AN-183 -- High Frequency Swept Measurements

AN-183 -- High Frequency Swept Measurements

.]:,..

F!.{ur

I

;!.'jfi; r ec rc ea o

L

L ffimwEcffieEsem ffimEe €ffiffi

ffigffiffiFffiEffiffiffiffiY ffiwtrPEffitr&ffiffiffitrffitrffi€ffi

DECEMBER

1978

HEWLETT

-

:v

PACKARD

L

ffiffiffi8trffi8ffi

INTRODUCTION

.

History

Measurement Fundamentals

..

Sweep

Oscillators

Detectors

Displays

Signal

Separation

IMPEDANCE

....

Coax Reflectometer

Measurements

. . .

.

Waveguide ReflectometerMeasurements

...

Waveguide Reflectometer Measurements with

RF Substitution

.

.

Reflectometer

Accuracy

Swept Slotted-Line Measurements

.

...

Swept Slotted-Line

Accuracy

Dilectivity

Measurements

...

.

Source SWR

Measurements

TRANSMISSION

.

Coax

Transmission Measurements with

Directional

Coax

Transmission Measurements

Couplers with

Power

Splitters

Coax

Transmission

Measurements

Summary

Coax Comparison

Measurements

Waveguide TransmissionMeasurements

Waveguide Transmission Measurements

.. with

RF Substitution

.

Transmission Measurement

Accuracy

.

..

.

.

Simultaneous Transmission/Reflection Measurements

.

Power

Measurements

....

Input

Power vs. Output Power or

Gain

PowerMeasurementAccuracy

APPENDIX

A

Source Match of Leveled or Ratio

Systems and Coupler vs. 2- and

3-Resistor

Splitters

B

Errors in

Reflection Measurements

. . .

.

C

Errors in Transmission Measurements

.

Digital

Storage and

Normalization

Other Literature on High Frequency Swept Measurements

....

.....

1

..........1

.....

1

.

..

.

..

..

1

........6

.... ...

... I

.

.. .. ..

.11

.....13

.. ..

.13

......16

..

...

..77

...

.18

.

...

..

.

..

.19

......27

......27

.....

.

.23

..........25

. . . . .25

. . . .

..

.

..28

......29

...

..

.

..

.30

.......31

........92

. .

.. ..

.33

.........34

.......'36

-

.

..

.37

,.....37

.........3e

....

..

'.41

'..

.

...

'43

.........45

.........46

\JI

I

I

I

I

I

I

I

\1

I

I l

L

L

8ffigffiffiffiffiffi88ffiffi

As engineers continue to design and test more complex high-frequency networks, the need for more accurate data over broader bandwidths has become more important. Satisfaction normally achieved quickly and accurately through the swept measurement of these data requirements is of two primary network parameters, impedance and transmission coefficient.

The objective cussion of this application note is the disof specific techniques for the swept measurement of

1.

Impedance fReflectionCoefficient)

2.

Transmission

Coefficient.

In the most general sense, impedance and transmission data possess on both magnitude and phase characteristics.

However, phase information is not required in many general applications, and the techniques discussed here will concentrate on the magnitude characteristics.l

Besides presentation the accuracy and of specific techniques, information particular applicability of the various measurement techniques is provided.

Further development has led to the introduction of multi-octave solid-state sources, allowing continuous swept measurements over several octaves of frequency.

Innovations in other areas of swept-measurement technology have complemented sweep oscillators.

For instance, the development of multi-octave, highdirectivity, coax-directional couplers have been developed at frequencies up to

18 GHz.

New diode detectors with more sensitivity can be combined with the appropriate ratiometer display to obtain greater measurement range and increased measurement confidence over broad frequency spans.

MEASUREMENT FUNDAMENTALS

Four fundamental components are common to all of the swept measurement techniques discussed in this note, These components are:

L.

Swept signal source

2.

Detectors

3.

Displays

4.

Signal separation devices.

Understanding the trol the operation of these components surement systems is important parameters that conimportant in in swept-meaunderstanding techniques, qualifying system accuracies, and configuring new measurement systems.

The major characteristics of swept signal sources, detectors, displays, and signal separation devices are outlined in the remainder of this section.

HISTORY

Early impedance and transmission measurements were made at fixed frequencies in a point-to-point fashion.

These measurements were tedious, time consuming, and often yielded incomplete results. For instance, resonances between points of measurement were often missed or the skirt responses of filters and amplifiers were inadequately defined.

The evolutionary process ficulties was initiated of overcoming these difin

1952 with the introduction of the first broadband, high directivity directional coupler.

The ability reflection to and separate waves on a broadband basis speeded fixed-frequency transmission swept-frequency measurements became the introduction forward and reverse traveling measurements.

In

1954 possible with of a sweeping signal source using a motor-driven, mechanically-tuned klystron. These instruments combined detectors with ratiometers and broadband to introduce swept measurements as a powerful new methodology for network optimization.

By L957 the mechanically-tuned obsoleted klystron had been by the voltage-tuned backward wave oscillator tubes [BWO] as

The voltage-tuned curate sweep times a source

BWO's allowed faster and more acwhile relieving the problems of moding and tracking associated of swept RF signals.

with klystrons. However,

BWO's are vacuum-tube devices which burn out after prolonged use. To overcome the finite lems associated life probwith the BWO, solid-state microwave sweep oscillators were developed in the late

1960's.

SWEEP OSCILLATORS

Sweep source of work impedance and transmission characteristics.

The major technical features portance oscillators (sweepersJ are the most common swept RF signals of in a systems sweeper measuring in qualifying measurements are: and their netim-

1.

Sufficient power output which can be made constant fleveledJ over the frequency range sweep also ance leads important match of the to a wide dynamic range' Leveling is in maintaining good source impedunder a variety of loading conditions.

2.

Accurate linear sweeps and calibrated frequency markers are important for frequency determination and horizontal calibration of

CRTs and

X-Y recorders.

S.

RF signal with low spurs and harmonics is required to minimize measurement inaccuracies when broadband detection is used.

4.

Frequency stability is necessary for making acc.r.ate single

- frequency ICW] and narrowband measurements.

I Phase measurement is discussed in

HP

Application Notes

1U-l and

121.

-

5.

Basic operating functions and capabilities are necessary for efficient calibration and measurements.

Several capabilities important to swept measurements are the various sweep modes, variable sweep speeds, frequency markers, leveling, and modulation.

Sweep oscillators covering the frequency ranges from

400 kHz to

40 GHz are offered by

Hewlett-Packard.

The

8620C solid-state sweeper and some of its major features are shown in

Figure

1.

START MARKER and STOP

MARKER

CONTROLS set independent start and stop frequency limits for broadband continuous sweeps,

'FULL

They also

SWEEP and control tUro markers in

AF.

' rr

PLUG-lN capability allows multiband operation.

AF

CONTROL sets calibrated symmetrical sweep about the

CW

MARKER for narrowband and broadband sweeps.

AUTOMATIC

LEVELING

CONTROL allows output power to be leveled internally (option) or externally with a diode detector or a power meter.

CW

CONTROL sets sweeper for single frequency measurements.

This control also sets one of the marker frequencies in addition to AF center frequency,

MARKER SWITCH actuates amplitude or intensity markers. Marker frequencies are indicated by cw

MARKER, START

MARKER and

STOP MARKER cursors.

FLEXIBLE

SWEEP modes, triggerinq and speeds.

Figure 1.

HP 8620C Solid-State Sweep

0scillator. Various plug-in oscillators for the 8620C mainframe operate over the

3

MHz range. The 86290A RF plug-in (above) covers the 2 to

18

GHz band in a single sweep.

to

18

GHz frequency

Leveling

Sweepers

The concept of leveling is illustrated by the photographs of unleveled and leveled sweeper outputs in

Figures 2

(aJ and 2

(bJ.

Leveling sweeper output accomplishes two major obj ectives:

7,

Output power is held constant as a function of frequency.

2.

Output power is held constant as a function of load impedance because source match is improved.

Flatness and a well-matched source are achieved by an automatic leveling circuit

(ALC) which is standard on HP 8620 and 8690

Series

[BWO] sweepers fsee

2a

2b

2c figure 2. 0scilloscope displays for:

(a) unleveled sweeper output resulting from insufficient

ALC gain, {b) maximum leveled sweeper output, and osiillatlons in the tically identical l'evelin! f6edback to the miiimum loop resulting from unlevelid power;-the excessive remaining

ALC gain.

From variation in leveled the display output is it is evident caused by that the coupling maximum variation with leveled power is frequency. Note:

(c) prac-

Since

\-l negative voltage diode detector was used, more power is down.

Oetector

Directional

Coupler or

Pomr

Spliner

Leveled

RF

Outtrrt

Sweep

Oscillator

_____l

Figure 3.

Basic automatic leveling circuit fiLC). lnternally-leveled sweepers contain the directional coupler and detector (point-contact diode) within the sweeper. Externally-leveled sweepers may use either a directional coupler or a two-resistor power splitter with a point-contact diode or thermistor detector.

Figure

3). tio,!a1

A dc feedback voltage is derived from a detector monitoring the sweeper output through the arm of a directional coupler or similar sampler.

The direccoupler is arranged so only the forward power will be held constant with frequency and load impedance amplifier comparing the dc feedback voltage ence variation. The leveling amplifier voltage; the dc output according to the applied of the voltage. is a amplifier is

Increasing differential to a referconnected to a

PIN attenuator which modulates the output power the leveling amplifier reference voltage calls voltage, hence

Bain more

RF output power. The overall loop is controlled by a variable resistor (ALC

Gain) at

'he other input to for more detector the differential amplifier. Increaiing the gain control raises negative feedback, reducing power peaks and improving leveling; too much gain will make the loop unstable and subject to oscillation,

Figure 2

[c).

duce

Observing Figures 2

[aJ the maximum leveled power unleveled maximum leveled power, the PIN attenuator must absorb any sweeper output power setting.

and

(b), one is very power. When the ALC in

2 is see that near the minimum operating excess of can the to proleveled

It is now easy to see how leveling improves source match: since the directional coupler in the feedback loop only couples power moving tion, the ALC will only in the forward direcrespond to variation in forward power.

Thus no matter what type of load is applied to the leveled sweeper output, the reverse power flow resultant from the load will not affect the

ALC or the sweeper output.

Reflections tors, etc. from cables, connecwithin the loop are also compensated by the

ALC.

Depending on the application, there are a variety of methods for leveling a sweeper.

Essentially, a sweeper may be leveled internally by a point-contact diode detector detector or externally by means of a point-contact diode or a power leveling arrangements are shown

6. The methods meter.

Some of in the most

Figures

4 common through for attaining maximum leveled power and the application of each configuration will be briefly discussed.

DIODE

DETECTOR LEVELING

Interhal Leveling leveling's major advantage is connections

4a and are required. It's

8620

Sweeper

8620

Sreper vs. External Leveling.

Internal a minimum primary its simplicity. No external of sweeper disadvantage

752C Directional Coupler adjustments is that it is usually farther from the point of measurement than external leveling, thus cable losses and connector mismatch reflections that might have been in an external leveling loop will compensated contribute to measurement inaccuracies.

The flatness for most internally leveled microwave sweepers is typically between

11 dB when measured at the output.

i0,5 and

To obtain maximum leveled power, turn the

ALC switch to

INT and the

POWER

LEVEL maximum

[clockwiseJ.

Turn the

UNLEVELED light

POWER goes out for maximum leveled power.

LEVEL down and the until the sweeper is set

11691D

Dir*tional

Coupler

Leveled

RF Output

4b

Figute

4.

Configurations for externally-leveling sweepers with directional couplers and point-contact diode detectors in a) waveguide

(2.6 to

40

GHz) and b) coax (0.1 to

18

GHz).

Note:

Low-pass filters are utilized to eliminate errors caused by harmonics. An oscilloscope and point-contact diode should be used to monitor the leveled output for undesireable loop oscillations.

Externally leveled sweepers using diode detectors are shown in

Figure

4 [a)

(waveguide) and in

Figure

4

(bJ (coaxJ.

The procedure for obtaining maximum leveled power is:

1.

Set up equipment as shown and set the sweeper for a rapid sweep over the desired frequency range.

An oscilloscope and point-contact diode detector should be used put to monitor the leveled source outfor undesirable loop oscillations.

2.

Set

ALC to EXT and turn

POWER LEVEL and

ALC

GAIN to maximum ffully clockwise).

3.

Decrease POWER LEVEL until

UNLEVELED light goes out or oscillations appear on the oscilloscope.

If oscillations appear reduce

ALC

GAIN until oscillations are removed.

4.

Repeat step

3 until

POWER LEVEL can be continuously adjusted before from the minimum to the point just the

UNLEVELED light comes on without oscillations.

5.

Set

POWER

LEVEL to the point just before the

UNLEVELED light comes on and the sweeper will be operating at optimum feedback gain and maximum leveled power.

In

Figure 5 secondary

[aJ a 3-dB directional coupler with the arm terminated has been placed between the 10-dB directional coupler and the ALC detector to obtain a super-leveled waveguide system.

The inverse coupling characteristic compensates coupler's of the 3-dB coupler's main arm for the coupling variation of the

10-dB auxiliary arm. The net result is the cancellation of coupling variation that would otherwise affect the

ALC feedback level. Note: the 3-dB and 10-dB directional couplers must be of similar design and manufacture for cancellation of coupling variation.

POWER METER LEVELING

Because sponse, of the thermistor's good frequency repower meter leveling setups like those in

Figures 6 (a) and 6 [b) provide consistently flat output over the microwave frequency range,

However, thermistors have inherently long thermal time constants, making of the power meter leveling the effective bandwidth loop narrow compared to the diode leveling loop.

Consequently, required long sweep times

(20 to

30 sec,/octave) are for the thermistor to respond to power peaks and the leveling to be efrective.

v

8620 Swper l------1\ Itlo4A ll ll t-e"elins ll

..ll

Amol

ll_+

tL-'--

Ext

AM

I

I

8620

Sreper

Reorder

\ groBTeminatiJ-\

752A

Dirctional

Coupler

Lwded

RF Output

752C

Dircctional

Coupler

752A

Directional

Coupler

Supor

Lfleled

RF Output

8620

Swe€per e't

I

AM

I

8620

Swssper

Figure and

5.

Configurations for a) waveguide super-leveling

?.6 to 40

GHz) b) coax directional detector

(l to

12.4

GHz). ln waveguide superleveling, the two-coupler arrangement results in mutual compensation of coupling variation with frequency, improving leveling performance.

A directional detector is a combination directional coupler and pointcontact diode optimizing coupling variation and detector frequency response for improved leveling flatness.

11691D

DirEtiglal

Couphr

Lileled

RF

Output

The optimum leveling flatness can be achieved in coax by using an HP

780 Series directional detector as shown in

Figure

5

[b).

A directional detector, a combination coupler and point-contact diode detector, has been optimized detector both in terms of coupling variation and frequency response to produce a flat RF output. Both the directional detector and the super-leveled waveguide system are leveled by the same procedure as the systems in

Figures

(aJ and a

[b).

Figure

6.

General systems for a) power meter super-leveling in waveguide (2.6 to

40

GHz) and b) power meter leveling in coax (0.1 to

18

GHz).

Temperature compensated frequency response resulting thermistor detectors have excellent in flat leveled output.

However, the thermistor's long thermal time consta,nt necessitates longer sweep times

(20 to 30 sec/octave) than the point-contact diode.

The basic procedure for power meter leveling is:

L.

Set up the equipment as shown and zero

HP

432

Series power meter. This establishes zero carryover.

L

2.

3.

4.

5.

6.

7.

8.

Set sweeper to

START-STOP and

MANUAL sweep. Set and

ALC to MTR and turn

POWER LEVEL

ALC GAIN to maximum

(clockwiseJ.

An oscilloscope, preferably variable persistence, is useful to monitor for oscillations.

Manually sweep and locate the the frequency range minimum power of point interest with the meter. power

Adjust meter range so that the minimum point is on the upper

% of the meter scale.

Decrease POWER LEVEL until deflection occurs froT the minimum power level. This is effectively leveling to the lowest power point of the unleveled sweeper.

Manually sweep the band of interest allowing time for the thermistor to respond. on the oscilloscope, remove them

ALC GAIN.

This results in maximum leveled power.

If another power meter is

If oscillations by available decreasing the appear output can be monifored directly.

The power meter RANGE may be used brated attenuation;

RANGE switch will i.e. leveled sweeper output.

a

S-dB produce a for calidecrease in the

S-dB decrease in

Continuously variable leveled power below the maximum leveled power is available by moving the meter to a more sensitive range and adjusting the

POWER LEVEL.

An HP

84O4A leveling amplifier may be used to increase the gain of a power meter leveling loop, moving fine-grain variations rein sweeper output power.

