preamplifier with a noise figure of less than 1.5 dB (120 degrees K) could be used. A loss of 8 dB between the preamplifier and receiver is assumed to allow for an adequate distance between the tracking system and control site. Even with these conservative parameters, the plots indicate that the maximum line-of-sight range of the Model LTS-06C utilizing a receiver bandwidth of 1 MHz exceeds 200 miles for a 10 watt radiated power.


While the mission requirements vary for various telemetry applications of tracking aircraft and helicopters a typical mission could be considered to be a maximum range of 150 miles with a maximum altitude of 25,000 feet.

Figure 4 is a computer plot of the range versus altitude for constant field intensity radiation patterns of the high-gain and low-gain antenna. The antenna is tilted so that the 1 dB point of the high-gain antenna pattern is on the horizon. The earth’s curvature and slant range are accounted for in plotting the constant field intensity radiation patterns. While the

antenna tilt was selected to optimize coverage for a flight profile of 150 miles range at

25,000 feet altitude it is noted that good coverage is provided for altitudes well in excess of 25,000 feet. As an example if an aircraft flew at 40,000 feet from the maximum range to directly overhead the received signal strength would always be greater than the signal strength received at minimum altitude maximum range for line of sight with the exception of approximately a 5 degree sector centered at about an 8 degree elevation angle. In this sector the received signal is only down about 1 dB so that good data reception would be experienced even at 40,000 feet.

The elevation of the antenna can be manually adjusted through a range of ±5 degrees so that the system is capable of being optimized for a variety of mission requirements.


Because of the dual-beam feature of the system presented herein the majority of airborne tracking and telemetry mission requirements can be satisfied without the necessity of employing a tracking elevation axis. The cost of the antenna pedestal/servo control system is reduced by a factor of two and reliability is improved by a factor of two. Since manual elevation adjustment is provided this system can easily be tailored to a variety of different mission applications. This system also is far superior in tracking near overhead passes than a conventional single-antenna-beam two-axis tracking system. This is because the system described in this paper is tracking near overhead passes with a wide-beam antenna having an acquisition angle of approximately 70 degrees, consequently the tracked vehicle can greatly exceed the dynamics of the tracking system. Providing that the tracked vehicle does not lead the tracker by more than 35 degrees the system will successfully track the near-overhead pass. The acquisition angle of a single beam 6-foot diameter two-axis tracking system is approximately 14 degrees. Therefore the two axis system would lose track in a near overhead pass when the aircraft led the tracker by 7 degrees.

The dual beam single axis tracking approach can be applied to almost any tracking mission by varying the gain in both the high-gain and low-gain antenna to optimize performance over a specific flight profile. To date high-gain antennas have been built with 4, 6, and 8 feet diameters and the low-gain acquisition antennas have been built using two element and four element arrays.


The author wishes to thank Charles W. Chandler for the computer programming of the graphs and valuable discussions contributing to the content of this paper.




James C. Wiltse

Associate Director

Engineering Experiment Station

Georgia Tech

Atlanta, GA 30332


Millimeter-wave technology has progressed rapidly during the past several years. A summary is given of the state-of-the-art in components, devices, and techniques and their uses in several system applications. Although present systems activity includes work on radar, guidance, remote sensing, radio astronomy and spectroscopy, the applications to communications are emphasized.


The strong revival during the past several years of research and applications in the millimeter-wave region is attributed to the advent of new technology, the evolution of new requirements for sensors and communication links, and the superiority of millimeter-wave systems over optical and infrared systems for penetration of smoke, fog, haze, dust, and other adverse environments. The improved technology includes better sources (such as

IMPATTS, Gunn oscillators, gyrotrons, extended interaction oscillators, and magnetrons) which have higher power outputs and/or operate at higher frequencies, and in some cases have longer lifetimes than earlier designs. Lower-noise mixers have also been developed, providing noise temperatures as low as 600 K (uncooled) or 200 K (cooled) near

100 GHz. Component development has progressed, too, particularly in the areas of integrated circuits, image lines, fin line waveguides, and quasi-optical techniques.

Nonetheless, there is room for improvement in components and devices, particularly in extending ferrite or other non-linear devices to frequencies above 100 GHz and in reducing losses in most components. For a variety of reasons the frequency range from 35 to

100 GHz (with some special examples at 140 GHz) has seen the heaviest system development, while the range above 100 GHz (and on into the submillimeter range) is seeing a concentration of research effort on components and techniques.


The presentation which follows better describes the technology improvements mentioned above, and provides many pertinent references. In the area of sources, both solid state and vacuum-tube types are available and in use in various ways. For low power applications

(watts of peak power, milliwatts average) solid state IMPATTS have seen extensive development and now operate to 300 GHz.

1-4 Gunn oscillators are available to 100 GHz, 5-7 klystrons to 200 GHz, and carcinotrons to 1000 GHz.

8,9 In medium power tubes, the magnetron has been improved in lifetime and is available to 95 GHz, and the extended interaction oscillator (EIO) has appeared and is in use up to 140 GHz.


An EIO has been operated at 270 GHz and pulsed units are being developed for 220 GHz. A breakthrough in extremely high power has been obtained with the gyrotron tube and related devices, which have already achieved megawatts peak at efficiencies greater than 30%, and hundreds of kilowatts CW at efficiencies near 500 11,12,13 The gyrotron has seen application in fusion research and undoubtedly will soon see use in communications and radar.

Optically-pumped lasers can also be used to produce significant power (up to tens of kilowatts peak) at specific millimeter and submillimeter frequencies, but efficiencies and pulse repetition rates are low.


The recent growth in millimeter wave applications is certainly attributable in part to the availability of improved building blocks, particularly solid state sources. While such sources provide relatively low power, they are smaller in overall size and require much lower voltages and prime powers than vacuum tube types, such as magnetrons, EIO’s, and klystrons. The most obvious improvements in solid state sources during the past several years have been the availability of higher power outputs and operation at much higher frequencies. Some of the other improvements are more subtle, but very important for systems; these include development of frequency stabilized or phase-locked sources, injection-locked IMPATT amplifiers, and frequency-doubled Gunn oscillators. (As an aside, frequency stabilization and phase-locking are being developed for pulsed EIO’s, also.) These types of improvements are permitting the extension of all-solid-state pulsecompression and coherent MTI (moving target indication) radars to as high as 94 GHz.

Similarly, all-solid-state receivers are being extended to at least 140 GHz.

In the mixer area both the microwave structures


and diode materials


have been extensively studied. In a decade of effort mixer noise temperatures have been lowered by an order of magnitude. Another innovation has been the use of subharmonic mixing, wherein employing two back-to-back diodes permits the use of a local oscillator frequency of 1/2 or 1/4 the normal fundamental L.O. with excellent performance. This permits building all-solid state receivers and radiometers at frequencies much higher (2 to 4 times) than those available from solid state-sources.

18 Of course, another approach is to use harmonic generators as sources for fundamental mixing


. The various types of diodes

investigated in recent years have included superconducting tunnel junctions, point-contact

Josephson junctions, Schottky devices, Mott diodes (a low-doped Schottkys with an epitaxial layer that is completely depleted at zero bias), and Mottky diodes


(Schottky barrier diode with a thin epitaxial layer, approximating a Mott). An additional advantage for some of these mixers is a lower requirement for local oscillator power (as low as the order of 0.1 milliwatt). Some of the best research has been done for radio astronomy applications, but the results carry over directly to other fields. Important considerations for some of these mixers are the needs for being ruggedized and being made more producible.

Nonetheless, the improvements have led to better receivers for radar, communications, and radiometry (remote sensing and radio astronomy).

In addition to the work on sources and mixers mentioned above, the literature on component and device development is extensive.

techniques to millimeter wavelengths was the subject of a special issue of the IEEE


15 Finline structures 30

5,6,7,28,29 Extension of integrated circuit

and insular guides (or dielectric image lines) been investigated in considerable detail.

31 have

In general, atmospheric propagation efforts dominate considerations relating to applications. This is true even for satellite communications outside the atmosphere, since frequencies may be chosen for which the atmosphere is opaque, thus preventing detection of satellite-to-satellite communications by ground-based receivers. Terrestrial systems desiring to prevent signal “overshoot” in range may similarly operate at a frequency of high atmospheric absorption to gain a degree of covertness. Typical values of atmospheric attenuation have been well-established for frequencies up to 100 GHz, including the attenuation and backscatter due to rain or fog, and programs have now been initiated to obtain better information about atmospheric effects above 100 GHz. Of particular interest are the effects in atmospheric “windows,” or attenuation minima, near 35, 94, 140, and

220 GHz. Atmospheric turbulence effects due to time-varying localized temperature and humidity variations in clear air produce amplitude scintillations and angle-of-arrival changes, and these are now being investigated in more detail.


The effects of smoke and/or dust are also being measured. One example is a test (“DIRT-I”) conducted in

September, 1978, at White Sands Proving Ground by the Army Atmospheric Sciences

Laboratory. Propagation through a dust and oil smoke cloud was measured at 94 and

140 GHz qnd at infrared wavelengths. In other programs the reflectivities and radiometric signatures of land and sea (i.e., “clutter”) , as well as vehicles and other targets, are also being obtained at the “window” frequencies.



One of the areas greatest activity in millimeter waves is that of guidance for missiles and projectiles.


The Defense Advanced Research Projects Agency (DARPA) and the three

services are now placing particular emphasis on terminal guidance of tactical air-to-surface missiles. In order to obtain narrow beamwidths from small antennas, millimeter-wave frequencies are being used, with the choice typically being 35 or 94 GHz. Extensive design and measurement programs are underway to develop “terminal seekers” optimized for various military requirements.


In general, all-solid state design has been emphasized for these seekers, consistent with the need for small size and low voltage. In addition to these types, semiactive systems are being investigated, with obvious analogy to laser designators and seekers. Similarly, millimeter-wave beam-rider missile guidance is being studied. For these examples higher power transmitters are needed, which generally rules out solid-state sources.

One of the newer radar systems under development is called STARTLE (Surveillance and

Target Acquisition Radar for Tank Location and Engagement), which sould provide US tanks with the ability to detect and track other tanks (or other armored vehicles) in adverse environmental conditions such as fog, smoke, dust, or rain.

35,38,40-42 The Army and DARPA are jointly sponsoring the program, which is being carried out with two competing contractors. Prototype all-solid-state models are being fabricated for operation at 94 GHz.

Radiometric sensors are being extensively used by NASA for remote sensing the Navy for navigation and imaging.


43-46 and by

For example, the Nimbus 6 satellite carried five superheterodyne radiometers with center frequencies between 22 and 60 GHz.

