Chapter 1: Data Converter History

Chapter 1: Data Converter History
DATA CONVERTER HISTORY
ANALOG-DIGITAL CONVERSION
1. Data Converter History
1.1 Early History
1.2 Data Converters of the 1950s and 1960s
1.3 Data Converters of the 1970s
1.4 Data Converters of the 1980s
1.5 Data Converters of the 1990s
1.6 Data Converters of the 2000s
2. Fundamentals of Sampled Data Systems
3. Data Converter Architectures
4. Data Converter Process Technology
5. Testing Data Converters
6. Interfacing to Data Converters
7. Data Converter Support Circuits
8. Data Converter Applications
9. Hardware Design Techniques
I. Index
ANALOG-DIGITAL CONVERSION
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
CHAPTER 1
DATA CONVERTER HISTORY
Walt Kester
Chapter Preface
This chapter was inspired by Walt Jung's treatment of op amp history in the first chapter
of his book, Op Amp Applications (Reference 1). His writing on the subject contains
references to hundreds of interesting articles, patents, etc., which taken as a whole, paints
a fascinating picture of the development of the operational amplifier—from Harold
Black's early feedback amplifier sketch to modern high performance IC op amps.
We have attempted to do the same for the history of data converters. In considering the
scope of this effort—and the somewhat chaotic and fragmented development of data
converters—we were faced with a difficult challenge in organizing the material. Rather
than putting all the historical material in this single chapter, we have chosen to disperse
some of it throughout the book. For instance, most of the historical material related to
data converter architectures is included in Chapter 3 (Data Converter Architectures)
along with the individual converter architectural descriptions. Likewise, Chapter 4 (Data
Converter Process Technology) includes most of the key events related to data converter
process technology. Chapter 5 (Testing Data Converters) touches on some of the key
historical developments relating to data converter testing.
In an effort to make each chapter of this book stand on its own as much as possible, some
of the historical material is repeated in several places—therefore, the reader should
realize that this repetition is intentional and not the result of careless editing.
SECTION 1.1: EARLY HISTORY
It is difficult to determine exactly when the first data converter was made or what form it
took. The earliest recorded binary DAC known to the authors of this book is not
electronic at all, but hydraulic. Turkey, under the Ottoman Empire, had problems with its
public water supply, and sophisticated systems were built to meter water. One of these is
shown in Figure 1.1 and dates to the 18th Century. An example of an actual dam using
this metering system was the Mahmud II dam built in the early 19th century near Istambul
and described in Reference 2.
The metering system used reservoirs (labeled header tank in the diagrams) maintained at
a constant depth (corresponding to the reference potential) by means of a spillway over
which water just trickled (the criterion was sufficient flow to float a straw). This is
illustrated in Figure 1.1A. The water output from the header tank is controlled by gated
binary-weighted nozzles submerged 96 mm below the surface of the water. The output of
the nozzles feeds an output trough as shown in Figure 1.1B. The nozzle sizes
corresponded to flows of binary multiples and sub-multiples of the basic unit of
1 lüle (= 36 l/min or 52 m3/day). An eight-lüle nozzle was known as a "sekizli lüle," a
1.1
ANALOG-DIGITAL CONVERSION
four lüle nozzle a "dörtlü lüle," a ¼ lüle nozzle a "kamuş," an eighth lüle a "masura," and
a thirty-second lüle a "çuvaldiz." Details of the metering system using the binary
weighted nozzles are shown in Figure 1.1C. This is functionally an 8-bit DAC with
manual (rather than digital, no doubt) input and a wet output, and it may be the oldest
DAC in the world. There are probably other examples of early data converters, but we
will now turn our attention to those based on more familiar electronic techniques.
NOZZLES
96mm
WATER
LEVEL
SPILLWAY
96mm
HEADER
TANK
HEADER
TANK
WATER
INPUT
FROM
DAM
OUTPUT
TROUGH
NOZZLES
(A) HEADER SYSTEM: Note-The spillway
and the nozzles need different outlets
(C) TOP VIEW OF
METERING SYSTEM
DETAILS SHOWING
BINARY WEIGHTED
NOZZLES
Adapted from:
Kâzim Çeçen, "Sinan's Water Supply
System in Istambul," Istambul Technical University /
Istambul Water and Sewage Administration,
Istambul Turkey, 1992-1993, pp. 165-167.
(B) SECTIONAL VIEW OF METERING SYSTEM
8 BINARY
WEIGHTED
NOZZLES
HEADER
TANK
WATER
INPUT
FROM
DAM
OUTPUT
TROUGH
SPILLWAY
Figure 1.1: Early 18th Century Binary Weighted
Water Metering System
Probably the single largest driving force behind the development of electronic data
converters over the years has been the field of communications. The telegraph led to the
invention of the telephone, and the subsequent formation of the Bell System. The
proliferation of the telegraph and telephone, and the rapid demand for more capacity, led
to the need for multiplexing more than one channel onto a single pair of copper
conductors. While time division multiplexing (TDM) achieved some measure of
popularity, frequency division multiplexing (FDM) using various carrier-based systems
was by far the most successful and widely used. It was pulse code modulation (PCM),
however, that put data converters on the map, and understanding its evolution is where
we begin.