If it is desirable to use an amplifier or the sweeper

ALC switch does maximum. Then,

AM; also adto a for basic power-meter leveling, using the 84044 POW-

ER

LEVEL and not have a MTR position, add the

84044 to the feedback loop and change the loop input to the sweeper from EXT INPUT to EXT just the

POWER LEVEL and

GAIN on the

8404A follow the same procedure as used

GAIN. Note, the

UNLEVELED light does not indicate leveled power when leveling with an amplifier. The meter or oscilloscope must be used to determine the leveled condition.

LEVELING AFTER

AN

AMPLIFIER

Most sweepers have adequate leveled power for the majority of swept-measurement applications. However, some networks may require large stimulating signals in order to recreate normal operating conditions.

In this situation, it is desirable to amplify output and then level, compensating the sweeper for any frequency variations in amplifier gain.

A sweeper and

TWT amplifier capable of producing one watt of leveled power are shown in

Figure

7.

Leveling after an amplifier eling any other sweeper with that care must be taken not to

Saturation occurs when there is very similar to leva diode detector saturate is excess the except amplifier.

power at the amplifier input and an increase in no increase in output power.

in input power results

Figure

7.

System for leveling sweeper output after amplification.

The configuration consisting of

Series the

HP 8620 Series sweeper and

HP 489-495

TM amplifier

(shown above) is capable of producing one watt of leveled power in the

1 to

12.4

GHz frequency bands.

Brief instruction for leveling after an amplifier are:

7.

Set the sweeper up for a rapid START-STOP sweep over the frequencies of interest.

Set

ALC to

EXT and mum, turn

POWER LEVEL and

ALC GAIN to maxi-

2.

Set

TWT

GAIN to

RATED POWER using the

CATHODE CURRENT meter, and properly terminate the output.

3.

4.

While observing the oscilloscope, decrease POW-

ER

LEVEL until a reduction in sweeper power results in a reduction a reduction in in amplifier output power. sweeper power results in an

If increase in amplifier output power, the TWT is saturated and the input power must be further reduced.

After removing the amplifier from saturation, the situation is the detector.

essentially leveling configuration;

LEVEL and power. that of any other

Also, care should be taken not diode i.e. use sweeper

POWER

ALC GAIN to obtain maximum leveled to saturate

The system guide in

Figure

7 can be duplicated in waveby substituting the appropriate components.

LEVELING

EFFECTIVENESS in

Figures 4,

5, and 6 is dependent on several factors on

'Leveling effectiveness of which have varying degrees the of sweeper significance illustrated depending the measurement situation.

Where separate coupler and detector are used, sweeper leveling performance is dependent on the coupling variation, detector frequency response,

ALC amplifier gain, and coupler-detector mismatch. formance

If a directional detector is used, is dependent on frequency response of perthe directional detector and ALC amplifier gain.

Sweeper harmonics will also affect leveling effectiveness if the appropriate low-pass omitted filters from the leveling loop.

[or bandpass filters) are

f

dB

At microwave frequencies, the leveling set-ups rn

Figures with coupling variation and detector frequency response

4

(a) and

4 [b] generally have flatness as of

(i1 the major problems. Above 72.4 GHz lhe frequency response of the point-contact diodes becomes the major limitation. As subsequent measurement setups will demonstrate, coupling variation and detector frequency response are between often compensated by tracking the leveling coupler and detector and a complementary coupler and detector at another point in the measurement.

Improved flatness of

(10.6 dB and typical when using the configurations

(!0.4 dB is in

Figures

5

(aJ and

5 (b) respectively. However, these setups are also limited above 72.4

GHz by the frequency response of point-contact diodes.

Further, super leveling and directional detector leveling destroy complementary tracking relationship between the leveling coupler and another coupler in the measurement.

Power meter leveling,

Figures 6 [a] and 6 (b), offers flatness of

(10.5 dB over most of the microwave frequency range. Since thermistor detectors do not have degraded

GHz, frequency response characteristics above 12.4

leveling flatness ble. The major of

(*o.a dB at 40

GHz is possilimitation in power-meter leveling is the long sweep times required for operation of the thermistor detector.

A two-resistor power splitter, such as the

HP

11667A pler may be substituted for a coax directional couin any of the externally leveled configurations below

18 GHz.

The two-resistor power splitter has excellent tracking response

(<0.2 dBl, especially in comparison of

=0.5 and to typical coupling variations

[-{-1 dB). Flatness dB is typical for leveling with a power splitter point-contact diode.

However, power splitters have a

6-dB loss in each arm, resulting in a

6-dB reduction in leveled power output. Comparatively, the main arm losses of 10 dB and 20 dB directional couplers are 0.46

dB and 0.04 dB respectively.

7. frequency response

2. reflection coefficient over full band

3. sensitivity

4. transfer characteristic

5. response time.

Diode Detectors

Because of their nonlinear characteristics, diodes are commonly used in a wide variety of electronic devices diode as rectifiers and demodulators. Specifically, a provides a dc output when an unmodulated

RF signal tens is applied or a low frequency ac output up to the of megahertz when the applied RF is modulated.

They are widely used ments because in swept microwave measureof their simplicity, sensitivity, fast response time, and broad frequency range.

A diode exhibits a square-law transfer characteristic when the output voltage variation portional to diode transfer characteristic is directly proapplied RF power variation. The law region and part of the lins61 lsgiea is illustrated of in a squaretypical

Figure

8.

Square-law response dynamic range is important because it limits in some measurement techniques.

the

E

E o

e o o.s

,/

,

Frequency

Stability

While sweepers generally have adequate frequency stability for most swept measurements, testing of some narrowband networks may require a more stable signal source.

To make these measurements, sweeper output may be phase-locked frequency to a harmonic of a stable lowsignal source.

When phase-locked, the sweeper source, stability approaches that of the low-frequency but operation is limited to

CW mode only.

DETECTORS

Broadband detection swept measurements require accurate of both absolute and relative power levels over wide frequency ranges, ments, detection complished either

In most swept for leveling and display measureof data is acby diode detectors or temperaturecompensated thermal detectors.

There are five fundamental detector characteristics:

-30 -20 lnput Power (dBm)

-10

0

Figure

8.

Typical point-contact diode transfer characteristic

(incident

RF power vs. output voltage).

The diode exhibits a characteristic (incident

RF power directly proportional square-law to output transfer voltage) for input power levels less than

-20 dBm.

POINT-CONTACT VS.

HOT-CARRIER

(SCHOTTKYJ

DIODES

The two most commonly used diodes in swept measurements are the point-contact (crystalJ and the hot-carrier. Point-contact diodes derive their name from the metal whisker that is inserted into a semiconductor chip during manufacture. Hot-carrier diodes have a similar Schottky fmetal-semiconductor] ev.er, the process junction; involves new techniques for howmetallization and is far more controllable than point-contact diode manufacturing processes.

As a result, hot-carrier diodes are more rugged and exhibit excellent uniformity in their transfer characteristics.

This allows the design

I j j i

\-.

of standardized instrumentation non-square-law to compensate for the portion of the hot-carrier diode's transfer characteristic, resulting in an effective 60 dB of dlmnmis range.

Hot-carrier diodes also exhibit a more pronounced non-linearity contact diodes, resulting vs.

-45 fsee dBmJ. the

V-I curve in

Figure

9J than pointin higher sensitivity [-S0 dBm

However, increa$ed sensitivity is normally achieved by dc biasing the hot-carrier diode for operation at the

V-I curve. point of maximum non-linearity on the

If biasing is employed, the applied RF signal must be modulated because signals could low dc levels from small

RF not be detected in the presence of the bias.

Thermocouples

[see

Figure principle that a dc voltage energy is applied to a

10

{b]l operate on the is generated when thermal junction formed by two dissim-

,"*i

il

II

Point Contacl tl

-0.1

0

Voltage

0.1

)

I

o.2

Hot ;arrier

Figure

9.

Voltage-Cunent ff-l)

Characteristics for point-contact and hotcarrier diodes. The hot-carrier diode exhibits hence greater sensitivity.

a greater nonlinearity,

0.4

0.3

E

0.2

(J

:

0.1

10a

10b

Figure

10. Simplified diagrams for a) bolometer

(thermistor or barretter) and b) thermocouple.

A bolometer is a temperature-sensitive resistor facilitating power measurement through bridge balancing techniques.

A thermocouple is a metallic junction of two dissimilar metals, generating a dc voltage when heat (incident

RF signal) is applied.

ate

Point-contact diodes like the Hp

4ZO

Series oper_ at frequencies between

100

HP

11664,4, hot-carrier diodes

)*ZO kHz and 40 GHz, operate dBm.

over the wiile

1S-MHz to

1B-GHz frequency range.

Both devices are subject to burnout power levels ilar metals. The magnitude of the dc voltage developed across the junction ergy generated is proportional by the applied

RF.

to the thermal en-

Thermistors are characterized quency ment range

[1 pW sponse.

100-kHz range ftypically to

1

MHz to to

1B-GHz frequency range, by their broad fre-

40

GHzJ, measure-

10 mW), and low-frequency

Similarly, thermocouples are noted for retheir low drift, measurement range (0.3 plW to

100 mWJ, and exceptionally low reflection coefficient (1.28 SWR at L8 GHzJ.

Both types of detectors have long response times in comparison to diode detectors. Also, specially matched pads are incorporated into thermocouple detector mounts, allowing power measurement up to

3 W.

Thermal

Detectors

Where accurate knowledge power is required, of average absolute temperature compensated thermal detectors such as bolometers and tliermocouples are utilized. As their name implies, thermal detectors sense

RF power levels energy

_or by determining the amount heat generated of thermal in a particular load impedance.

Bolometer detection thermal element's resistance changes er dissipated element absolute is by the inserted is based element. into power can be on the fact that with the

When such a bridge as measured in through a

RF powthermal

Figure

10

[a], baiancing techniques using a eters are differential amplifier.

Bolomdivided into two categories, barretters

(positive-temperature coefficient) and thermistors fnegalivetemperature coefficientJ; a negative-temperature coefficient implies that increasing the RF power dissipated in an element results in a decrease in the elementls resistance, Thermistors are lometer because by far the most common bothey are more rugged and less subject to burnout than barretters.

a

Measurement Techniques

Detector characteristics significantly choice affect the of a measurement technique as well as subsequent display instrumentation. Measuring any unknown parameter is always executed by comparison of the unknown quantity to a known quantity; which is taken as the standard. ment

There are

In other words, is performed, some kind of whenever a dio substitution, RF substitution, and measuresubstitution parison between an unknown and a known three common measurement techniques, au-

IF or is commade.

substitution, used in swept measurements. The choice of a technique for a particular application is based on three criteria: dynamic range; accuracy; and cost.

AUDIO SUBSTITUTION/SQUARE.LAW

DETECTION

As

Figure

11 indicates, audio substitution involves modulation voltage of as the detector output the RF source

The diode detector detects is in is directly proportional to the incident power. The change in at or square-law, inserting the

Device

Under Test an audio frequency.

demodulates providing an audio-frequency voltage output. the the amplitude of output amplitude produced

RF

As long its by is compared to a reference amplitude by substituting known calibrated audio gain or attenuation. The major advantages of audio substitution techniques

However, limited are economy and simplicity.

the dynamic range of the measurement is to the square-law region of the detector unless linearity compensation is employed.

The primary advantage of

RF substitution is that it does not rely on square-law performance of the detector, since the detector level both is held at a constant sigaal in calibration and in measurement.

The accuracy of the measurement is heavily dependent on the accuracy of the standard over the frequency range of interest. While much of the simplicity of audio substitution is maintained, precision RF attenuators usual-

Iy cover narrow-bandwidths, necessitating several expensive attenuators to make multi-octave measurements.

IF

SUBSTITUTION/FREQUENCY

CONVERSION

In

Figure

13, the local oscillator (LO) is tuned to mix with the

RF to produce a constant IF with the same amplitude information as the incoming

RF signal. The measurement is made by comparing the measured amplitude to the calibration amplitude using an

IF standard.

-DC

Figure

11.

Simplified Audio-Substitution measurement.

The sweeper is amplitude modulated at an audio frequency.

The detector detects or demodulates the

RF signal, allowing precise substitution of audio frequency gain or attenuation in the display unit.

Figure 13. Simplified lF-Substitution measurement. The RF is mixed with a signal from a local oscillator to produce a constant lF

(intermediate frequency) selectivity signal. Convenient lF filtering provides best sensitivity and ol all techniques.

RF

SUBSTITUTION

In an

RF-substitution measurement, a precision

RF attenuator between is utilized in the DUT and the measurement and placed the detector. Otherwise, the measurement mechanics are essentially the same as the audio-substitution measurement; a measured amplitude is compared to a calibration amplitude by varying the

RF attenuator.

RF l-lJ-l-[l-fl

Audio

Figure

12.

Simplified

RF-Substitution measurement. Measurement is accomplished and by comparison of the

DUT the detector to a precision

RF attenuator, is not required to exhibit a square-law transfer characteristic.

The narrow-bandwidth, single-frequency detection of the

IF allows high sensitivity and dynamic ranges greater than

60 dB. Frequency conversion is expensive relative to audio and RF substitution and is usually employed in precision calibration instruments, spectrum analyzers, or multi-channel, phase-measuring network analyzers.

More detailed discussions of IF substitution/frequency conversion are available in

HP Application

Notes 117-1 and72'L-7.

BROADBAND

AND NARROWBAND

DETECTION

The full-frequency spectrum simultaneously accepted of input signals is by broadband detection systems. Diodes, temperature-compensated thermistors and thermocouples are typical examples of broadband detectors.

While generally economical, broadband systems usually sacrifice noise and harmonic rejection.

However, harmonics can be eliminated through the careful use of filters.

tates

Narrowband detection systems employ frequency conversion to transform the swept RF to a constant

This eliminates problems due to harmonics and facilihigh sensitivity through the low-noise detection

IF' of the

IF.

However, narrowband systems are generally

Iess economical than broadband systems.

\-/

\-

DISPTAYS

Much of the convenience of swept measurements is dppendent on continuous, real-time displays for readout of data.

The basic technical features of a swept-measurement display system are:

1.

Broad dynamic range for the measurement of large variations calibrated in input signal level is facilitated by a log display and detector linearity compensation.

2.

Several detector inputs and processing channels are necessary for executing simultaneous measurements of the user's choice.

3.

Ratio measurement capability, or the capability of comparing two detected signals, is required to minimize source mismatch errors without a leveled source.

4.

Calibrated absolute-power measurement

5.

Straightforward operating functions and controls provide efficient operation in a variety of measurement modes.

is necessary to assure optimum power for both the device under test and the measurement system.

HP

8755

The HP

8755 Frequency Response

Test Set and some

14. of its basic operating features are shown in

Figure

It employs hot-carrier diode detectors and provides the appropriate linearity compensation dynamic range over for

60 dB of its

15-MHz to

1B-GHz operating band.

CALIBRATED LOG DISPLAY for accurate measuremen8 over

60 dB of dynamic range.

ABSOLUTE

POWER MEASURE.

MENT can be achieved at any the three detector inputs by of setting

OFFSET

CAL switch to

OFF.

RATIO MEASUREM ENT capability allows accurate measuremsnts without a leveted sourcs.

TWO PROCESSING

CHANNELS with independent controls for simultaneous measursments.

DIGITAL

OFFSET for accurate substitution measuremenG and direct readout of data.

HOT

CARRIER DIODE DETEC.

TORS with -50 dBm sensitivity for increased dynamic range.

!-

Figure dBm)

14.

HP 8755

Frequency Response Test Set. 0peration with hot-carrier diode detectors provides 60 dB of measurement range from 15

MHz to

18

GHz. The system also comes complete with an external modulator.

(a10 to

-50

The differences between a linear oscilloscope display fed by a point-contact diode and a calibrated log display may be noted loss through a high display) and 15

[b) by pass filter

(linear observing in the transmission

Figures oscilloscope

L5

[a] display).

(log

The uncalibrated linear display completely misses the

25-dB dip occurring at

2.4

GHz.

While adequately defined, the significance to the of the

11dB dip near

3.3 GHz with respect passband ripple assess.

This occurs

[3.8 to

4.0 GHz) is difficult to primarily because the linear display over-responds display to tle passband ripple; an equivalent of passband ripple without the confusion of rejection band characteristics may be obtained by increasing the log display's resolution.

v

0

6

!

.9

.9

-10

E o

'?

.9

-zo

E

F

-30 o!

EO

IE o:

'-r

CI

FI

0

Frequency (GHz)

15a

FEquency (GHzl

15b

Figure 15. Transmission loss through a high-pass filter as displayed on a) calibrated log display (8755 system) and b) uncalibrated linear display

(oscilloscope dB fed by a point-contact diode). Note that the linear display misses certain variations in the transmission characteristic, such as the

20dip at 2.3

GHz, while over-accentuating the passband ripple €.6 to 4

GHz).

HP

415E

The 415E calibrated detectors in is a low-noise tuned ampliffsl-veltmeter dB and SWR for use with square-Iaw in audio-substitution measurements.

It may be used for measuring SWR, attenuatiou, gain or other parameters determined by the ratio of two signal levels.