Temperature profiles were successfully measured for a variety of surface locations and atmospheric conditions. Recently, various radiometer receivers centered at frequencies between 90 and 183 GHz were flight tested in a Convair 990 aircraft. At 183 GHz, which is the frequency of an atmospheric water vapor transition, the radiometer provides atmospheric temperature data related to water content. A 93 GHz channel obtains surface or lower atmospher temperature information. A dual channel (93/183 GHz) system is now installed in an RB-57 aircraft.

Other areas where applications are developing include scale modeling (e.g., measuring radar returns from scale models at millimeter/submillimeter frequencies to simulate returns at microwave frequencies) 48,49 electronic warning receivers, 50 auto collision avoidance radar, 51-53 and Coast Guard obstacle avoidance.


Several types of millimeter-wave communications systems have been developed or designed. These include terrestrial point-to-point


or ship-to-ship systems whose carrier frequencies were chosen to take advantage of atmospheric propagation characteristics; enclosed, low-loss, TE


mode, circular waveguide for use in wideband, long-haul, heavyroute communication systems; and satellite-to-satellite or satellite-to-ground links.

The most straightforward scheme that has been developed for terrestrial (surface) communications is a shortrange (5-20 miles) duplex unit operating at 38 GHz, which is a low atmospheric-attenuation region.


A 35 GHz railway communication link has also been developed.


The advantages of a millimeter-wave system over a microwave system include the availability of a narrow beam width (high gain) from a small aperture, broader bandwidths (possibility of frequency agility), link privacy, and reduced interference.

In other cases the choice has been to use a frequency near 60 GHz, where attenuation is high, so signal overshoot is low and covert operation is possible. Examples include a lineof-sight communicator for use between ships.


The design employs 3-inch diameter paraboloid reflectors (4.5 degree beamwidth), an IMPATT transmitter/local oscillator and battery-pack power. Transceiver weight is 2 pounds and battery pack 6 pounds. Other systems have been devised which emphasize low cost and civilian application, 59,60 as well as advanced dielectric waveguide circuitry.


For many years Bell Laboratories and other organizations developed components and technology in relation to the so-called circular-electric mode (TE


) waveguide, which has been shown to provide extremely low loss (about 2 dB/km) over the frequency range from

33 to 125 GHz, while carrying about 220,000 voice circuits simultaneously. Complete systems have been developed in the U.S., Japan, and England.

63-71 Several years ago the

Bell System installed a 14-km long field test system in New Jersey, and a similar 22.7-km length of experimental waveguide has been installed in Japan. Tests have apparently been highly successful.


Early satellite communications in the 30-33 GHz region were carried out between the

NASA Applications Technology Satellites (ATS-5 and ATS-6) and round stations.


Satellite-to-satellite millimeter-wave relay links were also designed, 74 leading to the development of needed components at V-band, since most of these links would have taken advantage of the very large atmosphere attenuation near 60 GHz to provide isolation from ground receivers.

More recently a variety of satellite links has been reconsidered, partly for reasons of spectrum congestion and/or cost benefits.

75,76 It seems probable that further millimeterwave applications will evolve in the near future.



Millimeter-wave technology has advanced rapidly in recent years, and the state-of-the-art in components and devices is well-advanced and ready for numerous system uses. Several versions are well along in development, and in some cases, production potential has begun to appear.



H. J. Kuno and T. T. Fong, “Solid State mm-Wave Sources and Combiners,”

Microwave Journal, Vol. 22, pp. 47-48, 73-75, 85; June, 1979.


T. A. Midford and R. L. Bernick, “Millimeter-Wave CW IMPATT Diodes and

Oscillators,” IEEE Transactions on Microwave Theory and Techniques, Vol.

MTT-27, pp. 483-492; May, 1979.


T. T. Fong and H. J. Kuno, “Millimeter-Wave Pulsed IMPATT Sources,” ibid., pp.



F. J. Bernues, R. S. Ying, and M. Kaswen, “Solid State Oscillators Key to

Millimeter Radar,” Microwave System News, Vol. 9, pp. 79-86; May, 1979.


N. B. Kramer, “Solid State Technology for Millimeter Waves,” Microwave Journal,

Vol. 21, pp. 57-61; August, 1978.


N. B. Kramer, “Millimeter-Wave Semiconductor Devices,” IEEE Transactions on

Microwave Theory and Techniques, Vol. MTT-24, pp. 685-693; November, 1976.


F. B. Fank, J. D. Crawley, and J. J. Bernz, “InP Material and Device Development for Millimeter Waves,” Microwave Journal, Vol. 22, pp. 86-91; June, 1979.


G. Kantorowicz, P. Palluel, and J. Pontvianne, “New Developments in

Submillimeter-Wave BWOs,” Microwave Journal, Vol. 22, pp. 57-59; February,



M. B. Golant, et al, “Wide Range Oscillators for the Submillimeter Wavelengths,”

Pribory i Teknika Eksperimenta, pp. 231; 1969.


“Introduction to Extended Interaction Oscillators,” Data Sheet No. 3445 5M,

November, 1975, Varian Associates of Canada, Ltd., Georgetown, Ontario,



H. Jory, S. Heggi, J. Shively, R. Symons, “Gyrotron Developments,” Microwave

Journal, Vol. 21, pp. 30-32; August, 1978.


K. R. Chu, A. T. Drobot, V. L. Granatstein, and J. L. Seftor, “Characteristics and

Optimum Operating Parameters of a Gyrotron Traveling Wave Amplifier,” IEEE

Transactions on Microwave Theory and Techniques, Vol. MTT-27, pp. 178-187;

February, 1979.


Zeitscheff and Kisel, Radio Eng. Electron. & Phys. (Soviet), Vol. 19, p. 1056; 1974.


J. J. Gallagher, M. D. Blue, B. Bean, and S. Perkowitz, “Tabulation of Optically

Pumped Far Infrared Laser Lines and Applications to Atmospheric Transmission,”

Infrared Physics, Vol. 17, pp. 43-55; 1977.


Special Issue on Microwave and Millimeter-Wave Integrated Circuits, IEEE Trans.

on Microwave Theory and Techniques, Vol. MTT-26, No. 10; October, 1978.


Special Issue on Millimeter Waves: Circuits, Components, and Systems, IEEE

Trans. on Microwave Theory and Techniques, Vol. MTT-24, No. 11; November,



N. E. Swanberg and J. A. Paul, “Quasi-Optical Mixer Offers Alternative,”

Microwave System News, Vol. 9, pp. 58-60; May, 1979.


A. P. Cardiosmenos, “Planar Devices Make Production Practical,” Microwave

System News, Vol. 9, pp. 46-56; May, 1979.


R. Kawasaki and K. Yamamoto, “A Wide-Band Mechanically Stable Quasi-Optical

Detector for 100-300 GHz,” IEEE Transactions on Microwave Theory and

Techniques, Vol. MTT-27, pp. 530-533; May, 1979.


M. V. Schneider, “Low-Noise Millimeter Wave Schottky Mixers,” Microwave

Journal, Vol. 21, pp. 78-83; August, 1978.


N. J. Keen, “The Mottky-Diode: A New Element for Low Noise Mixers at

Millimeter Wavelengths,” AGARD Munich Conference on Millimeter and Submillimeter Wave Propagation and Circuits, Proceedings No. 245, pp. 16-1 to 16-9;

September, 1978.


A. Kerr, “Noise and Conversion Loss of Two-Diode Subharmonically Pumped and

Balanced Mixers,” IEEE International Microwave Symposium Digest, Orlando,

Fla., pp. 17-19; April 30, 1979.


D. Held, “An Approach to Optimal Mixer Design at Millimeter and Submillimeter

Wavelengths,” ibid,. pp. 25-27.


T. A. Smirnova and N. T. Cherpak, “A Millimeter Band Paramagnetic Receiver with Frequency Down Conversion,” Radioteknika i Elektronica, Vol. 20, No. 2, p. 348; 1975.


V. G. Mirovskiy, I. A. Strukov, and V. S. Etkin, “A Study of Centimeter and

Millimeter Band, Wideband Degenerate Parametric Amplifiers Employing

Semiconductor Structures Without Cases,” Radioteknika i Elektronica, Vol. 20,

No. 4, p. 796; 1975.


I. A. Strukov, Y.U.B. Khapin, and V. S. Etkin, “The Stability and Sensitivity of a

Radiometer Operating at 2.2 mm,” Radioteknika i Elektronica, Vol. 20, No. 5, p. 1058; 1975.


T. Takada and M. Ohmori, “Frequency Triplers and Quadruplers with GaAs

Schottky-Barrier Diodes at 450 and 600 GHz,” IEEE Transactions on Microwave

Theory and Techniques, Vol. MTT-27, pp. 519-523; May, 1979.


J. A. Calviello, “GaAs Schottky Barrier Devices and Components for Millimeter and Submillimeter Application,” Microwave Journal, Vol. 22; August, 1979.


E. P. Valkenburg and D. D. Khandelwal, “Bulk p-i-n Diode Millimeter Waveguide

Switch and Phase Shifter,” Microwave Journal, Vol. 22; August, 1979.


P. J. Meier, “Printed-Circuit Balanced Mixer for the 4 and 5 mm Bands,”

Microwave Journal, Vol. 22; August, 1979.

L. D. Cohen and P. J. Meier, “E-Plane Mm-Wave Circuits,” Microwave Journal,

Vol. 21, pp. 63-65; August, 1978.


R. Mittra and S. Bhooshan, “Multimode Waveguide Components for Millimeter-

Wave Integrated Circuits,” IEEE International Microwave Symposium Digest,

Orlando, Fla., pp. 211-213; May 1, 1979.


G. A. Andreyev, et al, “Intensity and Angle of Arrival Fluctuations of Millimetric

Radiowave in Turbulent Atmosphere,” Joint Anglo-Soviet Seminar on Atmospheric

Propagation at Millimetre and Submillimetre Wavelengths, Institute of

Radioengineering and Electronics, Moscow; November, 1977.


R. W. McMillan, J. C. Wiltse, and D. E. Snider, “Atmospheric Turbulence Effects on Millimeter Wave Propagation,” Proceedings IEEE Symposium on Aerospace and

Electronic Systems, Washington, D.C.; October, 1979.


R. N. Trebits, R. D. Hayes, and L. C. Bomar, “Millimeter Wave Reflectivity of

Land and Sea,” Microwave Journal, Vol. 21, pp. 49-53 and 83; August, 1978.


J. C. Wiltse, “Millimeter Waves - They’re Alive and Healthy,” Microwave Journal,

Vol. 21, pp. 16-18; August, 1978.