The material in the following sections has been extracted from a number of sources, but
K. W. Cattermole's classic 1969 book, Principles of Pulse Code Modulation (Reference
3), is by far the most outstanding source of historical material for both PCM and data
converters. In addition to the historical material, the book has excellent tutorials on
sampling theory, data converter architectures, and many other topics relating to the
subject. An extensive bibliography cites the important publications and patents behind the
major developments. In addition to Cattermole's book, the reader is also referred to an
1.2
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
excellent series of books published by the Bell System under the title of A History of
Engineering and Science in the Bell System (References 4 through 8). These Bell System
books are also excellent sources for background material on the entire field of
communications.
The Early Years: Telegraph to Telephone
According to Cattermole (Reference 3), the earliest proposals for the electric telegraph
date from about 1753, but most actual development occurred from about 1825-1875.
Various ideas for binary and ternary numbers, codes of length varying inversely with
probability of occurrence (Schilling, 1825), reflected-binary (Elisha Gray, 1878—now
referred to as the Gray code), and chain codes (Baudot, 1882) were explored. With the
expansion of telegraphy came the need for more capacity, and multiplexing more than
one signal on a single pair of conductors. Figure 1.2 shows a typical telegraph key and
some highlights of telegraph history.
Telegraph proposals: Started 1753
Major telegraph development: 1825 - 1875
Various binary codes developed
Experiments in multiplexing for increased channel
capacity
Telephone invented: 1875 by A. G. Bell while
working on a telegraph multiplexing project
Evolution:
Telegraph: Digital
Telephone: Analog
Frequency division multiplexing (FDM): Analog
Pulse code modulation (PCM): Back to Digital
Figure 1.2: The Telegraph
The invention of the telephone in 1875 by Alexander Graham Bell (see References 9 and
10) was probably the most significant event in the entire history of communications. It is
interesting to note, however, that Bell was actually experimenting with a telegraph
multiplexing system (Bell called it the harmonic telegraph) when he recognized the
possibility of transmitting the voice itself as an analog signal.
Figure 1.3 shows a diagram from Bell's original patent which puts forth his basic
proposal for the telephone. Sound vibrations applied to the transmitter A cause the
membrane a to vibrate. The vibration of a causes a vibration in the armature c which
induces a current in the wire e via the electromagnet b. The current in e produces a
corresponding fluctuation in the magnetic field of electromagnet f, thereby vibrating the
receiver membrane i.
1.3
ANALOG-DIGITAL CONVERSION
Extracted from U.S. Patent 174,465,
Filed February 14, 1876, Issued March 7, 1876
Figure 1.3: The Telephone
The proliferation of the telephone generated a huge need to increase channel capacity by
multiplexing. It is interesting to note that studies of multiplexing with respect to
telegraphy led to the beginnings of information theory. Time division multiplexing
(TDM) for telegraph was conceived as early as 1853 by a little known American
inventor, M. B. Farmer; and J. M. E. Baudot put it into practice in 1875 using rotating
mechanical commutators as multiplexers.
In a 1903 patent (Reference 11), Willard M. Miner describes experiments using this type
of electromechanical rotating commutator to multiplex several analog telephone
conversations onto a single pair of wires as shown in Figure 1.4. Quoting from his patent,
he determined that each channel must be sampled at
"… a frequency or rapidity approximating the frequency or average frequency of the
finer or more complex vibrations which are characteristic of the voice or of articulate
speech, …, as high as 4320 closures per second, at which rate I find that the voice with
all its original timbre and individuality may be successfully reproduced in the receiving
instrument. … I have also succeeded in getting what might be considered as commercial
results by using rates of closure that, comparatively speaking, are as low as 3500
closures per second, this being practically the rate of the highest note which
characterizes vowel sounds."
At higher sampling rates, Miner found no perceptible improvement in speech quality,
probably because of other artifacts and errors in his rather crude system.
1.4
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
There was no follow-up to Miner's work on sampling and TDM, probably because there
were no adequate electrical components available to make it practical. FDM was well
established by the time adequate components did arrive.
Extracted from: Williard M. Miner, "Multiplex Telephony," U.S. Patent 745,734,
Filed February 26, 1903, Issued December 1, 1903
Figure 1.4: One of the Earliest References to a Criteria for
Determining the Sampling Rate
The Invention of PCM
Pulse code modulation was first disclosed in a relatively obscure patent issued to Paul M.
Rainey of Western Electric in 1921 (Reference 12). The patent describes a method to
transmit facsimile information in coded form over a telegraph line using 5-bit PCM. The
figure from the patent is shown in Figure 1.5 (additional labels have been added for
clarity).
Rainey proposed that a light beam be focused on the transparency of the material to be
transmitted. A photocell is placed on the other side of the transparency to gather the light
and produce a current proportional to the intensity of the light. This current drives a
galvanometer which in turn moves another beam of light which activates one of 32
individual photocells, depending upon the amount of galvanometer deflection. Each
individual photocell output activates a corresponding relay. The five relay outputs are
connected in such a way as to generate the appropriate code corresponding to the
photocell location. The digital code is thus generated from an "m-hot out of 32 code" ,
similar to modern flash converters. The output of this simple electo-optical-mechanical
"flash" converter is then transmitted serially using a rotating electromechanical
commutator, called a distributor.