The standard tuned frequency is

1000

Hz. The

415E also contains a precision

60-dB attenuator gain measurements for attenuation or with audio substitution. measurements, the 415E's dc output is used

In swept to drive the

X-Y recorder.

\JI

16a

16b

Figure 16. a) HP

4l5E

StrYR

Meter, b)

HP

432A

Power

Meter, and c) HP 435A Power Meter.

16c

HP

432

Series and HP

435A ate ates

The 432 Series with

84788, and HP 486 with the HP

Power Meters are designed to operthermistor detector mounts (HP 47BA' HP

Seriesl; the 435,{ Power Meter oper-

8480 Series thermocouple

The

432 Series indicates absolute power on detectors' full scale

10 ranges between 10 pW and 10 mW while the

435A has full scale ranges between 3 pW and 3

W. Both meters are fully calibrated to the appropriate detectors and offer readout accuracies of. tlo/o of full scale' DC recorder outputs are available for making swept measurements with X-Y recorders.

SIGNAL SEPARATION

The three separation most commonly used devices for signal in high-frequency swept measurements are:

7. directional coupler

2. directional bridge

3. power splitter.

Directional couplers are devices used or sample the traveling wave moving in to separate one direction on a transmission line while remaining virtually unafiected by the traveling wave moving in the opposite direction. Directional bridges have similar characteristics to directional couplers with a different internal implementation. Power splitters separate a traveling wave on a transmission line into two equal components.

Coupling factor, directivity, mainline loss, and mainlins

SWR are the four principal parameters defining directional coupler performance. Coupling factor is the fraction (in dBJ of the power moving in the forward direction through the coupler main arm that appears at the secondary arm.

Mathematically

[see

Figure

77 (a)l:

Coupling factor

= -ro loc

f .

(11

Since directional couplers are not perfectly directional devices, directivity is used as a figure of merit for a coupler's ability to separate the forward and reverse traveling waves. the power

Directivity is the ratio (in dBl of in the secondary arm when all the power is flowing in the forward direction of the main arm to the power in the secondary arm when an equal power is flowing in the reverse direction of the main arm. Mathematically

[see

Figure rZ ft]l:

Directivity =

-10 l"t ;T.

The mainline loss the main arm is the by a traveling wave moving fraction of power lost in either direction through of a directional coupler. Mathematically

[see

Figure 17

[a) and 1z

[b]l:

Mainline loss = -ro los

ff.

Mainline SWR is a measure coupler output port and of is important mismatch at the in qualifying effective source match.

An

HP 11692D

(2 to

18

GHzl dual coax directional coupler for independently sampling both the forward and reverse traveling waves is shown in

Figure

18 [a).

Pzr

,}

17a

\t8b

Pza

+

17b

Figure

17.

HP 752 Series precision multihole waveguide directional couplers with a) all power moving in the forward direction through the coupler and b) an equal power moving in the reverse direction.

Coupling factor

-

-10 log

$, directivity =

-tO toc}, f2t and mainline loss

-

-10 rog

*.

77

Figure

18. a)

HP

11692D, directional coupler and

2 to

18

GHz high directivity, coax dualb) HP

11666A

40

MHz to

18

GHz coax reflectometer bridge.

Directional bridges utilize impedance bridge techniques to accomplish the same objectives as directional couplers. Main arm to auxiliary arm coupling, directivity, and mainline SWR are the important parameters in qualifying directional bridges.

Main arm to auxiliary arm coupling

The is equivalent to the sum of the mainline

Ioss and coupling factor

HP

116664 Reflectometer Bridge shown

1S

4o

(b) may be used

MHz and 18 GHz in to swept in sample a directional measurements coupler.

in

Figure between both forward and reverse verse traveling waves.

The forward wave and the rewave are sampled by a directional bridge, and two hot-carrier diode detectors are built in for use with the

8755

Frequency

Response

Test

Set.

Power splitters are usually defined arm_insertion loss, equivalent tracking. output

SWR, and output

Main arm insertion loss is the loss that occurs between the input and either of the by their main two output arms

(see Figure

191:

Main arm loss =

-10 log

#

= -ro los

fr|

Output tracking splitter maintains equality between

Pea11 and

Poo12

Figure

19J:

(in dB) indicates how well the fsee

Output tracking

= ro los

f,ffi

Equivalent a source output

SWR is a measure of how good match may be obtained when a power splitter is used in leveling a sweeper or in a ratio measurement. Power three-resistor splitters may be of either two-resistor or construction; three-resistor power splitters are greatly inferior to two-resistor power splitters in terms of equivalent output

SWR when used in leveling or ratio applications

(see

Appendix

AJ.

Pout

I

Pout

2

+

P.

tn

Figute 19.

HP 11667A dc to

18

GHz two-resistor power splitter.

Main arm loss

-

-10 bC

+ -

6

D dB and main arm tracking = 10 log

;:

(typically

<0.2 dB).

Two+esistor power splitters exhibit excellent roul2 equivalent sweepers output

SIIJR

(typically 1.2

SWR) when used in leveling or ratio measurements.

\,/

72

\-

8ffiffitrffiffiffiffitr

Characterization frequency is fundamental to both the design and test of broadband high frequency networks. However, the wavelengths of of impedance as a high-frequency signals function of are usually small compared sion-line theory to the physical size of the networks under test, necessitating the application of in obtaining impedance data.

transmis-

According applied to a wave.

These to transmission-line uniform, lossless two traveling waves add tionary standing wave along

The terminating impedance line impedance different from its characteristic impedance will produce both an incident and a reflected traveling the is theory, terminated to form a statransmission proportional a signal in to an line.

the magnitude variation of the reflected wave and the amplitude in the standing wave, resulting in two basic, but related, high-frequency impedance expressions.

Quantitatively, these relationships terms of are explained in reflection coefficient and standing-wave ratio.

Equations 1 through

4 present the common expressions for impedance both in linear and log form.

Reflection Coefficient

Magnitude:n

=

lffffil

frt

Return Loss

[dB]r RL (dBl

-

-20 log p

Standing Wave

Ratio:

SWR =

tfffl=

1*p

7-p

(21 t3l

Standing Wave Ratio [dB): SWR

[dBJ

:201og

SWR t4l

Swept standing-wave using a swept slotted line. of uniform transmission line with a longitudinal slot that allows insertlon device measures transmission the electrical field strength along the path of the slotted line, allowing determination of lE-",

I of a probing device. The probing and taneously sweeping

I ratios are measured directly

E-1, the

A l, slotted line is hence SWR. frequency and a section

By simulmoving the probe along termined on

Oscilloscope result the slotted-line path, SWR may be dea swept basis.

The 8755A/7BIT

Storage display can then be used to display the directly in

SWR

[dB).

Applications ience and ment. Return loss and SWR

[dBJ are directly related by equations 1 through 4, allowing the final measurement data matter of the reflectometer or slotted line are generally determined on the basis the accuracy desired to be converted to of the convenin a particular measurethe desired format no which technique was used in the original measutement.

The objective basic reflectometer surements. models of this section is to present the and slotted-line impedance mea-

To this end, accuracy considerations and will be discussed, as well as the relative applicability of both techniques.

COAX

REFLECTOMETER MEASUREMENTS

Figure 20 exhibits the results turn loss measurements tometer systems cause of of some typical rein coax using the

8755 refleceither Figure 21 (a) or

21 [b).

Be of its speed and versatility, the reflectometer is

Swept-reflection coefficient sured directly using a reflectometer measurement configuration.

A reflectometer tional couplers or return loss is meautilizes the ability of direcor directional bridges to individually sample the incident and reflected traveling waves.

Reflection coefficient or return loss is determined by performing the may be ratio of the detected signal levels. The ratio performed in either of two ways:

'1..

DIRECT RATIOING:

In direct ratioing both incident and reflected signal samples are fed directly into an instrument like the

8755 system which performs the ratio.

2.

LEVELING: both

1E1""

I is maintained at a constant level in calibration and test by externally leveling the sweeper output.

In both cases the net effect is the same; however, direct ratio measurements are usually more convenient because the sweeper does not have to be leveled prior to making the measurement.

The equivalence between ratioing and leveling is explained mathematically in

Appendix A.

6

20a :

;0

!zo

G

40

Frcquenry

(cHz)

13

;

20b

-i ro n20

30

Frequency

(cHz)

Figute

20.

Return-loss measurements pass of a) bandpass filter and b) lowfilter as seen on the display of the

8755 Frequency Response Test

Set.

one of the most widely used systems for obtaining broadband swept-impedance the reflectometers ble of return loss dataln coax. For instancel in

Figures 21

[aJ measurements and 21

[bJ are capafrom 1b MHz to-18

GHz, assuming the sweeper and directional couplers for directional bridgesJ operate over the frequency range of interest.

Det

A

DLJJ.

Since leveled as

Appendix convenient power to the

8755 for system bilities, the sweeper does wave are it was in many

A for a discussion of leveling vs. ratioing.) In compensated controlling has not older a leveled system the variations wide reflectometers. variations the DUT when making ratio

{See in the incident traveling for by ratio the measuring necessarily have sweeper leveling loop.

Ratio-measuring instruments sate and for any variation leveling in by instantaneously ratioing similarly compenthe incident traveling it with the reflected is not required unless the DUT is wave wave, inputpower sensitive.

Because setup and calibration are both quicker and easier, ratio measurements are generally preferred. However, internally leveled sweepers are in capato input measurements.

At this point,

8755 care system must be taken signals is a broadband detection system; therefore to eliminate spurious and harmonic that might otherwise introduce errors into the measurement. it should be re-emphasized that the

Incorporation of a low-pass filter, like the one shown in Figure

21, is usually sufficient to eliminate errors caused by source harmonics. However, the source produces sub-harmonics or other spurious signals, careful use may be required, of high-pass or if bandpass filters

A general outline of the reflectometer measurement procedure is presented below.

\./

116664 |

Refl

Bridge

\

-.-C!

Short

Figure

21.8755 coax reflectometer systems utilizing a) ll592D

18

GHz dual directional coupler and b)

11666A

40

MHz

2 to to

18

GHz reflectometer bridge.

In the reflectometer

[see

Figure 27

(a)1, the R detector of the

8755 system is placed on the forward arm of the directional coupler to. sample the incident traveling wave while the

A detector flected traveling wave through samples the rethe coupl-er's reverse arm, The

8755 system automatically performs the ratio of the reflected and the incident signals, displaying the results in bration is achieved by placing a short circuit dB or p

= terms of swept return loss

1) at the reflectometer

[dBJ.

Initial

[R.L. cali-

=

O output port. Actual measurement is accomplished by replacing the short with a DUT (device under testl and displaying the results on a CRT or X-Y recorder.

Sweep speeds may be adjusted either for continuous CRT displays or slower

X-Y recordings.

74

Set up and Galibration

7,

Set up the equipment as shown in

Figure 27 and adjust the sweeper to sweep the frequency range of interest.

2.

Use the

8755A

POSITION controls to adjust the

0 dB/O dBm position line to a convenient graticule, hence referred to as the position graticule.

3.

Remove the R detector and connect it to the measurement mately dBm, port, adjust

With position adjust the the

OFFSET sweeper

CAL output in for the

*10 dBm at the measurement port (if

OFF approxi-

(*10 for maximum). By using the

8755 system's capability to measure absolute power

[OFF-

SET CAL to

OFFJ, maximum dynamic range assured. Care should also be taken not input power specifications of the

DUT.

is to exceed

4.

Return

R detector to its normal position and set

8755A for a ratio measurement

(A/Rl.

5.

Connect a short to the test port. With

OFFSET dB at t00 and

OFFSET

CAL to

ON, use the

OFFSET

CAL vernier to average any variation in bration trace about the position graticule.

the cali-

Note: The

8755 reflectometer system brated is now califor swepit return loss measurements.

If the

OFFSET

CAL vernier is moved during the measurement, calibration will be destroyed.

Measurement

6.

Remove the short and connect the DUT to the measurement port.

Be careful to terminate the

DUT in its characteristic impedance ports.

if it has two or more v

7.

Use the

OFFSET dB thumbwheels and

RESOLU-

TION push-buttons to obtain a suitable display.

8.

OFFSET dB thumbwheels can be used for accurate substitution measurements and digital readout of return loss (to the nearest dBJ at any point on the trace. Once the thumbwheels have been used to bring the point of interest to the position graticule, resolution may be increased without the trace leaving the screen.

The advantages of increased resolution are demonstrated by the series of

CRT photos in

Figure 22. lJsing the appropriate offset, resolution may be increased for a better view area of interest, of the trace as a whole or a particular

During calibration the absolute power at the measurement port was adjusted for approximately

Assuming a nominal coupling factor rvould allow

40 dB measurement. of of dynamic range

20 dB, in a

*10 dBm.

+10 return dB loss

However, this is not a restriction since high return loss measurements are normally limited by the directivity of the reverse directional coupler which is usually

<40 dB.

If

X-Y recordings of data are desired, they are easily secured by adding an

X.Y recorder to the measurement setups volts,/scale in

Figure auxiliary outputs

IAUX division, a

21.

A

Since and recording of the

8755

AUX B) any trace system provide at

0.5

any resolution may be obtained initially calibrated. after the recorder has been

A typical X-Y recording is shown in

Figure

23.

22a

Frequency (GHz)

22b

-3 to

Ezo

In rv

\r

\A

\/ll lJ f'

Frequency (GHz)

3

Frequency (GHz)

Figure

23.

X-Y recording of a bandpass filter return loss. The recording was made with the

8755 reflectometer system in

Figure 21

(a).

o

E1

E

2

3

4

Frequencv (GHz)

22c

Figure

22.

Return loss tion, a)

10 dB/div, b) of a low-pass filter at various levels

5 dB/div, and c) of resolu-

I dB/div.

Note that increased resolution can be uSed for a better display of the trace as a whole,

(b), or close scrutiny of the rejection band return loss,

(c).

15

Because stored calibrated grid lines can be plotted and on an X-Y recorder, the recorder is particularly useful ments. in making high-resolution return loss measure-

Variations in the calibrated grid lines are caused by tracking errors ffrequency responseJ and calibration errors plotted in tribution the of measurement reflectometer system; source tion errors varies match to however, effective with the phase of the consystem calibrathe load at the port. Ideally, a calibration grid should be with a short but 180" different in

[p

=

1J phase and with from the an open [p short)

; the

=

1, mean of these two grids would be taken as the actual calibration grid. This procedure cause open circuits have the phase angle to is difficult finite differ from that to follow capacitance, becausing of a perfect open

[0o] at frequencies above 8 GHz.

In

Figure 24 the progressive steps are demonstrated for making a high-resolution X-Y recording of return loss data. The first grid is drawn with the measurement port shorted.

Because the impedance resulting from the connector capacitances is small between

2

GHz and

4

GHz, the measurement port may be opened and a second grids calibration grid plotted.

The mean of these two is assumed to be the true calibration grid. The

DUT is then connected at the test port and the measurement consummated, resulting in the plot in

Figure

24.

An alternative means of storing the reflectometer tracking errors ffrequency response) grease-pencil recording trace.

As in the case of

X-Y recordings, is it to is make on the CRT of the calibration normally a sufficient to record only the calibration trace resulting \/ from a short at the reflectometer test port. When the actual measurement is executed the OFFSET dB thumbwheels can be used to compare the measurement trace to the grease-pencil recording. Of course, all measurements involving made the grease-pencil recording must be at the same resolution as the original recording.

t1

{

OP

EN

)RT

X,

(-

><

WAVEGUIDE REFLECTOMETER MEASUREMENTS

Figure system

25 shows an

8755 waveguide reflectometer for making swept return loss measurements in waveguide bands between

2.6 GHz and 18 GHz,

The theory of operation is virtually identical to that of the coax reflectometer system.

However, there are some differences peculiar to the waveguide components themselves that are discussed below.

24a

3

Frequency (GHz)

\./

; a.10

-3 rr

&.e

\,

\

."-l

ME

,/l

\SUREI

-a

><

(v

Det

R 28-:A

DerA

L.P.

752C Directional Couplors

Filter g20A

D.U.T.

Figure

25. 8755 waveguide reflectometer system

(2.6 to

18

GHz in waveguide bands). required

Note HP

281 Series waveguide-to-coax adapters are to connect the

11664A detectors to the waveguide directional couplers.

234

24b

Frequency (GHz)

Figure

24.

High resolution

X-Y recording grids, one of return loss. Two calibration with the test port shorted and the other with it opened, are plotted in a), and the mean taken as the true grid.

The DUT is inserted; the

8755 system adjusted to the correct

0llset dB setting, and the high resolution measurement accomplished in b).

The open/short calibration improves measurement accuracies because it eliminates calibration and tracking

(frequency response) errors from system uncertainty,

In general, most coaxial connectors act as reasonable open circuits to

8

GHz. Complete elimination of source match and directivity errors is not possible without both phase and magnitude information'

16

The precision multi-hole waveguide directional couplers utilized in

Figure

25 have a nominal coupling coefficient of the high directivity, return loss measurements up to

40 of

10 dB and

)40 dB of directivity.