N. C. Currie, J. A. Scheer, and W. A. Holm, “Mm-Wave Instrumentation Radar

Systems,” Microwave Journal, Vol. 21, pp. 35-42; August, 1978.


C. R. Seashore, J. E. Miley, and B. A. Kearns, “Radar and Radiometer Sensors for

Mm-Wave Guidance Systems,” Microwave Journal, Vol. 22; August, 1979.


G. S. Sundaram,”Millimetre Waves: The Much-Awaited Technological

Breakthrough?,” International Defense Review 2, pp. 271-277; February, 1979.


D. T. Williams and W. H. Boykin, Jr. “Millimeter-Wave Missile Seeker Aimpoint

Wander Phenomenon,” AIAA Journal of Guidance and Control, Vol. 2, pp. 196-

203; May-June, 1979.


R. Balcerak, W. Ealy, J. Martins, and J. Hall, “Advanced Infrared Sensors and

Radar Systems for Tactical Target Acquisition,” Society of Photo-Optical

Instrumentation Engineers (SPIE) Symposium on Effective Utilization of Optics in

Radar Systems, Huntsville, Alabama, pp. 172-184; 1977.


P. H. Backus, “Electro Optics,” Journal of Electronic Defense, Vol. 2, pp. 24-35;

March/ April, 1979.


J. Fawcette, “Army to Test STARTLE Radar While Navy Readies Gyrotron,”

Microwave Systems News, Vol. 8, pp. 23-24; July, 1978.


J. Hank Rainwater, “Radiometers: Electronic Eyes That ‘See’ Noise,” Microwaves,

Vol 17, pp. 59-62; September, 1978.


J. M. Schuchardt and J. A. Stratigos, “Detected Noise Levels Guide Radiometer

Design,” Microwaves, Vol. 17, pp. 64-74; September, 1978.


J. Schuchardt, J. Stratigos, J. Galiano, and D. Gallentine, “Dual Frequency Multi-

Channel Millimeter Wave Radiometers for High Altitude Observation of

Atmospheric Water Vapor,” IEEE International Microwave Symposium Digest, pp. 540-542; May 2, 1979.


M. D. Blue, “Reflectance of Ice and Seawater at Millimeter Wavelengths,” ibid., pp. 545-546.


J. P. Hollinger, J. E. Kinney, and Ballard E. Troy, Jr., “A versatile Millimeter-Wave

Imaging System,” IEEE Transactions on Microwave Theory and Techniques, Vol.

MTT-24, pp. 786-793; November, 1976.


L. A. Cram and S. C. Woolcock, “Review of Two Decades of Experience Between

30 GHz and 900 GHz in the Development of Model Radar Systems,” AGARD

Munich Symposium on Millimeter and Submillimeter Wave Propagation and

Circuits, Proceedings No. 245, pp. 6-1 to 6-15; September, 1978.


W. Gabsdil and W. Jacobi, “Model Simulation of Target Characteristics and

Engagement Situations Employing Millimeter-Wave Radar Systems,” ibid., pp. 7-1 to 7-8.


R. Hartman, “Multisensor Warning Receiver Tackles Growing Threat,” Journal of

Defense Electronics, Vol. 2, pp. 79-86; May, 1979. (Also see Microwave System

News, Vol. 9, pp. 32, May, 1979.)


Bob Wollins, “Auto Anticollision Radar Accelerates Away from the Lab

Backburner to Global Test Sites,” Microwaves, Vol. 17, pp. 9-12; January, 1978.


F. Moncrief, “Car Radars Could be Standard in the 1980s,” Microwave System

News, Vol. 8, pp. 23-26; April, 1978.


D. Zur Heiden and H. Oehlen, “Radar Anticollision System for Road Vehicles,”

Electrical Communication, Vol. 52, pp. 141-145; February, 1977.


S. V. Bearse, “94 GHz Radar Undergoing Coast Guard Evaluation,” Microwaves,

Vol. 14, pp. 10; September, 1975.


S. J. Dudzinsky, Jr., “Atmospheric Effects on Terrestrial mm-Wave

Communications,” Microwave Journal, Vol. 18, pp. 39; December, 1975.


“Hughes Unveils New 38 GHz Radio,” Microwave System News, Vol. 6, pp. 13;

February/ March 1976.


H. Meinel, A. Plattner, and R. Breitschadel, “A 35 GHz Communication Link for

Railway Applications,” European Microwave Conference, Brighton, England;

September 19, 1979.


R. T. Davis, “Mm Transceiver Provides Covert Communications,” Microwaves, pp. 9, October, 1974.


Y. Matsuo, Y. Adaiwa, and I. Takase, “A Compact 60-GHz Transmitter-Receiver,”

IEEE Transactions on Microwave Theory and Techniques, Vol. MTT-24, pp. 794-

797; November,1976.


S. Becker, “Tunable Planar Hybrid Millimeter-Wave Transmitter and Receiver,”

Eighth DARPA/Tri-Service Millimeter Wave Conference, Eglin AFB, Florida;

April, 1979.


J. E. Keitzer, A. R. Kaurs, and B. J. Levin, “A V-Band Communication Transmitter and Receiver System Using Dielectric Waveguide Integrated Circuits,” IEEE

Transactions on Microwave Theory and Techniques, Vol. MTT-24, pp. 797-803,

November, 1976.


Y. Chang, J. A. Paul, and Y. C. Ngan, “Millimeter Wave Integrated Circuit

Modules for Communication Interconnect Systems,” Final Report No. DELET-TR-

76-1353-F on Hughes Aircraft Contract DAAB07-76-C-1353 with ERADCOM;

October, 1978.


R. W. White, M. B. Read, and A. J. Moore, “Recent British Work on Millimeter

Waveguide Systems,” IEEE Trans. Communication Soc. Vol. COM-22, pp. 1378-

1390; September, 1974.


K. Miyauchi, S. Seki, N. Ishida, and K. Izumi, “W-40G Guided Millimeter-Wave

Transmission System,” Review of Electrical Communications Laboratories, Vol. 23, pp. 707-741; July/August, 1975.


“Millimeter Waveguide Systems,” Microwave Journal, Vol. 20, pp. 24-26, March,



“Japanese W-40 G. Guided Millimeter-Wave Transmission System,” Microwave

System News, Vol. 6, pp. 91-97; February/March 1976.


W. D. Warters, “Millimeter Waveguide Scores High in Field Test,” Bell

Laboratories Record, pp. 401-408; November, 1975.


W. H. Steier, “The Attenuation of the Holmdel Helix Waveguide in the 100-125

GHz Band,” Bell System Technical Journal, pp. 899-906; May 1965.


T. E. Abele, D. A. Alsberg, and P. T. Huchison, “A High-Capacity Digital

Communication System Using TE


Transmission in Circular Waveguide,” IEEE

Trans. Microwave Theory, Tech., Vol. MTT-23, pp. 326-333; April 1975.


J. R. Mahieu, J. N. Marchalot, J. P. Boujet, G. I. Coz, P. Gibeau, and J. V. Bouvet,

“Technologies Avancees Developees Par l’Industrie Pour Les Etudes Sur le Guide

D’Ondes Circulaire,” Annales Telecommunications, Vol. 29, pp. 443-464;

September/October, 1974.


P. Bernardi, G. C. Corazza, G. Faliciasecca, R. Koch, G. Magni, G. A. Mastellari,

G. B. Stracca, and F. Vardoni, “Italian Experimental Equipment for High Bit Rate

Transmission on Circular Waveguide,” in Proc. 1974 European Microwave

Conference, pp. 619-623.


L. J. Ippolito, “Effects of Precipitation on 15.3 and 31.65 GHz Earth-Space

Transmissions with the ATS-V Satellite,” Proc. IEEE, Vol. 59, pp. 189-205; 1971.


J. L. King, J. W. Dees, and J. C. Wiltse, “A Millimeter Wave Propagation

Experiment From the ATS-E Spacecraft,” IEEE International Convention Record, pp. 248; March, 1968.


J. W. Dees, G. P. Kefalas, and J. C. Wiltse, “Millimeter Wave Communications

Experiments for Satellite Applications,” Proc. IEEE International Conference on-

Communications, San Francisco, pp. 22-20 to 22-26, June, 1970.


N. B. Hilson, J. J. Gallagher, et al, “Millimeter Wave Satellite Concepts,” Vol. 1,

Technical Report No. NASA CR-135227, NASA Lewis Research Center, Contract

NAS3-20110, September, 1977.


N. E. Feldman and S. J. Dudzinsky, Jr., “A New Approach to Millimeter-Wave

Communications,” Rand Report R-1936-RC, April 1977.


L. G. Mundie, N. E. Feldman, “The Feasibility of Employing Frequencies Between

20 and 300 GHz For Earth-Satellite Communications Links,” Rand Report R-2275-

DCA, May, 1978.


C.K.H. Tsao, W. J. Connor, and T. E. Joyner, “Millimeter-Wave Airborne Terminal for Satellite Communications,” International Conference on Communications, 11th, pp. 36-4 to 36-8, 1975.


A. Castro, and J. Healy, “Transmitter System for MM Wave Satellite

Communications,” International Conference on Communications, 11th, pp. 36-9 to

36-13, 1975.


The authors are D. M. Snider and D. B. Coomber, “Satellite to satellite Data

Transfer and Control”, Proceedings Seventh AIAA Communication Conference,

San Diego, pp. 457-470; April 23, 1978.


Stanley Becker


Applied Electronics Division

AIL Division of Eaton Corporation

Walt Whitman Road

Melville, New York


A communication link operating at 5 millimeters can be designed to take advantage of the properties of the oxygen absorption band to provide relatively interference free and secure communications. A tunable transmitter and receiver were developed to demonstate the adaptability of such a link to varying propagation and potentially hostile EMI conditions.


Communication link designs optimize the link parameters of radiated power, receiver sensitivity, antenna gains, and propagation losses for maximum signal-to-noise ratio at the receiver. When there is an added requirement to provide communications security, directive antennas are often used and radiated power is reduced. For some short-range applications, the spatial security offered by these antennas (a restriction of look angle) is not sufficient. High gain antennas do not limit the overreach of the transmitter emanations beyond the receiver terminal in line with the antenna main beam or in diverse directions from reflectors in the field of view.

Attenuation of a transmitter signal in free space is inversely proportional to the square of the range from the transmitter. Additional attenuation, increasing exponentially with range, is possible in frequency bands where molecular absorption of radiated power is prominent.

Such enhanced attenuation of radiated signal power will reduce the overreaching of the transmitter signal. Conversely, a receiver tuned to an incoming signal in an absorption band is less susceptible to radiated interference for a given interfering transmitter power and separation from the receiver. An experimental tunable transmitter and receiver * designed to operate on the skirt of the 5- mm oxygen line band is described, and some of the novel features that attend tunability in this band are described.