1.5
ANALOG-DIGITAL CONVERSION
RECEIVER
TRANSMITTER
5-bit "Flash"
Converter
Transparency
(negative)
Light
sensitive
plate
Photocell Bank (32)
Deflected
light beam
5-bit D/A Converter
Figure 1.5: The First Disclosure of PCM: Paul M. Rainey,
"Facimile Telegraph System," U.S. Patent 1,608,527,
Filed July 20, 1921,Issued November 30, 1926
The serial data is transmitted, received, and converted into a parallel format using a
second distributor and a bank of relays. The received code determines the combination of
relays to be activated, and the relay outputs are connected across appropriate taps of a
resistor which is in series with the receiving lamp. The current through the receiving
lamp therefore changes depending upon the received code, thereby varying its intensity
proportionally to the received code and performing the digital-to-analog conversion. The
receiving lamp output is focused on a photographically sensitive receiving plate, thereby
reproducing the original image in quantized form.
Rainey's patent illustrates several important concepts: quantization using a flash A/D
converter, serial data transmission, and reconstruction of the quantized data using a D/A
converter. These are the fundamentals of PCM. However, his invention aroused little
interest at the time and was, in fact, forgotten by Bell System engineers. His patent was
discovered years later after many other PCM patents had already been issued.
The Mathematical Foundations of PCM
In the mid-1920's, Harry Nyquist studied telegraph signaling with the objective of finding
the maximum signaling rate that could be used over a channel with a given bandwidth.
His results are summarized in two classic papers published in 1924 (Reference 13) and
1928 (Reference 14), respectively.
1.6
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
In his model of the telegraph system, he defined his signal as:
s( t ) = ∑ a k f ( t − kT).
k
Eq. 1.1
In the equation, f(t) is the basic pulse shape, ak is the amplitude of the kth pulse, and T is
the time between pulses. DC telegraphy fits this model if f(t) is assumed to be a
rectangular pulse of duration T, and ak equal to 0 or 1. A simple model is shown in
Figure 1.6. The signal is bandlimited to a frequency W by the transmission channel.
SAMPLER
TELEGRAPH
PULSES
CHANNEL
BANDWIDTH = W
T
sin x
x
T
T ≥ 1
2W
fs = 1
T
≥2W
t
Up to 2W pulses per second can be transmitted over a channel which has a bandwidth W.
If a signal is sampled instantaneously at regular intervals at a rate at least twice the
highest significant signal frequency, then the samples contain all the information in the
original signal.
Figure 1.6: Harry Nyquist's Classic Theorem: 1924
His conclusion was that the pulse rate, 1/T, could not be increased beyond 2W pulses per
second. Another way of stating this conclusion is if a signal is sampled instantaneously at
regular intervals at a rate at least twice the highest significant signal frequency, then the
samples contain all the information in the original signal. This is clear from Figure 1.6 if
the filtered rectangular pulses are each represented by a sinx/x response. The sinx/x time
domain impulse response of an ideal lowpass filter of bandwidth W has zeros at intervals
of 1/2W. Therefore, if the output waveform is sampled at the points indicated in the
diagram, there will be no interference from adjacent pulses, provided T ≥ 1/2W (or more
commonly expressed as: fs ≥ 2W), and the amplitude of the individual pulses can be
uniquely recovered.
Except for a somewhat general article by Hartley in 1928 (Reference 15), there were no
significant additional publications on the specifics of sampling until 1948 in the classic
papers by Shannon, Bennett, and Oliver (References 16-19) which solidified PCM theory
for all time. A summary of the classic papers on PCM is shown in Figure 1.7.
1.7
ANALOG-DIGITAL CONVERSION
Multiplexing experiments such as Williard Miner, "Multiplex Telephony," U.S.
Patent 745,734, filed February 26, 1903, issued December 1, 1903.
H. Nyquist, "Certain Factors Affecting Telegraph Speed," Bell System Technical
Journal, Vol. 3, April 1924, pp. 324-346.
H. Nyquist, Certain Topics in Telegraph Transmission Theory, A.I.E.E.
Transactions, Vol. 47, April 1928, pp. 617-644.
R.V.L. Hartley, "Transmission of Information," Bell System Technical Journal, Vol.
7, July 1928, pp. 535-563,.
Note: Shannon's classic paper was written in 1948, well after the invention of PCM:
C. E. Shannon, "A Mathematical Theory of Communication," Bell System
Technical Journal, Vol. 27, July 1948, pp. 379-423, and October 1948, pp. 623-656.
W. R. Bennett, "Spectra of Quantized Signals," Bell System Technical Journal, Vol
27, July 1948, pp. 446-471.
B. M. Oliver, J. R. Pierce, C. E. Shannon, "The Philosophy of PCM," IRE
Proceedings, Vol. 36, November 1948, pp. 1324-1331.
W. R. Bennett, "Noise in PCM Systems," Bell Labs Record, Vol. 26, December
1948, pp. 495-499
Figure 1.7: Mathematical Basis of PCM
The PCM Patents of Alec Harley Reeves
By 1937, frequency division multiplexing (FDM) based on vacuum tube technology was
widely used in the telephone industry for long-haul routes. However, noise and distortion
were the limiting factors in expanding the capacity of these systems. Although wider
bandwidths were becoming available on microwave links, the additional noise and
distortion made them difficult to adapt to FDM signals.
Alec Harley Reeves had studied analog-to-time conversion techniques using pulse time
modulation (PTM) during the beginning of his career in the 1920's. In fact, he was one of
the first to make use of counter chains to accurately define time bases using bistable
multivibrators invented by Eccles and Jordan a few years earlier. In PTM, the amplitude
of the pulses is constant, and the analog information is contained in the relative timing of
the pulses. This technique gave better noise immunity than strictly analog transmission,
but Reeves was shortly to invent a system that would completely revolutionize
communications from that point forward.