Because dB are feasible.

The L0-dB mean coupling, compared to

20-dB mean coupling allows

40 dB of dynamic range urement with 0 dBm of in coax directional in a return incident power at couplers, loss the measmeasurement port. Note that waveguide-coax adapters are employed to attach the

116644 detectors to the waveguide couplers.

Another particularly distinctive feature guide systems of waveis the availability of high quality sliding

short,

A waveguide sliding short has a constant reflection coefficient can be phased of unity

[p

:

1 or R.L. be simulated at the measurement port

- 0 dB] which through >180o. Since an open circuit can by a short onequarter wavelength removed, the total effect it is now possible to see of source match on effective system tracking errors.

This is easily observed on the CRT by sliding the short at least one-quarter wavelength of the lowest frequency in band being swept.

The measurement procedure flectometer for the waveguide reis similar to that of the coax reflectometer.

The important points and major differences are outlined below.

Set

Up and Calibration

7.

Set up the equipment as shown in

Figure

25 and set the sweeper for the frequency range of interest.

Adjust

POSITION controls on 8755A if necessary.

2.

Use the R detector and adapter to measure the absolute power sweeper at the measuremenf the output power measurement for

0 dBm to port,

*10

Adjust dBm at port

(assuming 10-dB directional couplers are usedJ.

3.

Reconnect the

R detector to the forward coupler, connect the sliding short to test port, and set 8755A for ratio measurement

[A/R].

4.

Phase sliding short and note the variation in effective system tracking errors. Establish

0 dB returnloss calibration by averaging the variation in the tracking errors about the position graticule.

Measurement

5.

Same procedure as the coax reflectometer. Figure

26 shows the return loss of the X486A thermistor mount measured on the system in

Figure

25.

X-Y recordings may be obtained by adding an

X-Y recorder to the measurement setup proceeding in

Figure 25 and in a fashion similar to that used for the coax reflectometer.

The only major difference between the coax and waveguide systems is in the plotting of calibrated grid lines. When plotting the grids, the sliding short should be rapidly phased to simulate the various phases of variation load in impedance.

The result the grid lines which is defines a the fine-grain limits of source match

The mean contribution to effective tracking errors.

of these variations is taken as the true calibrated grid.

As in the case of coax reflectometers, plotting of calibrated grids can be ignored where high resolution is not required, and grease pencils can be used to record calibration traces directly on the

CRT.

WAVEGUIDE REFLECTOMETER

MEASUREMENTS

WITH RF

SUBSTITUTION

An

RF-substitution technique for making swept reflectometer measurements from 2'6

GHz to

40 GHz is in

Figure

27.2

System operation is based on a shown leveled sweeper and a precision RF rotary-vane attenuator. Before the actual measurement cific values of return loss are pre-inserted via the attenuator, and the results stored grids on the X-Y recorder. in is the form executed, of specalibration

While

RF substitution is not as convenient as the 8755 reflectometer, above 18 GHz and it does operate is composed of economical instruments.

424A

Xtal

Det

/[

N

t*

D.U.T.

Frequency (GHz)

Figure 27. System techniques. With for measuring swept return loss using

RF substitution the test port shorted, values of return loss are preinserted using

The short the precision attenuator and calibration is replaced by the

DUT; the attenuator set grids plotted.

to a reference value; and the measured trace compared to the calibration grids.

27.

Obtaining maximum leveled power from the sweeper is the initial step

The sweeper in calibrating the system in

Figure is essentially externally leveled in

Figure

26.

Return loss of an

HP X486A thermistor mount. The maximum

SWR specification for the

X486A is

1.5

SWR

(R.1. -

14 dB). Note that the thermistor mount tested above is generally much better than the specification over the full operating band.

77

2

An 8690 Series sweeper must be used at frequencies above 18

GHz

I waveguide with a point-contact diode.

Since the sweeper output is leveled, forward {or incidentJ power must remain constant both as a function load impedance. Thus, it can be of frequency assumed th-at and iniident power at the measurement port calibration and measurement.

As noted sions of

8755 will be the same in both in the discusreflectometer system, leveling accomplishes the same effect as ratioing

{see

Appendix

AJ.

Operation of the

415E SWR

Meter requires the sweeper output to be amplitude modulated at a

1-kHz rate.

1-kHz amplitude modulation is available internally on all

8620 and 8690 sweepers.

With the reflectometer test port shorted, the

382 precision attenuator is used to pre-insert specific values of return loss lines in the reverse measurement arm.

By manually triggering the sweeper for single sweeps, grid at each attenuator setting are sequentially plotted on the X-Y recorder. During each calibration sweep, the sliding short is rapidly phased so that all phases of the source verse match error signal will be encountered at arm detector. The result is a fine-grain the revariation in the grid lines which defines the limits of source match flectometer, the mean as the true value source error contribution. As in the case of the

BZbb rematch of of these variations return loss for will each be taken grid.

Thus error is minimized, and better system accuracy is attained.

After the grid lines have been drawn, the short is removed and replaced with the DUT. A final sweep is triggered loss with the attenuator set to zero, and the return plotted on the X-Y recorder. The results return loss measurement are shown in

Figure of

28.

such a

^

17.0

a0

7 rg.o

! rs.o

€ zo.o

21.5

25.O

40.0

a'2

Frequency (GHz) figure 28.

X-Y recording of the return loss of an HP X486A therm.istor

mount bimilar t0 the one tested in

Figure 26) using the

RF-substitution system in

Figure

27.

be driven at a higher signal level and errors from recorder noise avoided. figurations,

Series function. a

432

RF-substitution measurement either as an ampliffer or a ratio measuring device. leveling loop would be eliminated. the basic measurement procedure remains unchanged, except the

OFFSET dB thlmlqrfueels purposes.

the

415E

In alternative could

In the latter is measurement conbe supplanted case by or a

435.4.

Power Meter, performing the sa-e

It is also possible to use an 8755 system in a

If the

8755 is used, that the

382A precision attenuator instead of used for the sweeper substitution ates

A full

40 dB of dlmamic range in a return loss measurement

As is provided by the system in most reflectometers, the directivity tional couplers limiting factor contact diodes in of

Figure the

27.

direc-

[)40 dB for

HP

752C couplersJ is the of dlmamic range. Further, the pointare not constrained square-law response since to operate with the detector always operat the same level both in calibration and measurement conditions.

r

5.0 nff*|lphr^\fl!yh*-.

Once they can be used as an underlay ments the calibrated grid lines have been drawn, with the actual return loss for many measureof the

DUT recorded on translucent paper. However, grids should be redrawn after long hours of operation or after the equipment has been turned off.

In the setup in

Figure 27, the

415E acts solely as an amplifier, amplifying the signal from the reverse arm detector. This allows the Y-axis of the

X-Y recorder to

18

REFLECTOMETER ACCURACY

Sources of uncertainty ment systems in reflectometer are generally classified measureas directivity, tracking, calibration, and effective source match errors.

These general four sources of error equation: error may be collected into a

Ap=A*Bp*Cp2.

(51

The value of the

A and C coefficients is determined by the directivity and source match errors respectively, while

B is composed of frequency response, instrument, and calibration errors.

A complete derivation of equation

(5) and numerical examples are provided in Appendix B.

It is important to remember that terms are functions of frequency and all will three usually significantly over the frequency range of interest.

error vary

An

HP

Reflectometer

Calculatof like the one in

Figure

29 pedance is useful both data and in changing the form of imin determining the measurement uncertainties caused the calculator by coupler directivity. For example, in

Figure

29 is set for a reflection coefficient of

0.20, using the blue arrow above the top scale.

The corresponding SWR equations value

(2) and

(3J. of

1.5 and of the directional couwith

40 dB of directivity is used to measure a return loss return loss of

14 dB are indicated by the blue arrows above and below the lower scales, avoiding the calculations required in

The

Coupler graduations, sigaify the worst-case window tainty caused by the directivity pler. For instance, a coupler of

Directivity

1.4 dB, of the

[dB) unceractual of return loss could be anywhere between

13.6

dB and

74.4 dB depending on the between vectorial relationship the directivity and reflected signal.

For directivity of

26 dB, the actual return loss could be as low as

12 dB or as high as 16.6 dB.

Similarly, the

Coupler

Directivity graduations values of

SWR and p.

also denote the worst-case

3A complimentary HP Reflectometer/Mismatch Error Limits

Calculator is avail.

able trom any Hewlett-Packard sales office.

*v

Figure tween

29.

HP Retlectometer Calculators are used for conversions bep

SWR, and

Return Loss. Worst-case uncertainties caused by

Coupler

Directivity and Mismatch Loss for absolute power measurements may also be calculated.

The calculator's present setting indicates that a p

-

0.2 is equivalent to a

SWR

-

1.5 and a

Return Loss

=

14 dB.

When the connectors on the

DUT on the reflectometer test differ from those port an adapter is often employed to make the measurement. However, the limitation to measurement accuracy caused by the adapter reflections is often severe. Suppose an adapter

SWR

(p

:

0.10 and R.L.

:

20 dB) is used tion measurement. Since the adapter has a p of

0.10, is not possible to measure

DUT p

Otherwise, the of in less with

7.22

a reflecthan

DUT reflection may be obscured by the adapter constant reflection.

Because the adapter reflection error quantity for all measured values p, reflectometer system directivity denoted by the it

0.10.

is it is

A term a added to the coupler directivity to obtain an effective in equation (5). Directional couplers like the

11691D and

116s2D are available with precision

APC-7 connectors on the main arm output, so low

SWR APC-7 adapters can be utilized.

The use of low-cost interseries adapters generally leads to an effective directivity under 20 dB in the microwave frequency range.

line including a stationary probe followed with adjustable depth by a well-matched 6-dB attenuator. The detector stationary probe is matched to the detector movable ing probe on the slotted-line carriage, thus compensatfor the frequency response of the probes as well as variations in sweeper output power. The

6-dB attenuator improves the source match. Note able-persistence oscilloscope that a 1B1T variis used as a display for the

8755,4..

Variable persistence or some other means of trace storage, such as time-exposure photography,

X-Y recordings, etc., is required for swept SWR measurement.

Operation stood of a swept slotted line is best underby first recalling a single-frequency slotted-line measurement.

From impedance other than transmission-line known that a uniform, lossless its characteristic impedance have two traveling waves on theory, line terminated it in is an will it.

Besides the forward or incident traveling wave,

Ei, there will be a reverse or reflected traveling wave, E", whose magnitude is dependent on the terminating impedance and of Ei.

While the reflectometer measures the value

Ei and

E" through the use measures of directional couplers, the slotted line the standing wave resulting from the interaction of

Ei and

E".

swR

=

E-u'

-

*,

++

E-in

Ei-E"

-\

*

1'-p

p

(Bl

Thus moving the slotted-line probe over mum a miniof one-half wavelength at a single frequency insures the detection mination of E-"" similar fashion. However, and the

E-io source and of

SWR.

The swept slotted line operates in a is the swept deterinternally, and the detector probe is moved over a distance of at least one-half wavelength at the lowest frequency so that both sampled. single

E-.*

Instead and of

E^i. reading of frequency the

8755A/1.81T can be used out the maxima and minima the out at standing

E-"* all and wave

E-i' to frequencies are at a read over which the source is swept. Since maxima and minima

8620

Swooper

8755/1817

Test

Set

SWEPT SLOTTED-LINE MEASUREMENTS

Slotted Iines accomplish high-frequency swept impedance measurements by sampling standing waves along a transmission line.

Because the directivity performance of above B GHz, many coax reflectometers deteriorates the swept slotted line with its consistent high-frequency performance measuring technique. measurement is an important impedance

It is particularly useful in the of small reflection coefficients fsmall swRJ.

Figure 30

Sweeper low-pass adapter. exhibits the basic

8755/8778 swept slotted-line system, operating output is fed to

18

GHz.

filter before entering the slotted-line sweep

The sweep adapter from 1.8 GHz through the modulator and a is a short piece of slotted

19

Mod

Dot R

L.P,

Filter

8178 Sw€pt

Slotd

Line

Figure

30.

HP

8l7B

Swept Slotted-Line System for swept SWR measurements between 1.8 and 18

GHz.

An 8755 system with

181T storage oscilloscope is used to detect and store the data.

.= o cc

= co cc o

.:

6

E

Frequencv (GHz)

3la

Frequency (GHz)

3lb

Figure and

31. Displays resulting from typical swept

SWR test with the

8178/8755 system in

Figure

30: a) low-pass filter with a

13

GHz b) a

10-dB pad. The measured SWR

(dB) of the pad is less than

I dB

(1.12

SWR) over the full band.

cut{ff frequency do not occur at the same point on the line at all frequencies, several sweeps probe at a are required, each with different position on the slotted line. result is an envelope like the ones shown the

The in

Figure

31.

The sents lower level of the envelope in

Figure g1 repre-

E-io while the upper level represents E*u*.

Because the display velope at the is logarithmic, SWR [dB) can be read directly by measuring the vertical thickness of the enfrequency of interest. This occurs because:

2o log

E*"*

-20 log

E*i. = zo log

20log

SWR = SWR

ffi

(dBJ

=

(6)

It follows directly from equation [a] in the introduction that: swR =roe-'(!lY*lqEf

)

(7)

The basic procedures for swept SWR measurements are using the 8178 swept slotted line outlined below.

with

BTSS/1,BIT

Set Up and Calibration

7.

Set up equipment as shown in

Figure 30 and set sweeper range for maximum power over the frequency of interest.

2.

Remove

R detector frbm sweep adapter and use it to measure the absolute power at the test port of the slotted appropriate line.

[Be careful that the power leve] is for the DUT.)

3.

Using the

OFFSET CAL vernier and

OFFSET dB thumbwheels, adjust the average insertion of the

A probe to sample power 25 dB below the absolute level measured in step 2.

4.

Return the R detector to the sweep adapter and terminate the slotted-line test port in its characteristic impedance.

Adjust the average insertion of the stationary probe [R probe) to sample power

19 dB below absolute level measured in step

2.

20

Measurement

5.

Connect the DUT to the slotted-line test port.

Be careful to terminate two-port DUT's in their characteristic impedance

[Ze).

6.

Set

8755 for ratio measurement the slotted-line carriage envelope similar to those until a

[A/R] and move swept SWR

The

OFFSET

CAL vernier, OFFSET dB wheels, and resolution push-buttons used may to obtain the most desirable display.

(dBJ in

Figure

31 is recorded.

thumball be

7.

Store the display of the SWR tdB) envelope and observe SWR

[dBJ at the frequencies of interest.

In the measurement procedure, the insertion of the

A-probe is adjusted to sample energy

25 dB below the incident signal level. The dynamic range available for

SWR

(dB) measurement absolute is the power level sampled difference between by detector noise level. For instance, if the the probe and incident the the signal level at the

DUT is 0 dBm and the detector noise level is

-50 dBm, the probe can sample maximum and minimum signal levels between allowing 25 dBm)l in dB

SWR of

-28 dBm dynamic range and

[-2b

-80 dBm dBm,

-(-50

(dB) measurements (17.8:1 SWRJ. Insertion may be increased to measure larger values of

SWR

[dBJ, however errors resulting from probe reflections will also increase and this technique is rarely used to measure large SWR's

{>3:1

SWRJ.

The

R-probe insertion is

6 dB above the A-probe insertion to compensate for the 6-dB attenuator in the sweep adapter.

the SWR of two port devices with very low insertion losses is measured, the quality of the termination used becomes important. This occurs because any reflection with the reflection from the DUT.

Because of this problem,

When it from the load termination has become will a standard practice load, such as the HP

905A, add to vectorially use a sliding to terminate the DUT.

\-f

i

L line

A sliding load is simply a length of transmission with a movable termination, permitting the phase angle of the separate the voltage reflected the phase angle of the load by the load to be varied while the magnitude is held constant. By manipulating reflection, it voltage reflected by the DUT is possible to from the voltage reflected by the load.

A useful technique is to mechanically couple the probe carriage to the sliding load so that the distance between is the slotted line probe and the load termination held constant. This keeps a fixed phase angle between the incident voltage,

Ei, and the part of the reflected voltage due to the termination

Er,.

Thus the thickness the swept

SWR

(dBl envelope becomes: of

E,

+ E,I

+_LEJ fB) i where

E. n'

I

>> swept is the voltage reflected it normally is, by the

DUT. the thickness of

If the is an excellent approximation of the

SWR being measured.

| n"

I as

SWR

(dB) envelope is due principally to E" and

The principal effect of the load reflection,

Er,, is to move the entire swept SWR

[dBJ envelope up or down on the display; the effect on envelope thickness is negligible.

SWEPT SLOTTED.LINE ACCURACY

The primary sources of uncertainty ted-line measurements are residual reflections, slope, probe reflections, and instrument errors. in

As swept in the slotcase of the reflectometer these terms may be combined into a composite enor equation:

Ap=A*Bp*Cp2 te)

The

A term is composed of the residual reflections and slope from the output connectors on the line and the slot ends, of the slotted line.