* The transmitter and receiver were built on Government Contract No. DAAB07-77-C-0118.


The signal power propagated to the receiver terminal of a communications link is:

(1) where P r

is the received signal power, P t

is the transmitted signal power, G transmitter antenna gain, G r the range, and " is the absorption factor (a function of 8).


is the

is the receiver antenna gain, 8 is the signal wavelength, R is

We see that received signal strength is inversely proportional to R 2 . When " is not zero, the signal strength is inversely proportional to a factor with a value of "R in its exponent.

The 5-mm band is characterized by an oxygen absorption band shown in Figure 1 (1). The absorption coefficient, ", is plotted in dB per nautical mile. Converted to dB/km, the maximum absorption is about 15 dB/km, obtained at the surface at a frequency of about

60.5 GHz. Figure 1 also shows how absorption decreases with increasing altitude, and how the individual resonance lines are evident when broadening due to atmospheric pressure is diminished. Rain is also a very significant attenuation factor at millimeter wavelengths (2). Figure 2 shows that the absorption coefficient at 60 GHz is somewhat less than 4 dB/NM or about 2 dB/km for a moderate rainfall.


By operating in the 5-mm band, terrestrial, airborne, and spaceborne communication links can be designed to minimize the effects of radiated jamming power and to reduce the undesired overreaching of a transmitted signal. The scenarios are different for links totally within or between each of the propagation media. In this paper, we will discuss the terrestrial link application.

Setting up a 5-mm terrestrial link over a given range requires consideration and allowance for: a) available radiated power b) modulation bandwidth c) receiver sensitivity d) practical aperture size e) propagation conditions

The highest power 5- mm solid-state sources that can be used for communications will provide up to about 0.2 W (InP Gunn) to 1.2 W (Impatt) in CW operation. With external-

cooling, the InP Gunn output power can be increased to about 0.4 W. Hard-tube sources with output power extending through the kilowatt range are possible using cooled TWT’s and magnetrons.

Modulation bandwidth and receiver sensitivity are intimately related. Depending on the application, the reliability of the link will dictate a given error rate, which will prescribe a minimum signal-to-noise ratio. For the purpose of this discussion, we will examine a 4-km link with a compatible modulation/receiver noise bandwidth of 200 kHz, receiver noise figure of 15 dB, signal-to-noise ratio of 10 dB, and a transmitter power of 50 mW.

If we were to set up a terrestrial link in a given geographic area, we would need to analyze the rainfall records in the area and compile statistics on rainfall rate versus probability of occurrence. Operation of the link during the day, night, or on a 24-hour basis would have to be factored into the analysis. Having compiled the rainfall data for the operational situation and with an acceptable percentage down-time decided upon, we can establish a maximum attenuation factor for rainfall. For this discussion we will use a maximum rainattenuation factor of 5 dB/km (moderate to heavy rainfall). In Washington, D. C., this rainfall would be exceeded about 10 hours per year or about 0.1 percent of the time.

If we select antennas that are well within the manufacturing state of the art and ones that can readily be aligned, we would limit ourselves to an aperture of about 12 inches. Such antennas would have a 3-dB beam width of about 1.1 degrees and a gain of about 43 dB.

The signal-to-noise ratio for the terrestrial link is:


At an operating frequency of 60 GHz we use the following values (expressed in the dB scale): F = 15 dB, kTB = -121 dBm, P t

= +17 dBm, G t

= G r

= 43.5 dB, R = 4 km, a = 15 dB/km, 8 = 5 mm, and we obtain an SNR of 10 dB in clear weather.


The degree of severity of overreach is a function of the sensitivity of the intercept receiver, or listener, and the propagation path between the intercept receiver and the link transmitter. For this example, we assume a receiver sensitivity/antenna-gain advantage of

10 dB for the intercept receiver. The overreach of the link signal in the direction of the main beam and in other directions, due to inadvertent reflection from surrounding terrain or reflectors, is a function of the factors 1/R 2 and 10 -"R as was previously shown. In a frequency band where " is insignificant, the interceptor could be at a range of more than

three times the link range; equivalent to an overreach of 8 km. However, a link operating at a frequency of 60 GHz would have an overreach of only 0.7 km.

The susceptibility of a 60 GHz receiver to interference is not inherently different from those in other bands, but the oxygen absorption band may be considered as a selective shield around the link receiver. By much the same way that an intercept receiver is at a disadvantage beyond the range of the link transmitter, an interfering transmitter must overcome an additional attenuation of 15 dB/km beyond the link range.


Operating the communications link in a moderate-to-heavy rainfall would result in an additional 5 dB/km or 20 dB additional loss over that for the free-space path. Such a condition would reduce the SNR in the example discussed to -10 dB. It is not practical to increase transmitter power, aperture sizes (for convenience), or receiver sensitivity to make up for such a loss of signal strength. It is feasible, however, to retune the link to a frequency of say 64 GHz, thereby reducing the oxygen-line absorption coefficient from

15 0dB/km to about 9 dB/km. Such a retuning would recoup 20 dB of signal attenuation over the 4 km link range, restoring the initial 10 dB signal-to-noise ratio. Furthermore, the susceptibility to detection and interference would be essentially the same as it was in clear weather.


The tunable 5-mm transmitter and receiver (3) that were built (Figures 3 and 4) incorporated several novel designs for the mixer, tunable oscillator and VCO (4 and 5). A summary of the component and system characteristics are shown in Tables I and II.

TABLE I. Millimeter-Wave Component Performance Characteristics

• Mechanically Tunable Oscillators

Frequency Range/Output Power

• Varactor-Controlled Oscillators

Frequency Range/Output Power

• Millimeter-Wave Mixer

Frequency Range

Conversion Loss

Noise Figure* (DSB)


* Including 4-dB IF amplifier noise figure

55 to 63 GHz/26 to 60 mW

64 to 70 GHz/20 to 40 mW

60 to 62.5 GHz/8 to 25 mW

63 to 66.5 GHz/3 to 25 mW

61 to 72 GHz

6.6 ±1 dB

10 dB

Dc to 12 Ghz

TABLE II. Transmitter and Receiver Performance

• Transmitter

Antenna (3-inch diameter)


Beam Width


Tuning Range

Low Channel

High Channel

Output Power

Low Channel

High Channel

Modulator Sensitivity

• Receiver

Antenna (3-inch diameter)


Beam Width


Tuning Range

Noise Figure

Predetection Bandwidth (3 Db)


Discriminator Sensitivity


4.5 degrees


60.8 to 61.5 GHz

64.3 to 65.8 GHz

22 ±6 mW

11 ±8 mW

20 MHz/volt


4. 5 degrees


5 GHz

10 to 15 dB

50 MHz

Up to 10 MHz*

0.2 volt/MHz

*Transmitter supplied with low-pass filter, reducing baseband to 100 kHz.


Some measurements of the effect of oxygen-line absorption were made in Melville, NY between a rooftop and line-of -sight points 1.0 and 1.5 km distant. Data were taken at 61 and 65 GHz on a clear day. Analysis of the measurement data show that a somewhat greater range cutoff factor was obtained than would have been predicted by the range equation given in this paper.

As the range is increased from 1 to 1.5 km, the theoretical difference in signal strength at

61 GHz is:


and at 65 GHz the differential is:


The field measurements showed a difference of 14.5 dB and 13 dB for the 61 and 65 GHz measurements, respectively. The greater than expected falloff of signal power cannot be explained except for possible site and environment anomalies, which were not measured or evaluated.


The use of a tunable transmitter and receiver in the 5-mm band can effect a decrease in susceptibility to radiated interference and a reduction of transmitter signal overreach. For a specified link reliability, the use of a tunable system can offset the effects of rain attenuation in a short-range communications application.

A tunable transmitter and receiver, developed for CORADCOM, were used in a field experiment to verify predictions of range cutoff as a function of frequency. The effects of increasing the range produced a greater than anticipated signal attenuation, but this might be attributed to anomalous site or propagation conditions.


The development of the tunable transmitter and receiver was performed for the U.S.

Army’s CORADCOM under the direction of E. Fliegler. Major contributors to the millimeter-wave design at AIL were L. Cohen and P. Meier. Assistance for the measurements described was given by D. Baird and S. Fung of AIL.


1. Vignali, J. A., “Millimeter Waves and Naval Communications,” NRL Report 7165,

10 December 1970.

2. Vignali, J. A. and Decker, K. M., “The Application of Microwaves to Low-Intercept

Short-Haul Communication,” NRL Report 6187, 8 February 1965.

3. Becker, S., “Tunable Planar Hybrid Millimeter-Wave Transmitter and Receiver,” DoD

ARPA Millimeter Wave Conference, Eglin AFB, FL, April 1979.

4. Meier, P. J., “4 and 5 mm Band Printed Circuit Balanced Mixers,” Microwave Journal,

August 1979.

5. Cohen, L. D. and Meier, P. J., “E-Plane Millimeter-Wave Circuits,” Microwave

Journal, August 1978.

Figure 1 - Atmospheric Absorption in the 60-GHz Region at Various Altitudes

Figure 2 - Attenuations Due to Rain at Sea Level, 68


Figure 3a - Tunable Millimeter-Wave Transmitter, Panel Side

Figure 3b - Tunable Millimeter-Wave Transmitter, Antenna Side

Figure 4a - Tunable Millimeter-Wave Receiver, Panel Side

Figure 4b - Tunable Millimeter-Wave Receiver, Antenna Side



C. Sun

Hughes Aircraft Company

Electron Dynamics Division

Torrance, California 90509


Recent trends in millimeter-wave solid state transmitter components are reviewed.

Specifically, the progress of the developments of IMPATT and Gunn Oscillators, Power

Combiners, phase and injection locking techniques and upconverters are presented.


In recent years, there has been a rapid growth of millimeter-wave system applications. This trend can be attributed to the decided advantages of using millimeter-wave systems over microwave and optical counterparts. The millimeter-wave systems require much smaller antennas and provide wider bandwidth than the microwave systems. In addition, the millimeter-wave systems provide better penetration capabilities through fog, clouds and dust than the optical systems. In this paper, recent progress and trends in the development of selected solid state transmitter components for millimeter-wave systems will be reviewed.