It was therefore the need for a system with the noise immunity similar to the telegraph
system that led to the (re-) invention of pulse code modulation (PCM) by Reeves at the
Paris labs of the International Telephone and Telegraph Corporation in 1937. The very
first PCM patent by Reeves was filed in France, but was immediately followed by similar
patents in Britain and the United States, all listing Reeves as the inventor (Reference 20).
These patents were very comprehensive and covered the far-reaching topics of (1)
general principles of quantization and encoding, (2) the choice of resolution to suit the
noise and bandwidth of the transmission medium, (3) transmission of signals in digital
format serially, in parallel, and as modulated carriers, and (4) a counter-based design for
1.8
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
the required 5-bit ADCs and DACs. Unlike the previous PCM patent by Rainey in 1926,
Reeves took full advantage of existing vacuum tube technology in his design.
The ADC and DAC developed by Reeves deserves some further discussion, since they
represent one of the first all-electronic data converters on record. The ADC technique
(see Figure 1.8) basically uses a sampling pulse to take a sample of the analog signal, set
an R/S flip-flop, and simultaneously start a controlled ramp voltage. The ramp voltage is
compared with the input, and when they are equal, a pulse is generated that resets the R/S
flip-flop. The output of the flip-flop is a pulse whose width is proportional to the analog
signal at the sampling instant. This pulse width modulated (PWM) pulse controls a gated
oscillator, and the number of pulses out of the gated oscillator represents the quantized
value of the analog signal. This pulse train can be easily converted to a binary word by
driving a counter. In Reeves' system, a master clock of 600 kHz is used, and a 100:1
divider generates the 6-kHz sampling pulses. The system uses a 5-bit counter, and 31
counts (out of the 100 counts between sampling pulses) therefore represents a fullscale
signal.
VOICE
INPUT
PULSE WIDTH
MODULATOR
SAMPLING
PULSE
6 kSPS
5-BIT
CK COUNTER
CLOCK
600 kHz
÷ 100
RESET
DATA
OUTPUT
READ-OUT PULSE
6 kSPS
Adapted from: Alec Harley Reeves, "Electric Signaling System,"
U.S. Patent 2,272,070, Filed November 22, 1939, Issued February 3, 1942
Figure 1.8: A. H. Reeves' 5-bit Counting ADC
The DAC uses a similar counter and clock source as shown in Figure 1.9. The received
binary code is first loaded into the counter, and the R/S flip-flop is reset. The counter is
then allowed to count upward by applying the clock pulses. When the counter overflows
and reaches 00000, the clock source is disconnected, and the R/S flip-flop is set. The
number of pulses counted by the encoding counter is thus the complement of the
incoming data word. The output of the R/S flip-flop is a PWM signal whose analog value
is the complement of the input binary word. Reeves uses a simple lowpass filter to
recover the analog signal from the PWM output. The phase inversion in the DAC is
easily corrected in either the logic or in an amplifier further down the signal chain.
1.9
ANALOG-DIGITAL CONVERSION
DATA INPUT
Adapted from: Alec Harley Reeves,
"Electric Signaling System,"
U.S. Patent 2,272,070,
Filed November 22, 1939,
Issued February 3, 1942
PWM
LOAD
6 kHz
CK
CLOCK
600kHz
5-BIT
COUNTER
VOICE
OUTPUT
R
FF
LPF
S
÷ 100
Figure 1.9: A. H. Reeves' 5-bit Counting DAC
Reeves' patents covered all the essentials of PCM: sampling, quantizing, and coding the
digitized samples for serial, parallel, phase-modulated, and other transmission methods.
On the receiving end, Reeves proposed a suitable decoder to reconstruct the original
analog signal. In spite of the significance of his work, it is interesting to note that after the
patent disclosures, Reeves shifted his attention to the shortwave transmission of speech
using pulse-amplitude modulation, pulse-duration modulation, and pulse-position
modulation, rather than pursuing PCM techniques.
PCM and the Bell System: World War II through 1948
Under a cross-licensing arrangement with International Telephone and Telegraph
Corporation, Bell Telephone Laboratories' engineers reviewed Reeves' circuit
descriptions and embarked upon their own pursuit of PCM technology. Starting in about
1940 and during World War II, studies were conducted on a speech secrecy system that
made PCM techniques mandatory.
The highly secret "Project-X" to develop a speech secrecy system was started in 1940 by
Bell Labs and is described in detail in Reference 6 (pp. 296-317). It used a complex
technique based on vacuum tube technology that made use of the previously developed
"vocoder," PCM techniques, and a unique data scrambling technique utilizing a
phonograph recording containing the electronic "key" to the code. This system was
designed at Bell Labs and put into production by Western Electric in late 1942. By April,
1943, several terminals were completed and installed in Washington, London, and North
Africa. Shortly thereafter, additional terminals were installed in Paris, Hawaii, Australia,
and the Philippines.
1.10
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
By the end of the war, several groups at Bell Labs were studying PCM; however most of
the wartime results were not published until several years later because of secrecy issues.
The work of H. S. Black, J. O. Edson, and W. M. Goodall were published in 1947-1948.