Residual reflections result while slope is due to mechanical tolerances causing the insertion of the probe to vary as the carriage. These it is moved on two terms correspond directivity in a to effective reflectometer system and represent the lower

Probe limit for p measurements using the slotted line.

reflections are the primary contributors to the

B term.

The C term reflections is composed of the slotted line; this is the same residual reflection term that combines entirely with of the residual the slope error to form A.

Since term. the swept slotted the measurement of

For the HP

8178 slotted line, the residual reflections are

1.04

SWR

[p

=

0.0196J at 18 GHz and slope error is

0.1 dB in

SWR

[dB]

[p two errors, an

A term equal to

0.02b6 (at 18 GHz) is obtained;

0.0256 can be converted dB may then be used small tometer measurements. line is primarily applied to p's sidual reflections and slope

(SWR'sJ, error to a return loss ters are utilized, adapter reflections as in the

B comprising by residual reflections and slope errors. and the

C terms in equation

(9J are usually not significant.

In this situation, caused error analysis can be concentrated on the re-

=

0.006J. Summing

A these of

32 which is equivalent to the directivity error in reflec-

The

Reflectometer Calculator to determine the uncertainty limits

If adapreflectometer measurements add directly the uncertainty.

to the

A term, increasing

In general, slotted lines offer greater accuracy than reflectometers but are more difficult to use.

Also, careful measurement techniques are required in slotted-line measurements, caused or such factors as probe reflections by excessive coupling can contribute major inaccuracies. However, the swept slotted line is essential to the measurement of low p's ISWRJ such as those encountered on cables and connectors, at 78

GHz.

DIRECTIVITY

MEASUREMENTS

Since the directivity of directional couplers is often the limiting factor in reflectometer measurements of impedance, it is an important quantity to

Cefine and measure. power in the secondary arm when all the power in the main arm power power

Directivity is defined as the ratio is flowing in the in dB of the forward direction to the in the secondary arm when an equal amount of in the main arm is flowing in the reverse direction.

Measurement achieved the configuration of or waveguide with the setup in

Figure

32.

By replacing the waveguide components with a similar coupler their coax directivity is counterparts, configuration employing the

8755 system may be used to measure coax couplers.

The setup to the basic reflectometer configuration of

Figure 27; however, the reverse coupler in

Figure 32 is identical is now the DUT. As in the reflectometer, the

415E

SWR Meter sweeper is employed as an amplifier, and the is accordingly amplitude modulated at a 1-kHz rate.

X-Y Reorder

6:1

/[

\

/\ lt

Figure 32.

Swept directivity measurement system for waveguide directional couplers (2.6 tution reflectometer to

40

GHz).

The system is similar to the

RF substiin

Figure

29 and may be replicated in coax by substitution of the appropriate components.

An

8755 system may be used in place of the

415E.

27

Calibration precision attenuator directivity of is established set to using a short specific values lines are drawn on the X-Y recorder with near the coupler under test. Calibration the the grid for each setting

of the precision attenuator.

If a sliding short is utilized, it should be rapidly phased during each sweep so source match variation with load phase will be averaged about the calibration grids.

After the grids have been drawn, the short is replaced by a sliding load and the precision attenuator set to zero.

With the sweeper set for a very long sweep time

()40 secondsJ, a final sweep is triggered and the sliding load continuously phased during the sweep.

X-Y recording is exhibited in

Figure

33.

A typical

18

16

€rl

I

E.,,

Ero

8

6

4

2

415E.

The short is replaced by a sliding load, and attenuator decreased for an on-scale reading on

415F,.

the the

24

22

20

SIGNAL SEPARATION CHART

DIFFERENCE IN dB IS BETWEEN

MINIMUM AND MAXIMUM

RETURN

LOSS MEASUREMENTS.

CORRECTIONS IN dB TO BE ADDED

SEPARATELY TO SMALLEST dB

R€TURN

LOSS READING-

12

14

CorEdion in dB

Figure 34. Signal Separation Chart used for separating two signals when their sum and difference are known.

\'./

$co

.E o'

642

$I a.2

Frequency (GHz)

Figure 33. Typical

X-Y recording for a swept directivity measurement of a precision, multihole, waveguide directional coupler. Note that the envelope created specified 40-dB by phasing the sliding load calibration grid. Thus is always below the it is not necessary to spot check the directivity at a

CW frequency.

By phasing the load and sweeping ble phase combinations slowly, all possiof the directivity signal and the load return loss are encountered at the reverse detector.

Thus sum the detected signal swings between the vector and difference of the two signals as the load is phased. The coupler under test is within its directivity specification if the swing, represented by envelope in

Figure

33, less the coupler transmission loss is below the specified directivity calibration grid. For instance, if the coupler under test had a coupling coefficient of

10 dB, a 0.46-dB transmission loss would be subtracted from the value of the swing noted on the

X-Y recording.

For coupling coefficients of

3 dB and

20 dB, the transmission losses are

3 dB and

0.04 dB respectively.

If the swing exceeds the specified directivity calibration grid, a more precise determination of directivity can be made aration Chart at a single frequency using the

Signal

Sepin

Figure

34.

The sweeper should be set to the frequency reconnected of interest fin

CW mode) and the short to the coupler under test. The precision attenuator can now be set to a value NEAR the expected used directivity, and the RANGE and

GAIN controls to establish a convenient reference level on the

22

Phase the load and note the maximum and minimum values on the

41.5E.

The precision attenuator may be used to determine the minimum and maximum nal levels

(in dBJ with respect to the original sig- reference .

\-

,i-

Ievel established on the 415E during calibration. is accomplished by phasing the load for either a

This maxi-

V mum or minimum and using the attenuator to re-establish the reference level on the 415E. Once the maximum and minimum are obtained their difference can be calculated and the

Signal Separation

Chart applied.

The difference signal levels

Separation is in dB of the maximum and minimum entered

Chart in

Figure

34.

The two curves on the chart are intersected ordinate entry, and the on the ordinate on a two horizontal of line the correction factors

Signal from the noted on the abscissa

(directly below the intersections). For instance, if the difference between the maximum and minimum signals is

12 dB, the two correction factors would be approximately 4 dB and 8.5 dB.

The two correction factors are added separately to the minimum dB reading noted in the sliding load test.

The two resultant numbers are the actual values of coupler directivity less transmission loss and the sliding load return loss. However, which signal level sliding load reflection. grade the sliding load

To it resolve is still not known is the directivity and which is the this ambiguity, deby taping a piece of solder on the load and repeat the test. nals

A second set of separated sigwill be obtained; one signal level in each set should remain unchanged.

The signal level plus the transmission loss of the coupler is the directivity.

In coax measurements, sliding loads. Therefore, it is difficult to degrade the it is usually convenient to either a different load or a different coupler use \-/ for the second measurement.

t

t

4

I

\-.

SOURGE SWR MEASUREMENTS

Because certainty source SWR, like directivity, causes unin high frequency measurements, ful term to quantify.

A system for put

SWR of microwave sweepers is it measuring shown is in a the useout-

Figure

X-Y Recorder

752D

Direc{ional

Coupler

415E

SWR

Merar

7ffi

Cal

9148

Sliding Load

\q

Test

920A

Sliding Short

Figure 35.

System output for measuring the source

SWR of sweepers with final in waveguide.

35.

Whether the sweeper ternallyJ leveling coupler, must be is in leveled (internally or not, the final output, either front panel or waveguide. The system

Figure 35 cannot be replicated in coax quality sliding shorts are not available.

because or exin high-

The results of a swept-source

SWR test are shown in

Figure

36.

Both X-Y recordings were obtained from a sweeper with an external waveguide leveling loop.

Figure 36

[aJ ter indicates that the leveled source has betthan aL.'J.:7 source SWR.

In

Figure

36 (b), the leveling loop is opened, and source SWR is close to

2:1.

With the coupler output terminated in a sliding load and the precision attenuator set to specific values of source SWR, three calibration grids are plotted.

The sliding load sweep the sweeper is replaced by a sliding short and a final triggered. secondsJ, and the sliding short should be phased continuously during will

The sweep time should be long flection coefficient the sweep. see all

As the short is

[)40 phased possible phase angles of rewith unity magnitude.

Any portion of the short reflection that is re-reflected by the sweeper will add with the incident signal and be coupled to the detector. The result is the envelope seen in

Figure

36.

Precision attenuator settings for the three bration grids are determined using the formula cali-

6 dB

*

20 log [1

-Lp"ptJ. p" is the reflection coefficient corresponding to a specific value of source SWR while p1 is equal to unity since a short is used as the load during test. about

6 dB which the is introduced as an arbitrary upper and lower offset calibration limits may be plotted.

Thus the precision attenuator settings for three calibration grids are:

Upper grid setting

Middle grid setting

Lower grid setting

=

6 dB

= 6 dB

=

6 dB

*

20 log

*

20 log

{1

[1

*p")

[10)

[11]

-p-] (72)

For example, a SWR of 2:1 corresponds to a p of

0.33; substituting into the equations [10), (11), and

[12J, the corresponding attenuator settings are 2.5 dB, 6 dB, and

8.5 dB respectively.

The accuracy of source SWR measurements is dependent on the high directivity

[)40 dBJ of waveguide directional couplers, assuming that all the signal variations seen at the measurement detector are caused by source match. Because coax couplers typically have lower directivities fusually 30 dB), small values of source SWR cannot be distinguished in coax systems.

!

t

!

E,

3

I

51.0

2.5

Frequency (GHz) Frequency (GHz)

36a 36b tigure 38.

Swept source

SWR measurement

From a) and b) above for a) a sweeper externally leveled in waveguide and b) the same sweeper with the leveling loop open.

it is evident that leveling improves sweeper source match from a near

2:l

SWR to less than 1.1:1

SWR.

23

During phasing the plotting the load will of cause calibration grids, rapidly fine-grain variations in the grid. Similarly to return-loss measurements, the effects of the load reflections on the measurement may be minimized by taking the mean as the true value of of the fine-grain variations the grid line. The

415E in

Figure

35 the acts solely as an amplifier for the vertical input of

X-Y recorder and could be supplanted by an

B7bb.

\-/

\-' l t

24

\-.

EffiffiffiffiffiHffiffi8ffiffi as a

Description function of the behavior of multi-port networks of frequency entails the measurement of transmission characteristics as well as impedance characteristics. quency

As in impedance measurements, high fretransmission measurements require the application of transmission-line theory and the subsequent sampling of the appropriate traveling waves.

Network transmission characteristics are usually expressed caused in terms of a transmission coefficient, defined as the change in power level (at the r. z is loadJ, by inserting a network between a reflectionless source and load. According reflectionless to transmission-line theory, implies that both source and load impedances are equal to

Zs, where

Zo is the characteristic impedance of the system. Under these conditions, the change in power level is purely a function of the device under test. Mathematically, z is defined incident and transmitted traveling waves: in terms of the the transmitted sigrral.

This by employing detectors with low reflections or is-olating the detector pler load with or the use in oltput the auxiliary arm detector of pads or may is usually of the main arm terminated also ferrite isolators.

a directional in a accomplished

Zo load. be isolated cou--

The through

The objective the techniques of this section is the development of for measurement of transmission coefficients (both gain and attenuation).

Advantages, applications, and accuracy discussed of the various techniques will be in addition to the basic measurement procedures.

Simultaneous reflection/transmission measurements will be presented in the conclusion of the section.

Transmission

Coefficient

Magnitude:

'

=

S1-]

111 llrnc

Loss or Gain

(dBJ: r(dBJ

= -20log

,

(2)

The transmission coefficient magnitude defined in equation

(1) may lossJ ment is the same be taken to be or greater than unity

(gain). The basic measurein assure less than either that power levels are consistent unity case; fattenuation however, care the incident and or must transmitted with the requirements of both the network under test and the measurement system

{i.e. amplifier linearityJ

[i.e. detector burn-out).-

Broadband swept-transmission measurements are accomplished by sampling both the incident and transmitted traveling waves der test. z with respect to the network unis determined by performing the ratio of the two detected signal levels. Directional couplers, directional bridges, and power splitters are ployed typically emto sample the incident wave. The transmitted wave may be sampled directly the network under test or at the output port through the auxiliary of arm of a directional coupler. However, the primary concern in any transmission measurement configuration is maintenance of the source and load impedances near

Zs.

A source impedance near Zo is obtained either by leveling the sweeper or directly performing the ratio of the incident and transmitted signal.

A leveled sweeper not only maintains but also makes the sweeper appear close leveling in lEq""

] at a constant level to a Z0 source.

However, instantaneously sampling and ratioing the incident and transmitted signals has the same effect as in maintaining a

Z0 source and in compensating for variations

I

Ei."

| {see

Appendix

A}.

Maintaining the load near 26 mizing the reflections from the is obtained detector used by minito detect

25

COAX TRANSMISSION MEASUREMENT WITH

DIRECTIONAL

COUPLERS

Swept-transmission data like that shown in

Figure

37 can be obtained from the

875b transmission measurement system in

Figure 38. Wide applicability in

!

o0

.9

F o

3go

'a

.2

E

E60

Frequency (GHz)

Frequency (GHz)

37a

37b

Figure 37. Transmission measurements

3.5

GHz) and b) low-pass of a) bandpass filter

(2.5 to filter

(10

GHz) as seen on the display of the

8755 Frequency

Response

Test

Set.

swept measurement characteristics, of both gain and attenuation is achieved as a result of the system's speed and versatility.

Because of the

8755 system's broadband detection the configuration shown in

Figure

38 is capable of measurements from

15 MHz to

18

GHz, assuming the sweeper and directional coupler operate over the frequency range of interest.

L- p.

Filter

T"*

Det

B conventional swept-transmission measurement systems. The incident-signal sample coupled through the coupler auxiliary arm may be fed either to the sweeper in a leveling loop or directly to a ratio measuring display like the

8755 system.

In both cases, tompensation for variations source

SWRJ. in incident power is achieved while match is improved to a near

Ze condition ((1.2

However, ratio-measurement systems are generally preferred because set quicker and easier than in up most and calibration are conventional leveled systems fsee

Appendix A).

Since the 8755 system a low-pass sion measurement setups utilizes broadband detectors, filter must be included to in many transmiseliminate sweeper harmonics and insure maximum dynamic range. While a low-pass filter is generally adequate protection against inaccuracies caused care must be taken

Feedthrough of measurements, the but in modulator feedthrough

27.8 by spurious siglal levels, some measurements to avoid and sweeper subharmonics.

kHz the

L1665B modulator may cause errors this can be avoided by adding an

HP 11668A 50-MHz high-pass modulation signal from in certain gaiu filter after the modulator.

Similarly, sweepers using.multiplier techniques generate subharmonics as well as harmonics, necessitating the use of bandpass rather than low-pass filters.

Both gain and attenuation measurements are possible with szSS transmission measurement system shown in

Figure

38.

In the discussion of measurement procedures, attenuation will be dealt with first followed by a review of any procedural or setup changes required to determine gain.

11691D

D.U.T.

Directional

Coupler

Figure

HP

38. 8755 coax transmission measurement system utilizing the l169lD 2 to

18

GHz directional coupler.

In transmission measurements as in the reflectometer, the

8755 system's

R detector samples the power incident on the DUT through the auxiliary arm of directional coupler. During calibration the

B detector is placed directly on the coupler main arm

(see Figure a

3B).

Using the 8755 system's capability to measure absolute power, sweeper output power at the coupler main arm is adjusted characteristics to a level consistent with operating of the DUT and the dynamic range !equirements of the measurement system.

After the appropriate power levels have been chosen, calibration for a

0 dB transmission coefficient

8755 system is established with the in a ratio measurement mode. The measurement is accomplished by inserting the DUT between the coupler main arm and the may be adjusted to

B detector. Sweep speeds accommodate continuous CRT display data readout or an

X-Y recorder.

on a

Because capabilities, the

8755 system has ratio-measurement the output of the sweeper in

Figure 38 is not necessarily leveled. Unless the

DUT is input power sensitive, leveling is not required as it was in most

26

Set

Up and

Calibration

7,

Set up the equipment as shown in

Figure

38 and set the sweeper to sweep the frequency range of interest.

2.

Use the 8755A position controls to adjust the

0 dB/0 dBm position line to a convenient graticule, hence referred to as the position graticule.

3.

Connect the B detector to the main arm of the directional coupler.

With

OFFSET

CAL position adjust sweeper output power in the

OFF for approximately dBm,

*10 dBm at coupler main arm adjust

[if

<+10 for maximumJ.

This assures maximum dynamic range; however, care should be taken not to exceed the

8755 system input power specification of the DUT.

Turning the

OFFSET

CAL to

OFF calibrates in absolute power

[dBmJ.

the

4.

Set 8755 system for ratio measurement tB/R).

5.