Figure 1 shows the state-of-the-art output power achieved at various frequencies with CW and pulsed double-drift IMPATT and Gunn oscillators. For IMPATT diodes, pulse powers of 25 W at 35 GHz, 15 W at 94 GHz, and 520 mW at 217 GHz have been obtained. CW powers of 2.3 W at 40 GHz, 1 W at 94 GHz, and 50 mW at 220 GHz have also been achieved. The peak power shown in Figure 1 is measured at less than 1% duty factor with

100 ns pulse width. With wider pulse widths and increased PRF, peak power output gradually decreases toward the value of CW operation. Below 100 GHz, the power output follows I/f characteristic, indicating that the device power is thermally limited. Beyond 100

GHz, the power output follows I/f 2 characteristic, implying that the device is limited by device-circuit impedance matching. Since IMPATT oscillators provide much higher output power than Gunn oscillators, INPATT sources are better suited for high power transmitter

applications. On the other hand, GaAs Gunn devices have lower AM noise, therefore, they are more useful for low noise, low power transmitter systems.


To achieve higher power output than a single diode can provide, power combining techniques are currently being developed to accumulate power from a number of IMPATT diodes for millimeter-wave applications. Two approaches have been pursued to achieve higher power. For systems where the bandwidth is important, millimeter-wave oscillators and amplifiers are combined using 3 dB hybrid couplers. The development in this area has been concentrated at 60 GHz for covert communications. Figure 2 shows a two-stage

V-band amplifier using four diodes in the output stage. Using silicon double-drift diodes on diamond heat sinks, output power of 1.5 W from two diodes and 2.4 W from four diodes have been obtained at 60 GHz.

The second approach is a resonant cavity combiner configuration first suggested by

Kurokawa (1). The advantage of this approach is the ability to combine a large number of diodes to achieve high power. The disadvantage of this approach is that the resonant cavity limits the bandwidth of the combiner to less than 1%. Figure 3 shows a four-diode combiner which produced over two watts of CW output power. A development program is currently underway to achieve 100 W peak output power at 94 GHz. Already 20 watts peak power from two diodes, and 40 watts peak power from four diodes have been achieved.


For certain system applications, the FM noise of the oscillator close to carrier must be kept low to ensure receiver sensitivity. To meet these requirements, phase and injection-locking techniques have been developed. Figure 4a shows the block diagram of a 94 GHz Gunn oscillator phase-locked to a crystal oscillator reference signal. The measured phase noise characteristics of the phase-locked oscillator is shown in Figure 4b. Significant improvement in phase noise can be seen within the locking band. The locking bandwidth is determined by the locking loop to typically 1 to 10 MHz.

Another method of achieving a low-phase noise spectrum is the injection-locking technique with a low noise signal input. One way of achieving a high CW output power and low FM noise simultaneously is to injection lock a CW IMPATT oscillator with a phase-locked oscillator as shown in Figure 5. Also shown is the spectrum of a CW

IMPATT source as a free running and injection-locked oscillator. It appears that the FM noise of the oscillator is improved by means of injection-locking.


In many millimeter-wave transmitter systems, upconverters are used. The upconverter receives an IF signal input f signal outputs at f


± f



and a RF millimeter-wave signal (f

. The lower sideband, f


- f



); and it generates RF

is rej ected by the bandpass filter connected at the output. The performance of the single sideband upconverter is shown in

Figure 6. A maximum power output of 7.5 dBm can be obtained with +18 dBm IF input power and +17.3 dBm RF input power; both single and balanced configurations are used.

The single-ended configuration offers a simple design, whereas the balanced configuration provides greater dynamic range.


The progress in the development of some millimeter-wave solid state transmitter components has been reviewed in this paper. With demonstrated high power capabilities and noise reduction techniques in these components, potential applications in the area of telemetry at millimeter-wave frequencies can be realized.



Kurokawa, K. and F. M. Magathaes, “An X-band 10-Watt Multiple-IMPATT

Oscillator”, Proc. IEEE, Vol. 59, No. 1, January 1971, PP 102-103.







SCALES 0.5 MHz/DIV., 10dB/DIV.






F. Bernues and P. Pusateri

Hughes Aircraft Co.

Electron Dynamics Division

3100 W. Lomita Blvd.

Torrance, CA 90509


This paper presents an overview of the state-of-the-art performance of mm-wave receiver front-end components. Topics covered include filters, diplexers, mixers (broadband and narrowband) and local oscillators (free-running and phase-locked). Examples of mm-wave receiver front-end configurations are given and performance tradeoffs are discussed.


Much of the impetus behind the successful development of mm-wave radar, communications and EW systems is due to the availability of small, rugged receivers with outstanding performance characteristics. The receiver designer today can optimize system tradeoffs by selecting the component technology best fit to meet the system performance requirements. Four types of components will be discussed in this paper: filters, frequency diplexers, balanced mixers and local oscillators.

Radar receivers are usually narrowband (IF bandwidth 500 MHz or less) and the main performance requirement is minimum noise figure. Some design tradeoffs are required for coherent and monopulse receivers, where requirements such as LO phase noise, frequency stability and channel-to-channel amplitude and phase tracking have to be met. EW receivers present a different problem; wherever signal frequency identification is needed, frequency accuracy and stability of local oscillators, instantaneous RF and IF bandwidth and intermodulation characteristics are important. Channelized receivers can be designed with channelized RF, channelized IF, or both. The general tradeoffs are:

1. Receivers with broader instantaneous RF and IF bandwidths (4 GHz or more) have higher receiver noise figure (partly due to higher mixer conversion loss and higher IF preamplifier noise figure) and degraded spurious performance (third order intermodulation and IF harmonics), but lower overall cost and volume, because of the need for fewer mm-wave mixers to cover a given band and the lower cost and complexity of IF diplexers.

2. Receivers with IF instantaneous bandwidths of 4 GHz or less have better noise figure and spurious performance, but if a large number of channels are needed to cover a given frequency band, the choice may be between RF diplexers and availability of space for several antennas (one antenna per channel), both of which result in increased cost and complexity.

Hybrid designs incorporating some RF diplexing and some IF diplexing present the optimum tradeoff in most cases. The selection of optimum channel bandwidth is critically dependent on several parameters: a) RF diplexer loss and isolation versus bandwidth b) Mixer noise figure versus IF bandwidth c) Mixer intercept point d) IF preamplifier skirt selectivity e) LO frequency stability f) Mixer LO-to-RF isolation g) IF diplexer cross-over characteristics

A good knowledge of component performance characteristics is needed to select the optimum receiver design. In the remainder of this paper we present an overview of existing mm-wave receiver component technology and we indicate trends and directions for the future.

Receiver front-end components, specifically filters and mixers, is one area where transmission media other than metallic waveguide are likely to challenge the present dominance of metallic waveguide components, if not in electrical performance at least in cost and complexity. Indications are that suspended stripline, fin-line, dielectric waveguide and quasi-optical techniques are likely to offer selective advantages for specific frequency ranges and performance levels. Further development efforts are required to elucidate potential advantages and disadvantages. It is also very likely that hybrid components will emerge integrating several transmission media in different parts of the circuit, specially for mixers. We believe that it is premature to dismiss metallic waveguide or to adopt simplistic arguments favoring one or the other media, particularly when the range of frequencies and systems needs to be covered is as broad as it is, and continually expanding.


Waveguide filters and frequency diplexers have been designed and fabricated at frequencies up to 110 GHz. Three types of waveguide filters exist: a) Low-pass filters, of the waffle-iron type, are limited to an upper frequency of 40 GHz due to limitations in fabrication techniques.

b) High-pass filters are based on waveguide cutoff properties. By careful design and fabrication of tapered transitions, insertion losses can be kept under 1 dB at 60 GHz and skirt selectivity is of the order of 30 dB at 200 MHz from band edge.

c) Bandpass filters are based on the classical iris-coupled cavity technique. Performance characteristics of a narrowband design at 94 GHz is given in Figure 1. The measured performance of two W-Band broadband filters is shown in Figure 2 (dashed line).

Insertion losses of 0.6 dB for a seven section filter at Ka-Band can be achieved by using electroforming fabrication techniques. Typical losses for frequencies between

30 GHz and 100 GHz can be estimated by extrapolating between the values shown.

Waveguide diplexers consist of a Y junction with two filters properly spaced along the colinear arms. The out-of-band rejection is determined by the filters and the insertion loss is the sum of three components: filter loss, ohmic loss of the junction and mismatch loss.

Contiguous-channel diplexers have relatively large mismatch loss at the crossover frequency, as shown in Figure 2 (solid line). A photograph of a W-Band diplexer is shown in Figure 3. Typical performance data for a Ka-Band non-contiguous diplexer is shown in

Table I.


Substantial development efforts in recent years involving both mixer diodes and circuits have resulted in the availability of new types of devices and the across-the-board improvement of critical mixer performance parameters, specially noise figure.

Three types of Schottky-barrier diodes are being used in mm-wave mixers: honeycomb, notchfront and beam-lead. Honeycomb diodes (Figure 4) are used with waveguide mixers and have the best noise figure performance potential, particularly above 100 GHz. The notch-front diode (Figure 5) is a variant of the honeycomb diode designed to fit the package requirements of suspended stripline circuit configuration. It has, however, one additional advantage in that it inherently reduces series resistance by reducing skin effect loss. Basically the notch-front diode is a honeycomb diode chip turned on its side.

Notch-front diodes offer significant advantages for operation above 100 GHz when used with subharmonically pumped mixers. They can also be used with fundamentally pumped mixers with suspended stripline IF output circuits.

Beam-lead diodes are whiskerless planar devices with a surface oriented configuration from which two beams extend for contact to an RF circuit (Figure 6). They are particularly well suited for fin-line, suspended stripline and dielectric waveguide circuits. Their performance capabilities below 100 GHz are comparable to that of honeycomb or notchfront diodes, whereas above 100 GHz the parasitic capacitance may result in higher noise figure.


Performance of a Ka-Band Non-Contiguous Diplexer

Channel 1 band-pass:

Filter insertion loss:

Filter rejection:

Diplexer insertion loss:

Diplexer rejection:

Channel 2 band-pass:

Filter insertion loss:

Filter rejection:

Diplexer insertion loss:

Diplexer rejection:

34.2 GHz to 35.4 GHz

<0.4 dB

>50 dB

0.8 dB

>50 dB

36.5 GHz to 38.6 GHz

<0.4 dB

>50 dB

<0.8 dB

>50 dB

Waveguide narrowband balanced mixers have been in operation at frequencies up to

140 GHz for several years. Figure 7 shows a 94 GHz mixer with 5.6 DSB NF (including an IF preamplifier with 10 MHz to 1010 MHz bandwidth and 3.5 dB NF), with a bandpass filter in the RF arm. Units similar to this one have been used in space programs and have been subjected to rigorous as well as environmental tests (e.g., 10000 g’s centrifugal acceleration on three axis for 30 seconds each).