(References 21, 22, and 23). Their emphasis was on speech encryption systems based on
PCM techniques, and many significant developments came out of their work. A PCM
system which digitized the entire voice band to 5-bits, sampling at 8 kSPS using a
successive approximation ADC was described by Edson and Black (Reference 21 and
22). W. M. Goodall described an experimental PCM system in his classic paper based on
similar techniques (Reference 23).
Some of the significant developments which came out of this work were the successive
approximation ADC, the electron beam coding tube, the Shannon-Rack decoder, the
logarithmic spacing of quantization levels (companding), and the practical
demonstrations that PCM was feasible. The results were nicely summarized in a 1948
article by L. A. Meacham and E. Peterson describing an experimental 24-channel PCM
system (Reference 24). A summary of PCM work done at Bell Labs through 1948 is
shown in Figure 1.10.
"Project-X" voice secrecy system using PCM, 1940-1943.
5-bit, 8kSPS successive approximation ADC
Logarithmic quantization of speech (companding)
Electron beam coding tube, 7-bit, 100kSPS
"Shannon-Rack" decoder (DAC)
Successful demonstration of experimental PCM terminals
Theoretical PCM work expanded and published by Shannon
Germanium transistor invented: 1947
Figure 1.10: Bell Laboratories' PCM Work: World War II through 1948.
A significant development in ADC technology during the period was the electron beam
coding tube shown in Figure 1.11. The tube described by R. W. Sears in Reference 25
was capable of sampling at 96 kSPS with 7-bit resolution. The basic electron beam coder
concepts are shown in Figure 1.11 for a 4-bit device. The early tubes operated in the
serial mode (Figure 1.11A). The analog signal is first passed through a sample-and-hold,
and during the "hold" interval, the beam is swept horizontally across the tube. The Ydeflection for a single sweep therefore corresponds to the value of the analog signal from
the sample-and-hold. The shadow mask is coded to produce the proper binary code,
depending on the vertical deflection. The code is registered by the collector, and the bits
are generated in serial format. Later tubes used a fan-shaped beam (shown in Figure
1.11B), creating the first electronic "flash" converter delivering a parallel output word.
1.11
ANALOG-DIGITAL CONVERSION
X Deflectors
(A) SERIAL MODE
Y Deflectors
Shadow Mask
Collector
Electron gun
(B) PARALLEL MODE
Y Deflectors
Shadow Mask
Collector
Electron gun
Figure 1.11: The Electron Beam Coder
Early electron tube coders used a binary-coded shadow mask, and large errors could
occur if the beam straddled two adjacent codes and illuminated both of them. The way
these errors occur is illustrated in Figure 1.12A, where the horizontal line represents the
beam sweep at the midscale transition point (transition between code 0111 and code
1000). For example, an error in the most significant bit (MSB) produces an error of ½
scale. These errors were minimized by placing fine horizontal sensing wires across the
boundaries of each of the quantization levels. If the beam initially fell on one of the
wires, a small voltage was added to the vertical deflection voltage which moved the beam
away from the transition region.
(A) 4-BIT BINARY CODE
SHADOW MASK
(B) 4-BIT REFLECTED-BINARY CODE
(GRAY CODE)
1111
1110
1101
1100
1011
1010
1001
1000
0111
0110
0101
0100
0011
0010
0001
0000
MSB
LSB
1000
1001
1011
1010
1110
1111
1101
1100
0100
0101
0111
0110
0010
0011
0001
0000
SHADOW MASK
MSB
Figure 1.12: Electron Beam Coder Shadow Masks
for Binary and Gray Code
1.12
LSB
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
The errors associated with binary shadow masks were eliminated by using a Gray code
shadow mask as shown in Figure 1.12B. This code was originally called the "reflected
binary" code, and was invented by Elisha Gray in 1878, and later re-invented by Frank
Gray in 1949 (see Reference 26). The Gray code has the property that adjacent levels
differ by only one digit in the corresponding Gray-coded word. Therefore, if there is an
error in a bit decision for a particular level, the corresponding error after conversion to
binary code is only one least significant bit (LSB). In the case of midscale, note that only
the MSB changes. It is interesting to note that this same phenomenon can occur in
modern comparator-based flash converters due to comparator metastability. With small
overdrive, there is a finite probability that the output of a comparator will generate the
wrong decision in its latched output, producing the same effect if straight binary decoding
techniques are used. In many cases, Gray code, or "pseudo-Gray" codes are used to
decode the comparator bank output before finally converting to a binary code output
(refer to Chapter 3 for further architectural descriptions).
In spite of the many mechanical and electrical problems relating to beam alignment,
electron tube coding technology reached its peak in the mid-l960s with an experimental
9-bit coder capable of 12-MSPS sampling rates (Reference 27). Shortly thereafter,
however, advances in solid-state ADC techniques quickly made the electron tube
converter technology obsolete.
Op Amps and Regenerative Repeaters: Vacuum Tubes to Solid-State
Except for early relatively inefficient electro-mechanical amplifiers (see Reference 5),
electronic amplifier development started with the invention of the vacuum tube by Lee de
Forest in 1906 (References 28 and 29). A figure from the original de Forest patent is
shown in Figure 1.13.