With the

OFFSET dB at

:t00 and OFFSET CAL to

ON, use the OFFSET

CAL vernier to average any variation in position graticule.

CRT may be used the calibration trace about the

A grease-pencil recording on the to store the variations in the calibration trace for future reference.

Note: The

8755 transmission measurement system is now calibrated for the swept measurement of transmission coefficient.

If the

OFFSET

CAL ver-

\-/

\./

\-.

nier is moved during measurement, the calibration will be destroyed.

Measurement

6.

Insert the DUT between the coupler main arm and the B detector.

7.

Use the

OFFSET dB thumbwheels and the

RES-

OLUTION push buttons to achieve a suitable display.

8.

The

OFFSET dB thumbwheels can be used for accurate substitution measurements readout (to the nearest dB) any point on the of attenuation trace. Once the position graticule, resolution may without the trace leaving the screen.

and digital thumbwheels have been used to bring a point of interest to the be loss at increased

Two additional factors are considered measurement: in a gain

9.

Add the 11668A

[bO-MHz high-pass with the

116658 modulator.

filter] in series

10.

Adjust the sweeper output-power level at the coupler main arm so consistent that the absolute power is with the input power specifications of the DUT and the dynamic range requirements of the

8755 system.

This is to avoid saturation of the

DUT.

The 116684 high-pass fiIter prevents amplification by the DUT conflicting, i.e. exceeds the

27.8

Two points are critical power level of in saturation does gain specified measurement range. feedthrough.

not occur. Also, the input power level plus the expected gain kHz modulation in of determining measurements. the the DUT must incident

The power not level must be in the normal input range of the DUT so that exceed

*10 dBm which is the upper limit of

1.1.664A detector's

If these two criteria are the normal DUT input power plus gain

*10 dBm, a pad may be introduced between the output gain can of the pad is used still the DUT and the detector. in calibration as well as measurement, be read directly from the

As long

OFFSET as dB thumbwheels.

Coupling

Compensation

In the single coupler configuration of

Figure

38, only the frequency detector

Using another cancels proving

R coupling variation sees the coupling which is typically

:L1 dB most the variation with coax identical coupler, the setup in couplers.

Figure

39 of the coupling variation as well as imeffective detector match. with frequency for is

The effect of reduced tracking between couplers which is typically to the i0,5 dB,

Impedance match at the the

SWR

DUT output is improved from of the detector, typically 1.5 SWR, to the main-line

SWR of the directional coupler, typically

1.3

SWR. The main arm output terminated of the coupler must in its characteristic impedance.

be

This technique trades dynamic range measurements

DUT for flatness. is the same as namic range is decreased by the coupling factor.

Assuming in

If the the in attenuation incident power at the uncompensated setup, dy-

*10 dBm incident power and 20 dB nominal

27 y

Mod

L.p.

Fitrer _

Det

R x l

Figure 39.

Coax transmission measurement system employing coupling compensation. The second coupler compensates the transmitted signal for the coupling variation seen nique at the

R detector.

However, the tech.

gives up some dynamic range in attenuation measurements to achieve flatness.

coupling, figuration the dynamic range of the compensated conis a0 dB as opposed to the

60 dB available in the uncompensated setup.

Of course, the dynamic range of gain measurements is unaffected by the addition of the compensating coupler

(see Figure

40).

Frequency (GHz) tigure

40.

Broadband gain measurement of a 4 to 8

GHz TWT amplifier.

The roll-off in gain outside of its normal operating region is clearly observable.

The measurement was accomplished with the system in

Figure 39.

The measurement procedure for the setup in Figure

39 is essentially the same as for the single coupler configuration in

Figure

38.

The two couplers are connected directly together for measurement.

for calibration and the DUT inserted

- - -cal- \

In certain attenuation measurements where limited sweeper occurs is available, the

6-dB loss that limit dynamic range.

In the corresponding configuration shown output power in both arms of the 116674 power splitter may in

Figure 28, the main arm transmission loss of a 20-dB means directional coupler is a negligible

0.04 dB. This that approximately

*10 dBm of the swept power is required at the coupler input for

60 dB range while

*16 dBm is required of dynamic at the power splitter input to obtain the same dynamic range'

The operating procedure for the measurement configuration in Figure 41 is fundamentally the same as in the two configurations using directional couplers.

The

R detector and

B detector are connected directly to the output arms power level optimized the DUT and the maximum dlmamis range.

Calibration is established of in the power splitter, and for the input the specifications ratio mode [B/R), and the incident of measurement accomplished by inserting the DUT between the power splitter and the

B detector. The results of several high resolution transmission measurements shown in

Figure

42, capitalizing on the flat tracking response of the

11667A power splitter.

The transmission characteristics of several other broadband devices are shown in

Figure

43.

\./

Figure 41.

Coax transmission measurement system

(dc to

18 GHz) power splitter.

usrng the

11667A

COAX TRANSMISSION MEASUREMENTS

WITH

POWER SPLITTERS

An

8755 transmission measurement system utilizing an 11667A power splitter is ing between output arms, source match, and dc

GHz frequency coverage are shown in

Figure 41.

Trackto

18 the primary advantages of using a power splitter instead of a directional coupler for couplers) in gain and attenuation measurements.

The

6 dB only real disadvantage of the power splitter is that more sweeper power is needed to obtain the same dynamic range in attenuation measurements.

Tracking errors are minimized, and the problem coupling variation vs. frequency is eliminated of because of the flat tracking response between output arms of the in a much flatter power splitter. This usually results calibration trace for grid lineJ than can be obtained conventional couplers.

Tracking error for the with

11667A power splitter is typically between 0.1 dB and 0.2 dB iompared errors splitter in generally ventional to i1 either vary between

1.05

SWR and 1.2 SWR. Conpower dB a in most couplers. are also minimized because characteristics

71667A. of two-resistor power splitters like the

When a powet splitter resistors, each equal pedance, to the system characteristic a source impedance very near

Zo is imrealized.

Typical source SWRs achieved with the 116674' power ratio splitters or a of is of leveling

Source-match the constructed configuration three-resistor matching of two construction do not exhibit good source-match characteristics.

(See Appendix

Figure

41 is

A.)

The measurement configuration in particularly convenient because the

11667,{ power splitter operates from dc to 18 GHz. Directional couplers like those used in

Figures 38 and

39 operate over much namorver bandwidths.

28

(n

.9

E o

3o

.9

1

E2

Pz

4

Frequency (GHz)

Frequency (GHz)

42a

42b

Figute 42.

High resolution transmission measurements ripple in a) bandpass of the pass-band filter

(2.5 to

3.5

GHz) and b) low-pass filter (10

GHz).

Both measurements were made with the system capitalizing on the power splitter's in

Figure

41, flat tracking. These are the same filters measured in

Figure 37.

\-/

When the power splitter is used to make attenuation measurements generated in excess of

40 dB, spurious signals at the R detector may feed through to the

B detector, causing added measurement uncertainty. Elimination of the spurious signals is achieved by introducing a

L0-dB pad between the power splitter and the

R detector.

Since incident power is reduced, spurious signal levels are correspondingly reduced, and greater isolation between the two delectors provided.

Because pads normally have reasonably flat frequency-response characteristics, between arms.

there is little deterioration in traiking

COAX TRANSMISSION

MEASUREMENTS

SUMMARY

Several transmission measurement configurations, two using directional couplers and one using a power splitter, have been presented thus mary is presented.

far. To clarify the typical advantages and applications, the following sum-

Measurements with

Directional

Couplers

SINGLE COUPLER (Figure

38):

Advantages: Maximum dynamic range measurements in attenuation is achieved with minimum sweeper power.

Disadvantages: Coupling variation of tr dB

(seen at

R detector onlyl makes high resolution measurements tracking errors.

difficult and contributes to

Applications:

General transmission measurements and high attenuation measurements,

40 dB to

60 dB (filter rejection bandsl.

COUPLING

3el:

COMPENSATION [Two Couplers,

Figure

Advantages:

Reduces the effect from of coupling variation t1 dB to 10.5 dB (typical).

Disadvantages: Usually sacrifice range

20 dB of dynamic in attenuation measurements and cost of additional coupler.

Applications: Gain measurements; high resolution measurements flow-loss devices).

E0

.9

@

320

.9

Eoo

F

60

Frequency (GHz)

43a

Measurements with a Power Splitter

Advantages: Tracking between output arms,

0.1 dB to

0.2 dB ftypicalJ, effective source

1.05 SWR

SWR, to

1.2 SWR [typical), widest operating range, dc to

18 GHz.

Disadvantages: 6-dB through loss in both arms may limit dynamic range, low isolation (12 dBJ tween output arms may allow of spurious signals.

befeed-

Applications:

All general transmission measurements, best vices technique for measuring low-loss dewith high resolution.

!0

o

E0 o

620 o

'E oo

F

60

43b

Another common technique for transmission measurements

[i.e. involves storing a measured reference level grease pencil, X-Y recorder] when the detector is directly connected to an unleveled source.

Next the

DUT is inserted and a second measurement made with the difference equaling insertion loss or gain. The advantages are low cost and maximum dynamic range.

However, the disadvantages are severe:

1.

2.

3.

dependence on source level stability source match error awkwardness of data manipulation.

Frequency (GHz)

Figure

(6.8

43.

Broadband transmission measurements

GHz) coupler. and b) the coupling coefficient of

2 of a) low-pass filter to

10

GHz directional fhe nominal coupling coefficient of the coupler shown in b) is l0 dB; notice the rapid rolloff in coupling after

11.6

GHz.

2S

X-Y recording quired by addition of transmission data may be acof an

X-Y recorder to any of the three measurement setups

Since

IAUX the

A auxiliary and a recording

AUX

BJ fsee outputs provide

Figures

38, 39, and +t).

on the

0.5

1B0T mainframe volts/scale division, of any trace at any resolution may be ob-

tained after recorder has been calibrated typical recording is shown in

Figure

44.

initially. A

I

^10

.A zo

E

t.o

.9

'E

40

0

E

'50

60

.-,ll

,l

t thumbwheels are used to compare the measured trace to the grease-pencil recording of the calibration trace; the

RESOLUTION must be at the same level as it was in calibration. This allows calibration tracking errors to be subtracted from the measurement, enhancing curacy in the same manner as grid lines on the

X-Y recorder. niques are ment

Both grease-pencil and widely used methods

X-Y of in transmission measurements.

recording accuracy ac- V techenhance-

\

\

L*t

'

,*qu"n"1y

(eg.)

Figure

44.

X-Y recording of the transmission loss through a band-pass filter.

The recording could have been made with any of the 8755 systems in

Figures

38,39, and

41.

4

Figure have been caused

45 exhibits the results of high-resolution attenuation measurement where calibrated plotted. The variations by the tracking errors ffrequency response] of the measurement system. in grid lines the grid lines are

Calibration tracking errors are essentially eliminated same by plotting the grids, and the grids may be used for many measurements as long as the system is not turned off or recalibrated.

COAX

COMPARISON MEASUREMENTS

It is often useful in both design and test situations to compare the transmission characteristics ponents. For instance, of two comit may be desirable to designate i particular component as a standard and align the transmission characteristics that of the standard,

In of other similar components to situations, such as troubleshooting, comparison data may be instrumental in locating

47. faulty or misaligned devices. Figure 46 shows a comparlson measurement made between two filters using the measurement configuration in notch

Figure

Aligament of a component under test to a standard is achieved quickly and easily using the real-time CRT display.

From the equipment setup seen that a comparison in

Figure 47, measurement is it can really be two simultaneous transmission measurements common utilizing

R detector.

Since the measurernent objective is comparison and both the signals seen at the

A and a

B detectors detector, will be ratioed with the signal seen at the

R the coupling variation with frequency is not

^18 g

.E

E o le

6zo

'd

'? zt e

'22

20 dB

CAt tr,t

IBRATION GRID-

../

Mt

)

TASURI

-l

14

D

Frequency (GHz)

Figure

45. High resolution attenuation measurement of a

20-dB pad.

Several calibration actual insertion grid lines are plotted on the

X-Y recorder before the of the

DUT, allowing calibration tracking errors to be subtracted from obtained using the measurement. a grease pencil on

Similar results could have the

CRT.

been

In a similar fashion, a grease pencil may be used to store the transinission calibration trace [or traces) on the

CRT. During the measurement, the

OFFSET dB

30

Figure

46.

Comparison measurement of two

2'GHz notch filters.

The

8755 system's real-time display allows rapid alignment of the two transmission characteristics.

a problem.

However, signal levels at it is important that the incident- \-/' the power-splitter output arms be nearly equal as possible both in calibration and test' as

\-.

L. P. Filter

11691

D

Diffitional

Couplei

'/

-

641

,/ DetA

-CAL \

Orre

D.U,T.

B

Figure

47. 8755 system for making comparison measurements in coax.

The system is essentially two simultaneous transmission measurements sharing a common

R detector, resulting in displays

The power splitter must be of like

Figure

46.

Note' three+esistor construction.

(See

Appendix

A.)

The calibration procedure ing the

A and

B detectors to the two output arms of the power splitter. Incident power appropriate is initiated is by adjusted connectto a level for both dynamic range and

DUT requirements, and concurrent calibration established on both channels.

The two

DUTs are inserted between the power splitter and the detectors, and the appropriate comparison accomplished.

If dynamic range is not a concern the and high resolution comparisons are desired,

R directional coupler may be replaced resistor power splitter. The basic by a twoquality of the comparison is not improved, but a flatter display at higher resolutions is obtained, low.

The basic measurement procedure is outlined be-

Set

Up and

Calibration

1..

Set up the equipment as shown in

Figure 47 and set the sweeper to sweep the frequency range of interest.

2.

Use the

S755A position controls to adjust the

0 dB/0 dBm position same line on both graticule, hence referred channels to the to as the position graticule.

3.

Connect the

A and

B detectors to the two output arms of the power splitter. With the

OFFSET

CAL in the

OFF position adjust the sweeper output power to an incident signal level appropriate for both the dynamic range and DUT input requirements.

4.

Set both channels of the

8755 system to ratiomeasurement mode

(A/R and

B/RJ.

5.

Set OFFSET dB to i00 and

OFFSET

CAL to

ON on both channels. Use the

OFFSET

CAL the vernier on

A channel to average any variation in the calibration trace about the position graticule. Use the

31

OFFSET CAL vernier on the

B channel to overlay, as closely as possible, the second calibration trace on the first. The calibration traces on both channels should be displayed simultaneously.

Note;

The

8755 system is now calibrated for a comparison measurement.

Calibration will be destroyed if either

OFFSET

CAL vernier is changed.

Measurement

6.

Insert the two DUT's between the power splitter and the two detectors.

7.

Use the OFFSET dB thumbwheels and

RESOLU-

TION push buttons to achieve an appropriate display. Of course, the controls on both channels must be identical for an accurate comparison.

be

X-Y recordings drawn of comparison measurements may with a single-pen recording by triggering two separate sweeps, one for each

DUT.

If a two-pen X-Y recorder is available the comparison measurement can be recorded in a single sweep.

WAVEGUIDE TRANSMISSION MEASUREMENTS

Transmission measurements formed in waveguide are perin a similar manner to measurements in coax.

Figure tem

48 shows an

8755 waveguide transmission syswith coupling compensation that swept measurements is capable of in waveguide bands between

2.6

GHz and L8

GHz The only major differences between

Adapt6r

Coupler

X-Y

Recorder gZ55

T6t

Set

I

D.u.T.

1

I

\ -cnl-./

752c

I ulr€stlonal counler tigure 48. 8755 waveguide transmission systems using coupling compensation. Note that

HP 281 Series waveguide-to+oax adapters are required to connect the

11664A detector to the waveguide directional couple rs.

the system in

Figure

48 and the corresponding coax system in Figure

39 are waveguide-to-coax adapters needed to connect the

1L664A detectors to the waveguide system and the

L0-dB coupling coefficients of the waveguide couplers.

Because 10-dB directional couplers are available in waveguide, the trade-off between coupling compensation and dynamic range is not as severe as it is in coax where coupling coefficients are usually

20 dB.

However, a single waveguide coupler configuration similar to the coax setup in

Figure

38 may be employed if sweeper power is not adequate range to obtain the required dynamic with coupling compensation. Since there is no common waveguide component with the same tracking characteristics as a coax power splitter, the waveguide transmission system the best results with coupling compensation offers for general measurements.

Again, it should be reemphasized that the two couplers must be of the same design and manufacture to assure adequate tracking. The salient points

Summary in the

Coax Transmission apply equally well to waveguide.

With the physical similarities in the 8755 coax and waveguide transmission systems, that the theory of operation is it is not identical. surprising

The setup, calibration, and measurement procedures guide system for the wavein

Figure 48 are the same as those for the coax system in

Figure

39,

The results are shown of a typical. waveguide in

Figure 49. The

X-Y recorder and greasepencil techniques discussed measurement in reference to

8755 coaxtransmission systems all apply equally guide system.

well to the wave-

WAVEGUIDE TRANSMISSION

MEASUREMENTS

WITH

RF SUBSTITUTION transmission coefficient

GHz

An

RF-substitution technique for measuring in to 4o

GHz is shown waveguide in bands

Figu-re 504.