Broadband balanced waveguide mixers can be implemented with the configuration shown in Figure 8. This particular unit was designed to be used with a scanning 50 GHz to

75 GHz local oscillator with 10 MHz to 500 MHz fixed IF bandwidth.

Other designs involving fin-line and dual-mode mixers are capable of achieving up to

20 GHz instantaneous IF bandwidth with a fixed local oscillator frequency. The

performance of these mixers above 100 GHz is critically dependent on the cutoff frequency and circuit parasitics associated with beam-lead diodes.

Figure 9 shows a prototype subharmonically pumped mixer at 94 GHz (with a local oscillator frequency of 47 GHz). While capable of excellent noise figure performance at frequencies up to 180 GHz, the ultimate advantages of this design depend on the feasibility of developing practical fabrication and assembly techniques. The lack of suitable local oscillators operating above 110 GHz makes this design very attractive for operation in the

110 GHz to 200 GHz range.

Above 180 GHz, quasi-optical techniques offer the best promise. Figure 10 shows a quasioptical 217 GHz single-ended mixer with an IMPATT local oscillator. A conversion loss of 10.6 dB was measured for this device.

Both dielectric waveguide and quasi-optical mixers are being designed to operate as subharmonically pumped mixers (i.e., with the local oscillator frequency equal to 1/2 or

1/4 the nominal frequency). Particular promise is shown by dielectric waveguide mixers in the 60 GHz to 200 GHz range for applications requiring large instantaneous IF bandwidth.

The third order intemodulation performance of a typical mm-wave balanced mixer/IF amplifier is shown in Figure 11. The 1 dB compression point is in the 0 dBm to +10 dBm

IF output power range (with 25 dB RF-to-IF gain), and is primarily determined by the IF amplifier saturation characteristics. This is particularly relevant for broadband mixers with

IF amplifiers in the 2-4 GHz or 4-8 GHz range.


The main requirement of mm-wave local oscillators is low AM noise. GaAs Gunn devices are used at frequencies up to 110 GHz and IMPATT oscillators (with suitable filters) can be used up to 220 GHz. The measured AM noise of a 94 GHz Gunn oscillator is shown in

Figure 12. It should be noted that oscillators with higher AM noise can be used in conjunction with mixers with good LO noise suppression (35 dB or better).

Phase-locked local oscillators are used for both frequency accuracy (EW receivers) and low phase noise (coherent radar receivers). Figure 13 shows the measured FM noise of a

94 GHz phase-locked Gunn oscillator. The dashed line shows the theoretical minimum noise, as determined by the 100 MHz crystal reference oscillator, and the solid line shows the measured FM noise.


Sophisticated mm-wave receiver front-ends have been designed and fabricated with performance parameters not significantly different than those achievable at lower microwave frequencies. Further development efforts will result in improved bandwidth and noise figure performance through the use of novel circuit and diode techniques.

























Senior Scientist

TRG Division, Alpha Industries

Woburn, Massachusetts


Recent developments in the technology for millimeter wave receiving equipment make a much more promising case for increased utilization of millimeter waves in telecommunication links. Low system noise figure coupled with large achievable antenna gain in small earth terminals make a good case for millimeter direct satellite broadcast links. Future technology trends indicate that use of the 80-100 GHz region of the spectrum will be beneficial and useful in the 1985-1990 timeframe.


For many years, millimeter wave receiver technology remained for the most part within the jurisdiction of University laboratories and a few government research institutes. This is no longer the case, with major emphasis on millimeter wave receiver and component technology becoming a major factor in both commercial and DOD planning for the next decade and beyond. Coupled with this new interest, is a series of major advances in Solid

State component design, particularity in the area of IMPATT, GUNN and GaAs Schottky

Mixer performance at millimeter wavelengths. Although transmitter sources at millimeter wavelengths tend to be somewhat lower power than what is commonly used in microwave links, the additional antenna gain resulting from modest aperature area at the shorter millimeter wavelengths makes systems both practical and cost effective. Receiver technology in the near term up through 100 GHz is comparable to that which can be attained at X band and below, and promises to be much better from the standpoint of instantaneous bandwidth in the 1985-1990 timeframe.


Unlike previous receiver technology, which centered around the use of whisker-contacted diodes mounted in sections of reduced-height waveguide to enable low conversion loss mixing and downconverting from millimeter wavelengths, new designs currently in

production utilize beam-lead Schottky Barrier devices. (1) Use of advanced beam lead diode technology has allowed room temperature downconverters at 94 GHz to be designed and manufactured with less than 8 dB Single Sideband Noise Figure including the IF and back-end contribution. Such performance is typically attained using a low-noise GUNN effect local oscillator and a FET IF amplifier. Bandwidths of greater than 2 GHz can be obtained at frequencies above 75 GHz. Similar performance and a high degree of reliablility can be obtained to at least 140 GHz in near term production-capable designs.

Present beam lead Schottky technology allows for conversion loss in ruggedized commercial mixers which is perhaps 0.5 to 1.5 dB greater than the minimum which can be obtained over the 40 - 140 GHz frequency range. Results very near the theoretical limit for room temperature devices can now be achieved routinely at 35 GHz. Advances in room temperature FET IF amplifiers allow the IF bandwidth to reach several GHz in practical mixer designs making millimeter wavelength receivers even more attractive from the standpoint of spectrum-available bandwidth.

A performance summary of many units developed by the TRG Division of Alpha

Industries, appearing in Figure One, points to the general slope of the mixer family over the millimeter frequency range. These are among the lowest noise figures ever achieved in practical production mixers at these frequencies. As is evident, no real performance difference exists for either fundamentally pumped, or subharmonically pumped balanced mixers. All noise measurements have been accomplished without the use of Local

Oscillator injection filters and have been independently verified by outside contractors in the systems-industry. The measured response of one such 94 GHz mixer, illustrated in

Figure 2, where a broadband 94 GHz noise source with a 12.7 dB ENR, produced a 8 to 9 dB Y factor over the entire 100-1000 MHz IF bandpass used in that design clearly illustrates the broad bandwidths over which low-noise performance can be realized. Less than 5 dB total system noise figure can be attained in room temperature mixers made using these technologies at a cost of less than $2000.00 in 1979 dollars for the entire mixer/ preamplifier when considered in large quantities (1000 pieces or more).

At 35 GHz several other technologies are catching up with resistive mixers such that over the near term it will be practical to field cooled mixers, cooled FET preamplifiers and parametric amplifiers with noise figures all less than 3.0 dB SSB. As is always the case various tradeoff’s between the technologies will result in the most practical and cost effective approach being accepted for telecommunications applications. These different approaches are summarized in Table 1, where it may be evident that in the near term the cooled mixer preamplifier is the most practical and highest performance approach to millimeter receiver implementation.









35 GHz




Room Temperature Mixer

Cryogenically Cooled Mixer

Room Temperature FET

Parametric Amplifier

Room Temperature Mixer

Cryogenically Cooled Mixer


5.5 dB

3.0 dB

5.0 dB

3.0 dB

7.5 dB

4.5 dB


1 GHz BW

1 Ghz

5 Ghz

0.5 GHz

2 Ghz

2 GHz

Such cooled mixer preamplifiers will utilize cooled FET IF amplifiers in the 4 to 8 GHz range for optimum overall noise performance when cooled. Typical Local Oscillator requirements for 35 GHz mixers are in the range of 1-2 mW at the LO port of the mixer, or closer to 5 mW when coupled into a cryogenically cooled mixer through appropriate waveguide circuits.

Cooling may be accomplished through the use of closed cycle refrigeration systems or through the use of multi-stage thermoelectric coolers. One such configuration for use at millimeter wavelengths is outlined in Figure Three, where the feedhorn is also cryogenically cooled to reduce the system noise figure to the lowest practical limit.


In the latter part of the next decade, several technologies now being investigated in various research establishments, may alter the most cost effective approach for telecommunications receiver implementation. As is evident in Table 2, Cooled FET preamplifiers at 35 GHz will be low noise, and probably broader in bandwidth than

comparable cooled mixers. Additionally, the RF gain provided by the FET preamplifier is advantageous from many standpoints in overall system design. Cooling of the FET device somewhere in the range of 60-100K is probably required for optimum operation at long millimeter wavelengths.




3 5 GHz




Cooled FET

Cooled Mixer

Cooled Mixer

Josephson Mixer **

Cooled 3 Terminal



2.5-1.5 dB SSB

3.0-2.0 dB SSB

4-6 GHz

2-4 GHz

5.0-3.5 dB SSB

1.0-0.5 dB SSB

2-4 GHz

4-6 GHz

3 dB SSB ??

** requires more complicated cooling system with cold temperatures near to 4K while all other cooled systems utilize easy to attain cooling at 60-100K.

It is also possible that the use of a cooled beam-lead varactor in conjunction with a FET preamplifier as a reactive downconverter may offer some hope of reducing the attainable noise figure well below 3 dB at 86 GHz at some point in the next decade.

As is evident from the photograph below, work is already well along on integrated millimeter mixer/ FET preamplifers for space communciations at millimeter wavelengths.

The present results indicate a bright future for millimeter communciations with a large payoff in increased channel capacity, and most importantly making possible the development of a high volume consumer direct broadcast system for use by the average citizen in the late part of the next decade.

A 40-50 GHz subharmonically pumped mixer with integrated FET IF amplifier is shown prior to final assembly. This device is a prototype for next-decade telecommunications systems at millimeter wavelengths.

1. Cardiasmenos, A. G. “Planar Devices make Production Practical” Microwave Systems

News May 1979, Vol 9 No. 5



Srini Raghavan and Jerry Armes

Spectrum Analysis & Frequency Engineering

Richardson, Texas


A major telecommunications growth area at this time is narrowband digital transmissions via satellites. With the availability of low cost 5 and 10 meter earth stations, and readily available digital ground equipment, this trend can be expected to continue for some time.

With the degree of frequency congestion which exists currently in the 6 GHz band, the frequency coordination of these earth stations will become more and more difficult. Since the data rate is generally 56 KBPS or 1.544 MBPS, the satellite uplink carrier frequency is often selected to give a degree of isolation from standard frequency plans used by the terrestrial common carriers. The amount of offset advantage in db to be conceded from a given frequency separation between satellite and terrestrial carriers is a matter of controversy however.

This paper describes a computer program written to provide the necessary calculations, the underlying models, and results in the form of parametric curves which can be used directly to obtain the offset advantage for a given carrier separation.









Nationwide Television Programming for the Public Broadcasting Service is now being carried by the largest integrated network of Earth Stations ever constructed by a single turn-key contractor. The 165 station network, including 5 remote origination transmit terminals, was completed significantly ahead of a schedule established two years before.