Extracted from: Lee De Forest, "Device for Amplifying Feeble Electrical
Currents," U.S. Patent 841,387, Filed October 25, 1906, Issued January 15, 1907
Figure 1.13: The Invention of the Vacuum Tube: 1906
1.13
ANALOG-DIGITAL CONVERSION
By 1914, vacuum tube amplifiers had been introduced into the telephone plant. Amplifier
development has always been critical to data converter development, starting with these
early vacuum tube circuits. Key to the technology was the invention of the feedback
amplifier by Harold S. Black in 1927 (References 30, 31, and 32). Amplifier circuit
development continued throughout World War II, and many significant contributions
came from Bell Labs. (The complete history of op amps is given in Reference 1). Figure
1.14 shows a drawing from a later article published by Black defining the feedback
amplifier.
INPUT
OUTPUT
+
µ
-
β
OUTPUT = µ INPUT – β•OUTPUT
1
OUTPUT
=
INPUT
β
1
1+
1
µβ
Harold S. Black, "Stabilized Feedback Amplifiers,"
Bell System Technical Journal, Vol. 13, No. 1, January 1934
Figure 1.14: Harold Black's Feedback Amplifier of 1927
The invention of the germanium transistor in 1947 (References 33, 34, and 35) was key
to the development of PCM and all other electronic systems. In order for PCM to be
practical, regenerative repeaters had to be placed periodically along the transmission
lines. Vacuum tube repeaters had been somewhat successfully designed and used in the
telegraph and voice network for a number of years prior to the development of the
transistor, but suffered from obvious reliability problems. However, the solid state
regenerative repeater designed by L. R. Wrathall in 1956 brought the PCM research
phase to a dramatic conclusion (Reference 36). This repeater was demonstrated on an
experimental cable system using repeater spacings of 2.3 miles on 19-gage cable, and
0.56 miles on 32-gauge cable. A schematic diagram of the repeater is reproduced in
Figure 1.15.
The Wrathall repeater used germanium transistors designed by Bell Labs and built by
Western Electric. The silicon transistor was invented in 1954 by Gordon Teal at Texas
Instruments and gained wide commercial acceptance because of the increased
temperature performance and reliability. Finally, the invention of the integrated circuit
(References 37 and 38) in 1958 followed by the planar process in 1959 (Reference 39) set
the stage for future PCM developments. These key solid state developments are
summarized in Figure 1.16 and discussed in greater depth in Chapter 4 of this book.
1.14
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
From: L.R. Wrathall, "Transistorized Binary Pulse Regenerator,"
Bell System Technical Journal, Vol. 35, September 1956, pp. 1059-1084
Figure 1.15: L. R. Wrathall's Solid State PCM Repeater: 1956
Invention of the (Germanium) transistor at Bell Labs: John
Bardeen, Walter Brattain, and William Shockley in 1947.
Silicon Transistor: Gordon Teal, Texas Instruments, 1954.
Birth of the Integrated Circuit:
Jack Kilby, Texas Instruments, 1958 (used bond wires for
interconnections).
Robert Noyce, Fairchild Semiconductor, 1959 (used
metallization for interconnections).
The Planar Process: Jean Hoerni, Fairchild Semiconductor, 1959.
Figure 1.16: Key Solid-state Developments: 1947-1959
With the development of the Wrathall repeater, it was therefore clear in 1956 that PCM
could be effectively used to increase the number of voice channels available on existing
copper cable pairs. This was especially attractive in metropolitan areas where many cable
conduits were filled to capacity. Many of these pairs were equipped with loading coils at
a spacing of 1.8 kM to improve their response in the voice band. It was natural to
consider replacing the loading coils with solid-state repeaters and to extend the capacity
from 1 to 24 channels by using PCM.
1.15
ANALOG-DIGITAL CONVERSION
For these reasons, a decision was made at Bell Labs to develop a PCM carrier system,
and a prototype 24-channel system was designed and tested during 1958 and 1959 on a
link between Summit, New Jersey and South Orange, New Jersey. This system, called the
T-1 carrier system, transmitted 24 voice channels using a 1.544-MHz pulse train in a
bipolar code. The system used 7-bit logarithmic encoding with 26-dB of companding,
and was later expanded to 8-bit encoding. The solid-state repeaters were spaced at
1.8-kM intervals, corresponding to the placement of the existing loading coils. The first
T-1 operating link went into service in 1962, and by 1984 there were more than 200
million circuit-kilometers of T-1 carrier in the United States.
1.16
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
REFERENCES:
1.1 EARLY HISTORY
1.
Walter G. Jung, Op Amp Applications, Analog Devices, 2002, ISBN 0-916550-26-5.
2.
Kâzim Çeçen, "Sinan's Water Supply System in Istambul," Istambul Technical University / Istambul
Water and Sewage Administration, Istambul Turkey, 1992-1993, pp. 165-167.
3.
K. W. Cattermole, Principles of Pulse Code Modulation, American Elsevier Publishing Company,
Inc., 1969, New York NY, ISBN 444-19747-8. (An excellent tutorial and historical discussion of data
conversion theory and practice, oriented towards PCM, but covers practically all aspects. This one is
a must for anyone serious about data conversion!).
4.
Editors, Transmission Systems for Communications, Bell Telephone Laboratories, 1964. (excellent
discussion of Bell System transmission systems from a technical standpoint).
5.
M. D. Fagen, A History of Engineering and Science in the Bell System: The Early Years (18751925), Bell Telephone Laboratories, 1975.
6.
M. D. Fagen, A History of Engineering and Science in the Bell System: National Service in War
and Peace (1925-1975), Bell Telephone Laboratories, 1978, ISBN 0-932764-00-2.