System swept from

2.6 li opera- v tion is based on a leveled sweeper and precision

RF rotary-vane attenuator. Before the actual measurement is executed, specific values of attenuation or gain are pre-inserted via the attenuator, and the results stored in the form of calibration grids on the X-Y recorder.

While

RF substitution is not as convenient as the

8755 system, it does operate above 18

GHz and is composed of more economical instruments.

,

X-Y Recordel

424A Xtal Det

/'

Dir

752C

Cuplers

D.U.T.

9108

Termination

Figure 50. System for measuring transmission coefficient using

RF substituti0n techniques. Similar cific values to the

RF-substitution reflectometer, speof attenuation

(or gain) are pre-inserted and stored using precision attenuator. The

DUT is then inserted and the measurement completed.

16

17

18 o

19

@ o o

20

21

E

22

G

F

23

24

Frequency (GHz)

Obtaining the appropriate amount of leveled power from the sweeper is the first step in calibrating the system in

Figure

50.

Essentially, the sweeper is externally leveled in waveguide with a point-contact diode.

Since the sweeper output quency and is leveled, power must remain constant load impedance. Thus, that the incident power level forward (or both as a function of freit bration and measurement.

As noted can be incident) assumed will be the same in caliin the discussion of

8755 transmission measurement systems, leveling accomplishes

415E the same effect as ratioing. Operation

SWR meter requires of the that the sweeper output be amplitude modulated modulation at a 1-kHz rate. 1-kHz amplitude is available internally on all

8620 and

8690

Series sweepers.

Figure

49. lnsertion loss of a waveguide flap attenuator set for 20 dB of attenuation. The measurement was accomplished with the system in

Figure 48.

32

4

An 8690 Series sweeper must be used at frequencies above 18

GHz.

With the DUT removed and the two couplers connected together, the

3B2A precision attenuator to pre-insert specific values manually triggering the sweeper is used of attenuation or gain. By for single sweeps, calibration grids are sequentially plotted of the precision attenuator.

After the grids have been plotted, the attenuator level, and the is set

DUT inserted. to a for each specific setting reference

A final sweep is triggered and the transmission coefficient plotted on the X-Y recorder.

If many similar devices are being tested, the grid lines, once plotted, may be used as an underlay for many measurements with the actual transmission coefficient plotted on translucent paper. However, the grid lines should be redrawn after long hours of testing or after the equipment has been turned off.

In

RF-substitution measurements, the

415E acts solely as an amplifier

In this function for the Y-axis of the X-Y recorder.

it could be easily supplanted by a

432A or 435A power meter. an

8755 system into an RF-substitution measurement configuration, instrument.

It is also possible to incorporate either as an amplifier or ratio-measuring

If the

8755 system is employed as a ratiomeasuring instrument, the sweeper leveling loop could be eliminated.

The results of typical measurement are exhibited in

Figure

51.

The RF-substitution system in

Figure

50 is capable of attenuation measurements of

45 dB to

50 dB. Gain measurements over the same range may be accomplished in a straightforward fashion. Of course, gain measurements possible of greater than 50 dB are by adding another attenuator in the measurement arm.

TRANSMISSION

MEASUREMENT ACCURACY

The primary sources of uncertainty in transmission measurements are mismatch and quency response and variations tracking euors. Tracking errors are essentially the result in of the

Mismatch errors result any time there differential magnitude freof z.

is a change in impedance in the system, such as a connector interface.

Interconnection of two impedances different from the system characteristic impedance

(Ze) results match uncertainty like that illustrated in misin

Figure

52.

A portion of the incident signal is reflected by the detector, pa, becoming an incident signal on the source impedance. Consequently, part of this signal is reflected by the source, p", resulting in the worst-case mismatch error of 1 tpap",

The mismatch uncertainty may be quantified using the HP Mismatch

Error Limits

Calculators in

Figure

53; the Mismatch

Calculator is on the reverse side of the

Reflectometer Calculator discussed error section, in the reflectometer

18.0

3

18.5

.E

19.0

1e.5

t

20.0

o

20.5

'.fi zt.o

fi

x.

g

F

22.O

5

A complimentary HP Reflectometer/Mismatch Error Limits

Calculator able from any Hewlett.Packard sales office.

is avail-

87s5

Test Set

Frequencl, (GHz)

Figure 51.

X-Y recording shows the transmission loss of a waveguide flap attenuator set for 20 dB of attenuation.

lncid€nt

Signal

As prior from in the coax systems, plotting calibration grids to the measurement calibration. allows

In the coupler tracking {frequency responseJ errors and

RF-substitution detector system, calibration grids also free the system from reliance on the square-Iaw performance of the to be eliminated readout detector.

This occurs because each grid was plotted at a known level of

RF attenuation or gain.

33

(1 i ps po)

Measurcd

Signal

Plus

Unertainty

Figure

52.

Schematic representation of mismatch uncertainty. A portion of the incident signal is reflected by the detector and then rereflected by source, generating the resulting uncertainty

11 -+ptpo).

rl

[email protected] !d ttt d€ w. db ,;lll

.lF Htb.-'oddrw f,dwfrded4lli.e

&-tLhSLlr=1.$d.p lrdd+t-!-ffihnrt11116

** gitrr = l-A k0=

'lbr.

I

Figure

53. HP Mismatch

Error Limits Calculator is used for computing the worst-case uncertainties resulting from the interface different from the system characteristic impedance.

0f two impedances

1.5

For instance, the SWRr scale is adjusted so that is under the black arrow. The MAX. MISMATCH

ERROR

LIMITS (dB) are indicated below and above the two

SWR: scales.

For a SWRz uncertainty are *O.225 dB and

-0.23 dB; this would be the maximum or worst-case calibration uncertainty that would result from interfacing a source of 1.5 SWR with a detector

=

1.3, the two limits of of

1.3 SWR.

Insertion of the DUT into the measurement system results in additional mismatch uncertaintv between the source and the DUT input port and the detector and the DUT output port. mismatch error analysis is presented

A detailed in

Appendix

C.

tions can require that these two parameters be measured separately, making measurements it difficult to observe the mutual interaction between the parameters. CRT photos of simultaneous transmission coefHcient/return loss made with the

8755 system are shown in

Figure

54.

The ability to make simultaneous measurements with real-time observation of the trade-offs between the transmission coefficient flatness and

SWR is a powerful tool for network optimization.

Uncertainty resulting from source mismatch duced is reby leveling the sweeper or making ratio measurements ftypically

1.1 to detector mismatch may

1.3 SWR].

In a similar manner, be reduced by isolation with a pad or directional coupler ftypically pad and <1.3

SWR for a couplerJ.

(1.2

SWR for a

I

6

54a

-3 o a

E20 c

,E

208 o o

+o2

.2

E ooF

Tracking calibration sponse errors resulting quency response of the trace and from differential fremeasurement subtracting the system can minimized by making a grease-pencil recording of frequency be the refrom subsequent measurements.

The magnitude of tracking errors resulting from ratio measurement inaccuracies fvariations with ?] is usually obtained from the appropriate technical data.

Appendix C offers further discussion of tracking errors.

40

I a

54b

-3 o

-

E20

Frequency (GHz)

I c0

'6 o zo3

.9

.2

+oE

F

SIMULTANEOUS TRANSMISSION/REFLECTION

MEASUREMENTS sary

In aligning broadband networks, it is often necesto compromise between transmission coefficient and reflection coefficient.

However, equipment limita-

34

Frequency (GHz)

Figure

54.

Simultaneous measurement coefficient of a) low-pass of return loss and transmission filter

(10

GHd and b) bandpass lilter (2.5 to

3.5

GHz) using an 8755 system.

\.1'

8620

Srveeper

L.P.

_J

"^

Filter

Det

R

/

\ IALBJ

11692D Dual

DiEtional

Coupler

875b

T6t

Ser

8620 Sweper

BZS5

Test

Set y

"'o

L. p.

Filter

Det

R

Cat

A

74J

I D.u.T.

Det

B

Figute 55. System for simultaneous measurement transmission coefficient using of return a single

116920

(2 to

18 directional coupler.

loss and

GHz) dual

Two measurement setups for making simultaneous return loss/transmission coefficient measurements are shown in

Figures

55 and 56.

Similarly to the comparison measurement, nal and the R detector samples the incident sigis common to both measurements.

The tector samples the reflected signal as the

B

A dedetector concurrently samples the transmitted signal.

Both setups incorporate the basic coax reflectometer while coupling compensation in transmission system measurements is offered by the system in

Figure 56.

The setup in

Figure 55 is essentially the single coupler transmission measurement with no compensation for the coupling variation seen at the R detector. Hence, the choice of setups is purely a function of the type of transmission data required. The criteria for the choice are the trade-offs between dlmamic range and compensation for coupling variation with frequency; these trade-offs are discussed in the transmission measurement summary.

Another setup for making simultaneous measurements using in

Figure

57. the

116664 reflectometer bridge is shown

It is also possible to use a hybrid combination of a power splitter and a directional coupler for simultaneous measurements.

While the bridge is convenient because of its broad frequency range and built in detectors, the dynamic range of transmission measurements suffers because of its high main arm loss

[9 dB]. Also, no complementary tracking relationships flike those in

Figure

56J exist when either the bridge or the power-splitter coupler combination are employed.

However, tracking errors are negligible compared to

Figure

56.

Simultaneous measurement employing an

11691D

GHz) directional coupler for coupling compensation.

(2 to

18 directivity errors in high-return loss measurements.

Directivity errors may be reduced by employing a directivity, narrow-band, coax directional coupler.d

high

The actual measurement involves calibration of one channel of the

8755 system and the other channel

DUT is inserted and the systems shown the in for return loss measurement for transmission coefficient; the measurement completed.

Figures

55 through 57, the

In

A channel is calibrated for a reflectometer measurement like the setups brated in

Figure

21, and the B channel is califor transmission measurement like the setup in

Figure 38.

After the DUT is inserted both traces may be displayed simultaneously with the

A channel controlling the return-loss trace and the B channel controlling the transmission trace.

862O

Sweper

Cal

A

7

-[r

I o.u.r.

8755

Test

Set

D.t

B

I

High-dlrectivity (some available

>40 dB), single.octave, coax.directional

from manufacturers such as

Narda and

Wavecom.

couplers are

35

Figure 57. Simultaneous measurements using the

11666A reflectometer bridge

(40

MHz to

18

GHz).

any and

Since the two measurements and the two measurement channels of the

8755 are virtually independent, of the previous techniques, such as grease-pencil

X-Y recordings, may be applied. Throughout this

A of the

8755 system has application note,

Channel been arbitrarily utilized for reflection measurements, while

Channel B has been used urements.

Except for transmission measfor the common use of the R detector both channels are independent and may be used to make any two measurements of the user's choice.

GAIN control to draw grid lines if required.

Once the

X-Y recorder has been calibrated, the detector may be connected to the point of measurement and a single measurement sweep triggered. be long enough (at least

2O

The sweep time should to

4O sec/octave) for the detector when to respond to all variations in power.

Thermocouple detectors have particularly long response times power levels under 300 pW are measured.

POWER MEASUREMENTS

Knowledge of absolute power in milliwatts or dBm is necessary when network parameters vary as a function of the input-signal power level. The power level at a selected point in a swept measurement configuration may be measured using the system in

Figure

58.

The

Hp

432 Series power meter with appropriate thermistor detectors measures absolute power between 10 mW and

L prW over a 1

MHz to

40

GHz frequency range. Measurements in

1

MHz to

18

GHz range are possible in coax while waveguide measurements are made between

2.6

and 40 GHz. Similarly, the HP 435A and

436A power meters measure and the appropriate thermocouple power levels detector in coax between

3

W and

0.3 pW at frequencies ranging from

100 kHz to

18

GHz.

After the appropriate power meter has been calibrated, system calibration may be established by varying the power incident on the detector and noting the corresponding variation on the

X-Y recorder. This is accomplished at a

CW frequency, using the

X-axis be

An

8755 system like the one in

Figure

59 may also utilized for swept measurements of absolute power in coax. POSITION controls are used to establish a convenient graticule as a

0 dBm reference and tem the sysis calibrated to readout absolute power from *10 dBm (10 mWJ

18 to

-50 dBm

(10

GHz frequency range.

If nW) over the 15 MHz to the

OFFSET

CAL is set to

OFF, the display power is automatically calibrated in absolute with respect to the 0 dBm reference graticule.

The sweeper may be adjusted for a continuous

CRT display during measurement, and the

OFFSET dB thumbwheels and

RESOLUTION push buttons may be used to obtain the desired display. X-Y recordings may be obtained in a normal fashion utilizing the

8755 system's auxiliary outputs.

Power can be measured at any of the

8755 system's three detector inputs, and the response of any two of these detectors displayed simultaneously.

8755

While the is sensitive to low levels of absolute power and has fast response time,

Series and

435,4. it is not power meters poorer detector match fsee urement accuracyJ and thb as accurate as the

432 primarily because of its discussion of modulator loss.

power meas-

X-Y

Reoder

/B5A

Pow€r Meter

(rR2A Pow€r Meter)

8481A Powei

Sonsot

(8478 Thermistor)

Figure

58.

System for swept measurement either the

435A power meter

(or

432 series).

of absolute power using

36

Figure

59.

Configuration for measuring swept absolute power using the

8755 system.

INPUT

POWER VS.

OUTPUT POWER OR GAIN

A system for measuring output power or gain as a function of input power at a fixed frequency is shown in

Figure 60.

The system is particularly useful for determining amplifiers. the saturation or compression point of

In the system of

Figure 60, the incident power level is swept over approximately a 30-dB range by amplitude modulating the

8620 sweeper output with its own sweep voltage. The maximum incident power level can be further manipulated using the step attenuator. Using provide the auxiliary outputs of the

BZ55 system to a voltage proportional to the incident power measured

CRT vertical axis may gain may be calibrated for absolute output power by displaying the signal at the B detector or for by at the R detector, the horizontal axis be performing for absolute input power.

The calibrated the ratio B/R. of the

Measurement is accomplished by inserting the DUT between the power splitter's output arm and the

B detector.

+10

E

@

I

!0

@

3 o

-10

Figure

61. lnput power vs. output power for an amplifier at

1

GHz.

Output power increases linearly with input power until the input reaches

-8 dBm. Note that the diagonal trace or the input power seen at the

R detector must be displayed in order to obtain horizontal deflection on the screen.

8620

Swesper

8755

Test Set

SEp Atten t18674

Pomr

Splitter

- rcAL-

\

Ampl lJnder Test

Det

B

-20 -10

0

+10

Power ln (dBm)

Figure 60. System for measuring input power vs. output power or input power vs, gain.

The input power is swept using the

8620's sweep ramp to drive its internal

PIN attenuator. The horizontal axis of the

8755 system with is calibrated in absolute power by driving the horizontal input the auxiliary output proportional to the signal seen at the

R detector.

Because some sweepers modulators and others do not have internal PIN are always internally leveled

{usually at frequencies

(1

GHzJ, it may not be possible to sweep output power using the EXT AM input. In these situations; sweep an external modulator may be used to power. The sweeper ramp

[0 - 10

V for

8620 and

0

-

15

V for

8690 sweepers) external modulator, may be used to drive provided compatibility exists.

the

The results of input power vs. output power and an input power vs. gain measurements are shown in

Figures

61 and

62 respectively. Note representing the power sweep is that diagonal trace always present on the display. Using

SET the

OFFSET dB thumbwheels and OFF-

CAL vernier it is possible to measure parameters like the 1 dB compression point.

37

Figure

62. lnput power vs. gain for an amplifier at 1

GHz.

Gain remains constant (22 dB) with input power until the input reaches saturation occurs. The diagonal

-8 trace must be displayed as dBm and it was in

Figure

61.

POWER MEASUREMENT ACCURACY

The primary sources of inaccuracy in measurements of absolute power are source-detector mismatch, detector frequency response, and instrument errors. Mismatch errors are or display by far the most serious uncertainty.

A portion of the power incident on a detector from a Zs transmission line will be reflected and lost. For any particular detector reflection coefficient, this

MIS-

MATCH LOSS (dBl may be calculated using the

HP

REFLECTOMETER

CALCULATOR. Similarly, tion of the reflected power a porwill be re-reflected by the source fprovided uncertainty to it the is not total reflectionless), adding mismatch loss.

A

HP an

MIS-

MATCH

ERROR

LIMITS

CALCULATOR may be used to determine the uncertainty limits.

For example, when measuring the

Zo power sweeper tector

SWR loss of of

1.S

0.18 the

REFLECTOMETER

CALCULATORJ and match of which has

1.5 SWR, a a source mismatch uncertainty of

*O.g+ dB and

MISMATCH LIMITS

CALCULATORJ tered. The net uncertainty in dB ffrom a mis-

-0.96 dB (from will the power with be from a a dethe encounmeasured is

+0.18

At a dB

CW

[-0,1s

-[-0.36]l and frequency mismatch

-0.52 errors dB can

(-0.18 -0.34).

be eliminated for all practical purposes, using a slide-screw tuner.