Such an accomplishment required carefully orchestrated efforts between customer and contractor in the areas of system and equipment design and qualification, site selection and design, frequency coordination, equipment scheduling, on site construction, site installation and test effort mobilization.

Unusual and unexpected constraints were often encountered, requiring resourceful and innovative solutions. Performance records since early 1978 when initial operations began indicate the reliability, availability and cost improvement goals of the system have been exceeded by a significant margin.

A similar system for the National Public Radio, involving 205 stations (15 with transmit capability) is currently in the early stages of implementation.


Giuseppe G. Russo Paul R. Hartmann

Collins Transmission Systems Division

Rockwell International

Richardson, Texas, U.S.A.


Rapid growth and modernization of microwave communications is currently taking place in both the commercial and military environments. Much of this growth is being accomplished through digital transmission, and calls for efficient utilization of frequency spectra allocated for LOS communications. As the bit efficiencies and data rates of digital radio systems have increased, the problems associated with multipath propagation phenomena have become more evident. This is due to the fact that the transmitter signal is randomized before transmission and the system behaves as though it has full loading at all times. To minimize outages and to meet path availability objectives, diversity protection alone, in many instances, is not sufficient. However, diversity protection, coupled with an acaptive equalizer, will meet the objective in all but a few severe cases.

This paper describes a digital radio system operating in the 8 GHz band with 8 PSK modulation and nominal data rates of 45 Mb/s or 90 Mb/s. Multipath propagation and its effects on digital transmission are discussed. Improvements in bit error rate threshold obtained through the use of an adaptive equalizer designed into the receiver are also presented.


The recognized benefits of digital transmission include: transmission quality, ease of encription, reduced system cost, and direct interface with digital switching systems. These factors constitute some of the driving forces that account for the rapid expansion of digital transmission. This modern, high-performance digital microwave radio, MDR-8-( ), uses

8 PSK digital modulation at RF, and can be configured to transmit one or two 44.736 Mb/s

DS-3 signals in a 20 MHz or 40 MHz RF bandwidth on a single polarization. These configurations correspond to 672 and 1344 PCM voice channels, respectively. When auxiliary channels and overhead bits are multiplexed with the DS-3 data stream, the spectrum efficiency is 2.26 bits/Hz.

Frequency diversity or hot stand-by protection configurations are easily implemented, due to the straightforward RF architecture. To compensate for distortion effects due to multipath propagation, the receiver uses an IF amplifier containing a recently-developed adaptive equalizer.


The MDR-8-( ) microwave digital radio operates in the 7.125 - 8.500 GHz band, and can be configured for frequency diversity or hot standby protection. The system is all solid state, except for the high efficiency TWT power amplifiers. The general characteristics are shown in table 1. The MDR-8-( ) rack structure, shown in figure 1, contains two transmitters, two receivers, the DS-3 interface, the order-wire, and protection equipment.

The mechanical configuration of the MDR-8-( ) features front access for all modules, card cages, waveguide, and rack wiring, and removable front covers. The following description is of the MDR-8-5N narrow-band system; the wide-band system is an easy extension of the one described. [1]

Figure 1.






As shown in figure 2, the bipolar signal at the 44.736 Mb/s data rate is applied to the DS-3 interface unit. The primary function of the interface is that of conditioning the DS-3 input signal to the required MDR-8-5N system format. This unit is used only at terminals and at repeaters with drops. The transmitter portion converts the bipolar signal into an NRZ signal, removes the bipolar three zero substitution (B3ZS) encoding, and recovers the clock. Bit stuffing techniques are used for the system framing code and auxiliary channel bit locations. Three data lines plus the clock signal constitute the output of the DS-3 interface which is then applied to the RF transmitter. The receiver portion of the DS-3 interface receives the three data lines and the recovered system clock. The data lines are then combined and the overhead bits are removed (destuffing). The B3ZS coding and bipolar format are restored to meet the DSX-3 interface specification at 44.736 Mb/s.

Figure 2.




The MDR-8-5( ) transmitter is similar to a conventional remodulating microwave radio.

The input signal to the transmitter comes from the DS-3 interface for a terminal or repeater with drops, and from the receivers decoder for a repeater without drops.

The block diagram of the MDR-8-5( ) transmitter is shown in figure 3. The auxiliary channel insert module inserts fault alarm and service channel information into the data streams, and provides protection for the radio frame. The encoder accepts the three data lines and clock from the auxiliary modulator. To prevent generation of line spectra, the data is scrambled and then differentially encoded. The output of the encoder is applied to the 8 PSK modulator. The modulator is a path length modulator, where the output of the

RF source is phase modulated by the encoded data. The output of the modulator is then amplified in the high efficiency traveling wave tube amplifier. The amplified signal is then applied to the transmit filter for spectrum shaping. The longterm stability of the transmitter spectrum shape is achieved by using a sealed 5-element cylindrical cavity filter made of

Invar. The transmitted RF spectrum is shown in figure 4, together with the Federal

Communications Commission spectrum limits for a 20 MHz wide RF channel.

Figure 3.


Figure 4.



The digital receiver is a conventional superheterodyne receiver with a low-loss, sixelement cylindrical cavity preselector filter and low-noise single balanced hybrid mixer to achieve a low overall noise figure. A block diagram of the receiver is shown in figure 5.

The RF input signal is coupled through the preselector filter into the down converter mixer.

The down converter module also contains an AGC amplifier to extend the dynamic range of the receiver. The 70 MHz output of the mixer is then coupled to the IF amplifier.


The 70 MHz IF amplifier provides AGC amplification, system delay equalization, and establishes the receiver noise bandwidth. The AGC amplifier provides 55 dB of receiver dynamic range with a constant output level. In addition, the IF amplifier has an adaptive amplitude equalization capability that can automatically correct for linear slope and inband null resulting from multipath propagation.

Clock recovery, carrier recovery, and data detection are accomplished in the 8 PSK demodulator, together with prediction of RF channel performance obtained through an eye pattern monitoring circuit. Decoding and descrambling of the data streams occurs in the decoder unit to restore the three data lines applied to the digital transmitter input. The auxiliary channel drop module locates and extracts the twelve 32 kb/s auxiliary channels and makes them available to the service channel and fault alarm equipment.


The 1:1 switch and logic for the MDR-8-( ) is integrated into the radio bay. The switching logic provides 12 levels of receiver priority and 8 levels of transmitter priority. It provides error-free transfer of the receiver output signal for diversity and maintenance. The system is configured so that conversion to 1:N can be accomplished with minimum changes.


Microwave multipath propagation is a metereological phenomenon that has been recognized as a source of fading for many years. It arises from destructive interference between the primary signal ray and one or more reflected rays that arrive at the receiver antenna delayed in time from the primary signal. The path geometry for refraction due to a single layer is shown in figure 6. Although the multipath propagation is a complex phenomenon, the use of a two-ray propagation mod l for laboratory studies has been proven to be a fair representation of this phenomenon. [2,3] During multipath conditions, deep fading arises from the fact that, at some frequency, the primary and reflected rays arrive out of phase. Amplitude and delay patterns for two-ray interference are shown in figure 7. Deep amplitude nulls occurring in the proximity of, or within the RF channel constitute the greatest problem.

Figure 6.




Figure 7.





The degradation of BER on a digital system during multipath depends on the magnitude of the resulting passband distortion which, in turn, depends on the fade depth and the time delay between multipath signals. This degradation is caused by intersymbol interference resulting from frequency-dependent gain and group delay. Laboratory studies have shown that, for a given echo delay and fade depth, gain slope causes more severe BER threshold degradation than delay slope or parabolic delay. [4]

The adaptive equalizers used in the MDR-8-( ) radio system are quite effective in controlling multipath, since they compensate directly in the receiver for passband distortion created by multipath. [5] This compensation is made possible due to the broad and well-defined pseudorandom digital spectrum transmitted after scrambling the data.


System BER threshold improvement using adaptive equalizers with multipath simulated by a two-ray model is shown in figure 8, where the effects of amplitude slope and in-band null are shown separately. The results show that, under multipath conditions creating amplitude

slope or amplitude null, the system with the adaptive equalizer off, for most fades, is inoperative but, when the adaptive equalizer is turned on, the BER threshold performances under the same conditions are brought well within the BER specifications of the radio with no multipath.

Figure 8.







Figure 9.





The critical notch depth vs. offset frequency for a constant BER is shown in figure 9. This

“W” shaped curve shown the sensitivity of the system when a multipath disturbance (null) sweeps across the bandpass. The improvement ( decrease in sensitivity to multipath) is

measured by the offset frequency required and by the additional fade necessary to achieve the same level of BER measured without the adaptive equalizer. [6]

The adaptive equalizer used in the MDR-8-( ) is incorporated into the,70 MHz IF amplifier. It consists basically of two slope monitoring circuits tuned near the edges of the passband and a null detector tuned to the center of the passband. Utilization of these circuits reshapes the passband spectrum close to the non-multipath response.


The MDR-8-( ) microwave radio system just described is intended for use in short and medium haul applications, and is a high-performance, high-capacity system. The use of adaptive equalizers to counteract the detrimental effects of in-band distortion due to multipath propagation will improve the systems availability and performance margin.


Frequency Range

Transmit Stability


Voice Circuits

Data Input/Output

System Data Rate

Receive Noise Figure

Receiver Level

Receiver Threshold

(for 10


) BER)

7.125 - 8.500 GHz





One DSX-3

45.129 Mb/s Nominal

8.8 dB Maximum (2)

-8 dBm Maximum

-70 dBm Maximum (2)

Power Output




System Gain

(@ BER = 10



(1) North American Standard

(2) At Branching Circulator



+30 dBm

+35.5 dBm

105.5 dB

107 dB



+33 dBm min.

+38.5 dBm min.

108.5 dB min.

110 dB typ.


1. P.R. Hartmann, J.A. Crosset, “A 90 Mb/s Digital Transmission System at 11 GHz using 8 PSK Modulation.” Record of ICC, pp. 18.8 - 18.12, Volume II, 1976.

2. P.R. Hartmann, E.W. Allen, “An Adaptive Equalizer for Correction of Multipath

Distortion in a 90 Mb/s 8 PSK Microwave System.” Record of ICC, pp.5.6, Volume

I, 1979.

3. B.C. Ruthroff, “Multiple-Path Fading on Line-of-Sight Microwave Radio Systems as a Function of Path Length and Frequency.” Bell System Tehnical Journal, pp. 2375 -

2398, Volume 50, No. 7, September 1971.

4. M. Ramadam, “Availability Prediction of 8 PSK Digital Microwave System During

Multipath Propagation.” Record of ICC, P. 32.4, Volume II, 1979.