7.
S. Millman, A History of Engineering and Science in the Bell System: Communications Sciences
(1925-1980), AT&T Bell Laboratories, 1984, ISBN 0-932764-06-1.
8.
E. F. O'Neill, A History of Engineering and Science in the Bell System: Transmission Technology
(1925-1975), AT&T Bell Telephone Laboratories, 1985.
9.
Alexander Graham Bell, "Improvement in Telegraphy," U.S. Patent 174,465, filed February 14, 1876,
issued March 7, 1876. (this is the original classic patent on the telephone).
10. Alexander Graham Bell, "Improvement in Electric Telegraphy," U.S. Patent 186,787, filed January
15, 1877, issued January 30, 1877. (this and the preceding patent formed the basis of the Bell System
patents).
11. Willard M. Miner, "Multiplex Telephony," U.S. Patent 745,734, filed February 26, 1903, issued
December 1, 1903. (a relatively obscure patent on electro-mechanical multiplexing of telephone
channels in which experiments describing voice quality versus sampling frequency are mentioned).
12. Paul M. Rainey, "Facimile Telegraph System," U.S. Patent 1,608,527, filed July 20, 1921, issued
November 30, 1926. (although A. H. Reeves is generally credited with the invention of PCM, this
patent discloses an electro-mechanical PCM system complete with A/D and D/A converters. The patent
was largely ignored and forgotten until many years after the various Reeves' patents were issued in
1939-1942).
13. H. Nyquist, "Certain Factors Affecting Telegraph Speed," Bell System Technical Journal, Vol. 3,
April 1924, pp. 324-346.
14. H. Nyquist, "Certain Topics in Telegraph Transmission Theory," A.I.E.E. Transactions, Vol. 47,
April 1928, pp. 617-644.
15. R.V.L. Hartley, "Transmission of Information," Bell System Technical Journal, Vol. 7, July 1928,
pp. 535-563.
16. C. E. Shannon, "A Mathematical Theory of Communication," Bell System Technical Journal, Vol.
27, July 1948, pp. 379-423 and October 1948, pp. 623-656.
1.17
ANALOG-DIGITAL CONVERSION
17. W. R. Bennett, "Spectra of Quantized Signals," Bell System Technical Journal, Vol. 27, July 1948,
pp. 446-471.
18. B. M. Oliver, J. R. Pierce, and C. E. Shannon, "The Philosophy of PCM," Proceedings IRE, Vol. 36,
November 1948, pp. 1324-1331.
19. W. R. Bennett, "Noise in PCM Systems," Bell Labs Record, Vol. 26, December 1948, pp. 495-499.
20. Alec Harley Reeves, "Electric Signaling System," U.S. Patent 2,272,070, filed November 22, 1939,
issued February 3, 1942. Also French Patent 852,183 issued 1938, and British Patent 538,860 issued
1939.
21. H. S. Black and J. O. Edson, "Pulse Code Modulation," AIEE Transactions, Vol. 66, 1947, pp. 895899.
22. H. S. Black, "Pulse Code Modulation," Bell Labs Record, Vol. 25, July 1947, pp. 265-269.
23. W. M. Goodall, "Telephony by Pulse Code Modulation," Bell System Technical Journal, Vol. 26, pp.
395-409, July 1947. (describes an experimental PCM system using a 5-bit, 8-kSPS successive
approximation ADC based on the subtraction of binary weighted charges from a capacitor to
implement the internal DAC function).
24. L. A. Meacham and E. Peterson, "An Experimental Multichannel Pulse Code Modulation System of
Toll Quality," Bell System Technical Journal, Vol 27, No. 1, January 1948, pp. 1-43. (describes the
culmination of much work leading to this 24-channel experimental PCM system).
25. R. W. Sears, "Electron Beam Deflection Tube for Pulse Code Modulation," Bell System Technical
Journal, Vol. 27, pp. 44-57, Jan. 1948. (describes an electon-beam deflection tube 7-bit,100-kSPS
flash converter for early experimental PCM work).
26. Frank Gray, "Pulse Code Communication," U.S. Patent 2,632,058, filed November 13, 1947, issued
March 17, 1953. (detailed patent on the Gray code and its application to electron beam coders).
27. J. O. Edson and H. H. Henning, "Broadband Codecs for an Experimental 224Mb/s PCM Terminal,"
Bell System Technical Journal, Vol. 44, pp. 1887-1940, Nov. 1965. (summarizes experiments on
ADCs based on the electron tube coder as well as a bit-per-stage Gray code 9-bit solid state ADC. The
electron beam coder was 9-bits at 12 MSPS, and represented the fastest of its type).
28. Lee de Forest, "Device for Amplifying Feeble Electrical Currents," U.S. Patent 841,387, filed October
25, 1906, issued January 15, 1907.
29. Lee de Forest, "Space Telegraphy," U.S. Patent 879,532, filed January 29, 1907, issued February 18,
1908.
30. H. S. Black, "Wave Translation System," U.S. Patent 2,102,671, filed August 8, 1928, issued
December 21, 1937 (The basis of feedback amplifier systems).
31. H. S. Black, "Stabilized Feedback Amplifiers," Bell System Technical Journal, Vol. 13, No. 1,
January 1934, pp. 1-18 (A practical summary of feedback amplifier systems).