Detector frequency response and display mentation errors are provided or instruwith the technical data on most power meters and similar equipment.

All

Hewlett-Packard instruments measuring absolute power are National

Bureau tainty of

Standards traceable. The uncerin this tracing process is often added as part of the worst-case measurement error.

\,/

3B

ffiffiffitrffiffiHH

ffi

SOURCE

MATCH OF LEVELED OR RATIO

SYSTEMS

AND

COUPLER

VS.

2- and

3-RESISTOR SPLITTERS

COUPLERS:

The effective source reflection coefficient of a coupler-leveled or

4' ratio system has been shown'to be ru - rc

-

TD

Where

(11

I"

T

= output reflection coefficient of couplers main or through orrl

=

S22

: transmission coefficient of through arm

={16il6r transmittecl =

Sz, e.g.,

10-dB coupler

20-dB coupler

:

0.95

:0.995

D

=

"

= lsz

[isolalionJ

Sar fcoupling)

The above is a vector equation.

Since the techniques described do not derive phase information, the maxirnum source match can be calculated by adding terms of the equation assuming worst-case vector addition.

EXAMPLES:

HP

11692D Broadband 2 to

78-GHz coupler at

18 GHz is specified:

1.

1.4 main line

SWR

2.

2O-dB coupling

:

=

0.167

0.995 reflection coefficient transmission coefficient

3.

26-dB directivity

=

0.05

Maximum

Source

Match

= 0.167

+

0.995 x

0.05

=

0.21,7 [p)

=

1.55 swR

@ g

GHz specs are:

1.

1..3

SWR

2.

2O-dB coupling

3.

30-dB directivity

Maximum

Source

Match

@

B

GHz

:

0.131

+

0.995

1.39 SWR x

0.03 = 0.164

The above solutions match do represent absolute worst cases but are usually modified somewhat. The through line misof a coaxial coupler is largely due to the effects of both input and output connectors, Since the input connector is within the loop and its effects thus removed, some recommend including only

50 to

7O"h of the specified through line match.

POWER SPLITTERS:

Using similar flow-graph techniques the effective source match of a power splitter on either arm EQUALS z our

3out r.=s22-s21 sgz

*Ei o, szg

=s3-s3txg,

\-' or = output reflection coef - tracking x directivity

{equivalent to the equation for couplersJ.

7 Paul C. Ely,

Jr., "Swept

Frequency Techniques," Proc. of the

IEEE vol. 55

#6

June, 1967.

39

TWO.RESISTOR

SPLITTER:

For the

HP 11667A, a 2 resistor

-50

O configuration is employed

2 out

3 out

First, let's calculate

Sgg: the equivalent resistance in a 50 O reference s-parameter measurement is

R=50+-=83.330

50 + 100

R

- Zo

33.33

'33-R*Zo-133.33-'" szz

=

Sss

S21

=S31 =.5

\./

S23=S32=.25

APPLYING EQ

(2) rs = .zs

-

.s x'!= o

.5

Since all elements are resistive, phase angles are zero equivalent source match is specified as: and perfect cancellation occurs in the ideal case.

Actual

<1,20 SWR

<1.33 SWR at

B GHz at 18

GHz due mainly to connector imperfections.

THREE-RESISTOR

SPLITTER:

The conventional three-resistor splitter is configured:

\-l

At any port equivalent input

R = 10%

* J34;-!9:

50 o

Szz=Ssr=Srr:0

Srz

=

Szr

=

Srs

=

Sgr

:

Se:

=

S:s = 0.5

Again applying Equation

(21 : r"

-

0

- 0.5 x$ :

0.5

6.5

A

I. of

0.5 is an equivalent source SWR of

3:1.

A

3:1 source SWR will cause a 1 dB ripple when measuring a device with a 1,25 input

SWR. Thus, the three-resistor power splitter should never be used in leveling or ratio applications.

40

I

ffiPPtrffiffiEH

ffi

ERRORS

IN

REFLECTION MEASUREMENTS

Any reflection measuring system can be represented by:

POWER IN

POWER OUT f measured ( fM

)

F actual

( f4

)

REFLECTION MEASUBING TEST

\-.

A represents power leaked directly from the input terminal of the test set to its output, independent of the device under test.

Directivity in the coupler is the major contributor in a practical test set.

T represents the system scaling factor.

The coupling of the test set and the detector and display sensitivities are the major contributors to the scaling factor.

C represents the source match or the equivalent reflection looking back into the test coupler as developed in

Appendix

"A."

By application of

Mason's non-touching loop law, it can be shown:

(11

This is vector relationship, i.e., all beyond the scope of this note, we will

I's,

A, T, employ and

C are complex this formula only to quantities. relate

Since practical phase-measuring coupler and test to the maximum errors than can be expected by assuming worst-case vector addition of all terms.

set devices are parameters

::Maxerrof=E=l--Ie

I1

_ .r

_rl_l_eil_

_r

A

=

A*(1

+

TJ|A

[1

+

CIA*

C2I,r2

*

C3Ia3 ignoring the higher order terms, C2l,r2

*

*...J -1"

C3ln3

*

. .

.

:

A

*

(1+

TJIA

[1

+

Ct^)

-]tA

=A*T1-A+Cr^2+TCfee the term

161,,, is negligible since

T and C are both small

: A*Tl^+Cl^2

(2)

Since tometer

T is not known, both a calibration standard {short is employed, and circuit) and the unknown are measured when a reflecit is necessary to apply equation (2) to both conditions:

1."=1.+[A+Tln*Cr.t2J

=-1*[A-T+C] where f r

= -1 for short circuit

{standard)

Ir,*,.o.orr.*

= f.r t

(A +

Tl^ +

Cl'.\:l where l,t

= the actual reflection coefficient of the utlknown'

41

The measurement is performed quantities are measured by taking the ratio of luo.rxNowN and in return loss).

rsg for the difference in dB, if the two lors""et.o

-

IuN'!'gowN

-

-r""

-

[fo*

A*Tfe* cfo2]

IA t

A i

TIA

+

CfA2

1-[A-T+C]

11

+ [A-

T

+ c]

+

(A

-

T

+

CJ'

+...1

-

[lA

+ A

*

Tre+cr2l

[1

+

A-

T

+ c]

=I.a*A*TI,r*

CIe2

+AlrA+A+TrA+crA2I

(3,l

-Tlro+A+Tr^+cl^21

+cFA+A+TrA+CrA2l eliminating higher order terms

-

Ia * tA

+

(T

+

A

-

T

+

Clr.{

*

CIe2l

:

Ia *

[A

+ tA

+

CJ|A

+

CtA2] thus the uncertainty in rorspr-evro

-

Ie

=

AI

- A

+

[A+C)IA

*

Cf,r.2

=A*Bl^+Cl^e where B =

A

*

C

=

CALIBRATION

ERROR

By replacing

I with its scalar equivalent, p, it is possible to obtain equation

(5J on page 18:

Ap:A*Bp*Cpj

{4) t5l

(61

Discussion

The B term or calibration error is the error due flectometer to directivity

IAJ and source match (B) that occurs when the reis calibrated with the short circuit. This error may be removed for all practical purposes by the open

short calibration discussed on page

16.

Studying equation tion.

However, i

[

), it should be noted that the system scaling factor

T cancels out of the uncertainty equacaution is in order.

If instrumentation errors fdetector and displayJ cause

T to change as a function of amplitude, cancellation does not occur.

Also, ibration reference in a swept measurement, frequency response errors or other storage technique may be if

T varies with frequency and will a display graticule is enter into the B term. used

A as the grease utilized to store the exact calibration, eliminating frequency response errors.

calpencil

EXAMPLE:

B =

0 where calibration error has been removed with an open moved with a grease pencil.

- short calibration and frequency response re-

With a 77692D coupler at

8

GHz.

Minimum directivity =

26 dB =

A

0.164.

:

0.05 and from Appendix

"A" effective source reflection

= C

=

:.E=0.05+0.164IA2

Thus directivity errors dominate the measurement of low

SWR.devices while source match and tracking,terms overshadow the measurement of high reflections.

Instrument errors may be determined from appropriate technical specifications.

42

The mismatch graphs: ffiffiffitrffiffiBH ffi

ERRORS

IN TRANSMISSION MEASUREMENTS terms in a transmission measurement system can be represented by the following flow

I

I lr

CALIBRATION ps and pr represent the effective source and detector reflection coefficients, respectively.

p1 and p2 represent the input and output reflection coefficients of the DUT, respectively.

z1 and

12 are the forward and reverse transmission coefficients of the

DUT, respectively.

By the application urement of

Mason's non-touching loop of the forward transmission coefficiepl, rr, is: law, it can be shown the mismatch uncertainty in the meas-

Mismatch Uncertainty

= r, ' p"pt) r!1

[1

,- oto") t p:p"J

* fp.trtspo)

(11

The numerator is the calibration uncertainty represented by the first flow graph while the terms in the denominator are a result of the

DUT insertion. The term [p.rr.rp,r] may be ignored signs are if z1r2

)10 inserted so that the worst-case errors resulting from the vectorial additions dB.

All of the will be obtained.

terms in equation

L are complex quantities, i.e., they have both magnitude and phase. Because phase is not knowni plus and minus

The HP MISMATCH ERROR

LIMITS CALCULATOR number

1.

Consider a case where: is ideally suited for evaluating the terms in equation p"

=

0.33

ISWR

:

2J, typical for unleveled sources.

pn

= o.2 (SWR

:

1.5], typical for most diode detectors without pads.

Pt

P:

= 0.09 ISWR: 1.2)

T1 rz=

1O dB of loss.

43

Using the calculator, the calibration uncertainty can be determined by

MAX. MISMATCH

ERROR

LIMITS [dB) above and below 1.5 on the two

+0.56 dB and

-0.6 dB may be graphically represented: setting the

SWR1 to

2.0 and reading the

SWRz scales.

In this case, the limits are

+.56 dB

(1 l ps po)

\.1

In a similar fashion, the measurement terms may be evaluated.

The limits are

*0.26 dB while they are

*0.157 and

-0.16 for

(1

:b pzpo). Graphically this is represented as follows:

-0.268 dB for

(1 a p"pr),

+.157 dB

(1 + pSptl

11

+ pZpOl

Tct

Value in dB

A composite of the calibration and measurement uncertainties can now be obtained:

+.56 dB

+.157 dS

{+.417)

10.9S8 dB

\-/

Test value 10 dB

For a 10-dB loss measurement,

72 were less than 6 dB, the total mismatch uncertainty causes a-7.O77 dB and

+0.988 dB uncertainty.

*(pstrt:po] term would have added further uncertainty to the measurement.

If z1 and

It should be noted that these are the worst-case errors; typical mismatch errors are much less than worst-case, measurement fsee pad between

SWR po

Source match can be the improved

Appendix "A"J. The detector match may be improved detector and to a SWR between

1.1 and

1.3 either by leveling the sweeper or making a ratio the DUT; if the coupling to

SWR compensation techniques is approximately 1.3 or the mainline

SWR of the coupler. Reworking the previous example

=

7.2, the worst-case mismatch uncertainty is reduced to -0.14567 and +0.1473.

of

<1..2 are by introducing a used the effective with p"

=

10-dB detector

1.1 and

Transmission Uncertainty

=

[1

[1 t pspnJ t pspr) (7

! p2ei)

+

[Psz1r2Pp)

+

TRACKING

ERRORS t2)

Low

Loss

After the mismatch uncertainty has been evaluated, only tracking errors remain. The portion caused by differential frequency response is usually specified

However, most of this error can be eliminated, using a grease-pencil recording errors caused for measurement systems of the like of tracking errors the HP

8755 calibration trace. system. by variations in z are also specified for most detector/display systems like the HP 8755 system. There is no practical way to eliminate the tracking error caused by variation in z in a ratio measurement.

44

I i

DIGITAL

STORAGE AND

NORMALIZATION

The effectiveness and convenience techniques described of several measurement in this application note can be further increased with the addition of digital storage and normalization. The improvement contribute is achieved by removing several of the factors that to

CRT display inaccuracy. Much less human analysis of the display is required, significantly reducing operator time, effort, and associated potential errors.

System frequency response variations (dotted line above) are easily removed.

Digital storage is a method for obtaining flicker free displays at slow sweep speeds.

An even, clear, and continuous trace can always be produced regardless of sweep rate. Normalization removes

CRT image of system frequency response and provides for simplified comparison measurements. Grease lines the effects or visual interpolation are is an accurate graph not of needed, since amplitude pencil reference the resulting versus frequency.

Deviation between test devices displayed directly in dB with a single trace.

INPUT is displayed a constant trace regardless of sweep speed.

INPUT_MEMORY subtracG stored trace from current input.

HOLD button freezes display for convenient photographing.

Stores calibration or reference trace for normalized m€asuremsnts,

Compatible with several

HP instruments.

X.Y

PLOT current display from memory.

Simultaneous two channel operation.

Returns quickly to normal real-time mode,

Figure 63.

HP 8750A Storage-Normalizer.

Digital storage and

INPUT minus

MEM0RY provide flicker-free displays and trace noimalization.

The

Hewlett-Packard 8750A

Storage-Normalizer and some

CRT dated data are digitized and stored.

The screen at the sweep speed.

A separate memory is used and memory tem allows is of its main features are shown in

Figure

63.

is then refreshed at a flicker-free rate while the memory is upautomatically displayed, resulting two channels of information in to store a reference trace.

The difference between input a normalized display. The

8750's unique to be stored, normalized, and viewed simultaneously.

four memory sys-

The HP

8750A

Storage-Normalizer is also used with several

Hewlett-Packard

Spectrum

Analyzers and

Network Analyzers. For compatibility and other information, refer to the data sheets for the 8750A or the HP 87555

Frequency

Response Test

Set.

45

OTHER

LITERATURE ON HIGH FREQUENCY SWEPT MEASUREMENTS

787-2

CONFIGURATION OF

A

2-18

GHz

SYNTHESIZED FREQUENCY

SOURCE

USING

THE

8620C SWEEP

OSCILLATOR

Describes. a configuration source for using the

8620C sweeper, a calculatbr controlled

2-78

G}{z synthesized frequency

UHF Synthesizer, and Hewlett-Packard Interface Bus.

187.3

THREE

HP-IB CONFIGURATIONS

FOR

MAKING MICROWAVE SCALAR

MEASUREMENTS

This application note describes three HP-IB configured systems scalar transmission and impedance characteristics for measuring the of microwave components.

One employs the

HP 436A

Digital Power Meter, another the HP

8755

Frequency Response Test

Set, and the third the HP 84108

Network Analyzer. The specific hardware requirements are discussed and the relative merits of each approach compared.

187.4

CONFIGURATION

OF

A

TWO-TONE

SWEEPING GENERATOR

Describes a configuration foi' a source which will allow front ends, etc. The source outputs a local oscillator signal and a receiver signal whose offset from each other

{the

IF) is phase-locked,

IF stability of ing the

RF from

2 to

L8 GHz.

(t sweep

Hz is testing realizable of mixers, receiver even while sweep-

187.5

CALCULATOR CONTROL OF

THE

8620C

SWEEP OSCILLATOR

USING THE

HP-IB

Describes programmable control it sweeper with the HP capabilities

9820A, gB21.A, of the

8620C and the procedures required to and 9830A/B Desktop Computers via the

HP-IB.

It also contains sample programs which can effectively improve the CW accuracy of to approximateiy io.oos% of the bandwidth of ihe plug-in by using the ! / a"counter feedback scheme.

-

155-1 ACTIVE DEVICE

MEASUREMENTS

WITH

THE HP

8755 FREQUENCY

RESPONSE

TEST

SET

Describes measurements amplifier characterization using the

8755.

Four configurations are presented for of:

Swept-frequency gain and power output, CW frequency gain compression, swept frequency gain compression, and swept frequency harmonic content.

'5512

1OO dB

DYNAMIC

RANGE

MEASUREMENTS USING

FREQUENCY RESPONSE

THE HP

8755

TEST

SET

Describes measurement transmission measurements. configurations for making 100 dB dynamic range swept

The note includes the. theory of operation, equipment limitations and accuracy considerations.

22I

SEMI-AUTOMATIC MEASUREMENTS USING

THE

B41OB

MICROWAVE NETWORK ANALYZER

AND THE

9825A

DESKTOP COMPUTER

Describes methods the configuration ments; fundamentals of a semi-automatic

Packard Interface Bus (HP-IB). Topics treated include: network block diagram of suggested equipment; of digitizing magnitude and phase readings; sources of error of one-port vector error correction; a analyzer using in microwave measuresample program the for

Hewlettthe

98254

Desktop Computer; and typical results and operating procedures.

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

Download PDF

advertisement