5. T.S. Giuffrida, “Measurements of the Effects of Propagation on Digital Radio System

Equipped with Space Diversity and Adaptive Equalization.” Record of ICC, p 48.1,

Volume 3, 1979.

6. C.W. Lundgreen, W.D. Rummler, “Digital Radio Outage due to Selective Fading -

Observation vs. Prediction from Laboratory Simulation.” Bell System Technical

Journal, pp. 1073 - 1100, Volume 58, No. 5, May-June 1979.


Data-Contol Systems, Incorporated




The Boeing Company

Seattle, Washington



Martin Marietta Corporation

Denver, Colorado






Rockwell International

Seal Beach, California


Dept. of Army

White Sands Missile Range

New Mexico


Harris Electronic Systems

Melbourne, Florida


Naval Air Systems Command

Washington. D.C.


Sandia Laboratories

Albuquerque. New Mexico


Grumman Aerospace Corporation

Calverton., New York


Jet Propulsion Laboratory

Pasadena, California


Teledyne Telemetry

Los Angeles, California


Marshall Space Flight Center - NASA

Huntsville, Alabama


Goddard Space Flight Center - NASA

Greenbelt, Maryland


Naval Missile Center

Point Mugu, California



Introduction. The Telemetering Standards Coordination Committee (TSCC) is chartered to serve as a focal point to represent the Telemetering Community. It serves to: a) receive, coordinate and disseminate information and b) recommend and endorse standards, methods and procedures to users, manufacturers and supporting agencies.

The TSCC is organized to: a) determine what standards are in existence and published, b) determine the technical adequacy of planned and existing standards, c) determine the need for additional standards and d) direct the derivation of new standards.

To perform these tasks the committee is composed of 14 members with approximately equal representation from industry and from government and/or not for profit organizations.

The TSCC was organized in 1960 under the sponsorship of the National

Telemetering Conference (NTC). Since 1967 the TSCC has been under the sponsorship of the International Telemetering Conference (ITC) and the Instrument Society of

America (ISA). The Annual Reports of the TSCC, since the year 1967 can be found in the NTC and the ITC Conference Proceedings.

Committee Activities. This year’s single meeting of the Telemetering Standards

Coordination Committee was held in Los Angeles on November 13, 1978 to coincide with the opening of the ITC Symposium. Ten members and alternates were in attendance at the meeting. The first order of business to be conducted was the elections for renewal of membership terms and election of committee officers. D. R. Andelin of

Rockwell International was nominated and unanimously accepted for a five year membership term. Messrs. McAnally and Trevathan were unanimously noted to new five year membership terms. Verne Jennings was nominated and unanimously elected to the position of Committee Vice-Chairman. Claude McAnally was unanimously elected to the position of Secretary-Treasurer.

Pursuant to the Committee’s Charter responsibilities to determine the need for new standards, discussions were conducted concerning the field usage on the following subjects: a) pseudo-random noise (PRN) generators for NRZ PCM transmission and magnetic tape recording systems to minimize the effects of non-DC response systems.

PRN provides an alternative to the use of DM or Bi-Phase codes more commonly used; b) constant bandwidth (CBW) FM formats utilizing subcarriers scaled from those published in the 106 document have been used to obtain additional channels of lower frequency data, or channels of greater frequency response than is presently possible. The

Committee considered recommending binary expansion of the standards to include

1 kHz deviations, as well as 16 and possibly 32 kHz deviations.

Joe Ramos, Chairman of the Range Commanders Council (RCC) Data Multiplex Committee, and Ken Scoeck, Member of the RCC Recorder/Reproducer Committee discussed with the Committee various changes being considered for inclusion in the next issuance of the IRIG 106-77 Standards Document.


Armes, J. . . . . . . . . . . . . . . . . . . . 639

Arndt, G. D. . . . . . . . . . . . . . . . . . 207

Baker, J. . . . . . . . . . . . . . . . . . . . . 288

Barry, J. D. . . . . . . . . . . . . . . . . . . 202

Bates, L. . . . . . . . . . . . . . . . . . . . . 570

Bebb, J. E. . . . . . . . . . . . . . . . . . . 143

Becker, S. . . . . . . . . . . . . . . . . . . . 602

Bernues, F. . . . . . . . . . . . . . . . . . . 620

Bolton, G. R. . . . . . . . . . . . . . . . . 350

Bowman, R. M. . . . . . . . . . . . . . . . 145

Cardiasmenos, A. G. . . . . . . . . . . . 633

Carpenter, D. D. . . . . . . . . . . . . . . 384

Chan, R. J. . . . . . . . . . . . . . . . . . . 261

Chernoff, R. C. . . . . . . . . . . . . . . . 230

Chi, A. R. . . . . . . . . . . . . . . . . . . . 568

Coomber, D. . . . . . . . . . . . . . . . . . . 55

Dahl, E. A. . . . . . . . . . . . . . . . . . . 570

Dickinson, R. M. . . . . . . . . . . . . . . 229

Dobrotin, B. M. . . . . . . . . . . . . . . . 112

Downs, H. R. . . . . . . . . . . . . . . . . 252

Dunstan, J. . . . . . . . . . . . . . . . . . . 295

Dutcher, G. L. . . . . . . . . . . . . . . . . 271

Eastwood, Jr., L. F. . . . . . . . . 203, 204

Einhorn, A. J. . . . . . . . . . . . . . . . . 206

Foldes, P. . . . . . . . . . . . . . . . . . . . 168

Gilcrest, A. S. . . . . . . . . . . . . . . . . 288

Gomberg, L. . . . . . . . . . . . . . . . . . 167

Goodwin, F. E. . . . . . . . . . . . . . . . 205

Gran, R. . . . . . . . . . . . . . . . . . . . . 190

Green, S. I. . . . . . . . . . . . . . . . . . . 203

Hartmann, P. R. . . . . . . . . . . . . . . 641

Hedeman, Jr., W. R. . . . . . . . . . . . 566

Hoagland, J. C. . . . . . . . . . . . . . . . 311

Hoedemaker, R. . . . . . . . . . . . . . . . 64

Holmes, J. K. . . . . . . . . . . . . . . . . 387

Holmes, Jr., W. M. . . . . . . . . . . . . 263

Hull, M. L. . . . . . . . . . . . . . . . . . . . 84

Huth, G. K. . . . . . . . . . . . . . . . . . . 517

Kantak, A. V. . . . . . . . . . . . . . . . . 231

Kasser, J. . . . . . . . . . . . . . . . . . . . . 21

Kasulka, L. H. . . . . . . . . . . . . . . . . 541

Kelley, R. L. . . . . . . . . . . . . . . . . . 144

Kellow, R. S. . . . . . . . . . . . . . . . . 640

King, M. A. . . . . . . . . . . . . . . . . . . 448

Kobylecky, G. M. . . . . . . . . . . . . . 134

Lamoreau, L. W. . . . . . . . . . . . . . . . 84

Lankford, J. G. . . . . . . . . . . . . . . . 271

Law, E. L. . . . . . . . . . . . . . . . . . . . 566

Lindsey, W. C. . . . . . . . . . . . . . . . 231

Long, F. M. . . . . . . . . . . . . . . . . . 125

MacPhee, J. . . . . . . . . . . . . . . . . . . 55

Martin, D. H. . . . . . . . . . . . . . . . . . 458

Maynard, J. A. . . . . . . . . . . . . . . . 203

McElroy, S. M. . . . . . . . . . . . . . . . 167

Montgomery, J. H. . . . . . . . . . . . . 142

Montgomery, R. C. . . . . . . . . . . . . 140

Morran, P. C. . . . . . . . . . . . . . . . . 143

Mote, Jr., C. D. . . . . . . . . . . . . . . . . 84

Mukhopadhyay, A. K. . . . . . . . . . . 112

Ng, W-H . . . . . . . . . . . . . . . . . . . . 486

Olshaw, R. Z. . . . . . . . . . . . . . . . . 204

Olson, E. . . . . . . . . . . . . . . . . . . . 260

Omura, J. K. . . . . . . . . . . . . . . . . . 477

Park, Y. H. . . . . . . . . . . . . . . . . . . 371

Patel, D. C. . . . . . . . . . . . . . . . . . . . 56

Pauley, J. D. . . . . . . . . . . . . . . . . . 125

Pautler, J. A. . . . . . . . . . . . . . . . . . 204

Proise, M. . . . . . . . . . . . . . . . . . . . 190

Pusateri, P. . . . . . . . . . . . . . . . . . . . 620

Raghavan, S. . . . . . . . . . . . . . . . . . 639

Raymond, H. A. . . . . . . . . . . . . . . 144

Rhodes, R. . . . . . . . . . . . . . . . . . . . 28

Rodriguez, T. M. . . . . . . . . . . . . . . 288

Russo, G. G. . . . . . . . . . . . . . . . . 641

Schoeck, K. O. . . . . . . . . . . . . . . . 134

Seely, R. . . . . . . . . . . . . . . . . . . . . 125

Semprucci, M. D. . . . . . . . . . . . . . . 28

Sesak, J. R. . . . . . . . . . . . . . . . . . . 145

Sklar, B. . . . . . . . . . . . . . . . . . . . . 402

Smith, B. L. . . . . . . . . . . . . . . . . . 569

Srinivas, D. N. . . . . . . . . . . . . . . . . . 45

Staloff, C. . . . . . . . . . . . . . . . . . . . . 64

Strock, O. J. . . . . . . . . . . . . . . . . . 100

Sullivan, A. . . . . . . . . . . . . . . . . . . 581

Sun, C. . . . . . . . . . . . . . . . . . . . . . 612

Teasdale, W. E. . . . . . . . . . . . . . . . 341

TeBeest, R. . . . . . . . . . . . . . . . . . . 167

Thompson, G. P. . . . . . . . . . . . . . 201

Torres, F. . . . . . . . . . . . . . . . . . . . 497

Torres, J. . . . . . . . . . . . . . . . . . . . . 56

Tu, K. . . . . . . . . . . . . . . . . . . . . . . 341

Udalov, S. . . . . . . . . . . . . . . . 424, 517

Vest, S. . . . . . . . . . . . . . . . . . . . . . 288

Weeks, R. W. . . . . . . . . . . . . . . . . 125

Wilkinson, D. D. . . . . . . . . . . . . . . 541

Wiltse, J. C. . . . . . . . . . . . . . . . . . 589

Wolf, J. D. . . . . . . . . . . . . . . . . . . 204

Woo, K. T. . . . . . . . . . . . . . . . . . . 387

Wood, T. G. . . . . . . . . . . . . . . . . . 141

Zislin, A. . . . . . . . . . . . . . . . . . . . . 190

Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project