32. Harold S. Black, "Inventing the Negative Feedback Amplifier," IEEE Spectrum, December, 1977
(Inventor’s 50th anniversary story on the invention of the feedback amplifier).
33. C. Mark Melliar-Smith et al, "Key Steps to the Integrated Circuit," Bell Labs Technical Journal, Vol.
2, #4, Autumn 1997.
34. J. Bardeen, W. H. Brattain, "The Transistor, a Semi-Conductor Triode," Physical Review, Vol. 74, No.
2, July 15, 1947 pp. 230-231 (the invention of the germanium transistor).
1.18
DATA CONVERTER HISTORY
1.1 EARLY HISTORY
35. W. Shockley, "The Theory of p-n Junctions in Semiconductors and p-n Junction Transistors,"
Bell System Technical Journal, Vol. 28, No. 4, July 1949, pp. 435-489 (theory behind the germanium
transistor).
36. L. R. Wrathall, "Transistorized Binary Pulse Regenerator," Bell System Technical Journal, Vol. 35,
September 1956, pp. 1059-1084.
37. J. S. Kilby, "Invention of the Integrated Circuit," IRE Transactions on Electron Devices, Vol. ED23, No. 7, July 1976, pp. 648-654 (Kilby's IC invention at TI).
38. Robert N. Noyce, "Semiconductor Device-and-Lead Structure," U.S. Patent 2,981,877, filed July 30,
1959, issued April 25, 1961 (Noyce's IC invention at Fairchild).
39. Jean Hoerni, "Planar Silicon Diodes and Transistors," IRE Transactions on Electron Devices, Vol. 8,
March 1961, p. 168.
40. Jean A. Hoerni, "Method of Manufacturing Semiconductor Devices," U.S. Patent 3,025,589 , filed
May 1, 1959, issued March 20, 1962 (the planar process— a manufacturing means of protecting and
stabilizing semiconductors).
1.19
ANALOG-DIGITAL CONVERSION
NOTES:
1.20
DATA CONVERTER HISTORY
1.2 DATA CONVERTERS OF THE 1950S AND 1960S
SECTION 1.2: DATA CONVERTERS OF THE 1950s
AND 1960s
Walt Kester
Commercial Data Converters: 1950s
Up until the mid-1950s, data converters were primarily developed and used within
specialized applications, such as the Bell System work on PCM, and message encryption
systems of World War II. Because of vacuum tube technology, the converters were very
expensive, bulky, and dissipated lots of power. There was practically no commercial
usage of these devices.
The digital computer was a significant early driving force behind commercial ADC
development. The ENIAC computer development project was started in 1942 and was
revealed to the general public in February 1946. The ENIAC led to the development of
the first commercially available digital computer, the UNIVAC, by Eckert and Mauchly.
The first UNIVAC was delivered to the United States Census Bureau in June 1951.
Military applications, such as ballistic trajectory computation, were early driving forces
behind the digital computer, but as time went on, the possibilities of other applications in
the area of data analysis and industrial process control created more general interest in
digital processing, and hence the need for data converters. In 1953 Bernard M. Gordon, a
pioneer in the field of data conversion, founded a company called Epsco Engineering in
his basement in Concord MA. Gordon had previously worked on the UNIVAC computer,
and saw the need for commercial data converters. In 1954 Epsco introduced an 11-bit,
50-kSPS vacuum-tube based ADC. This converter is believed to be the first commercial
offering of such a device.
The Epsco "Datrac" converter dissipated 500 watts, was designed for rack mounting (19"
× 15" × 26") and sold for $8,000 to $9,000 (see Reference 1). A photograph of the
instrument is shown in Figure 1.17. The Datrac was the first commercially offered ADC
to utilize the shift-programmable successive approximation architecture, and Gordon was
granted a patent on the logic required to perform the conversion algorithm (Reference 2).
Because it had a sample-and-hold function, the Epsco Datrac was the first commercial
ADC suitable for digitizing ac waveforms, such as speech.
During the same period, a few other companies manufactured lower speed ADCs suitable
for digital voltmeter measurement applications, and there were offerings of optical
converters based on coded discs for measuring the angular position of shafts in avionics
applications (see Reference 1). Converters of the mid to late 1950s used a combination of
vacuum tubes, solid-state diodes, and transistors to implement the conversion process. A
few of the companies in the data converter business at the time were Epsco, Non-Linear
Systems, Inc., J.B. Rea, and Adage. In order to gain further insight to the converters of
the 1950s, References 1, 3, 4, 5, and 6 are excellent sources.
1.21
ANALOG-DIGITAL CONVERSION
19" × 15" × 26"
500W
150 lbs
Courtesy,
Analogic Corporation
8 Centennial Drive
Peabody, MA 01960
http://www.analogic.com
$8,500.00
Figure 1.17: 1954 "DATRAC" 11-bit, 50-kSPS Vacuum Tube ADC
Designed by Bernard M. Gordon at EPSCO
Commercial Data Converter History: 1960s
During the mid 1950s through the early 1960s, electronic circuit designs began to migrate
from vacuum tubes to transistors, thereby opening up many new possibilities in data
conversion products. As was indicated earlier, the silicon transistor was responsible for
the increased interest in solid state designs. There was more and more interest in data
converter products, as indicated in two survey articles published in 1964 (Reference 5)
and 1967 (Reference 6). Because these devices were basically unfamiliar to new
customers, efforts were begun to define specifications and testing requirements for
converter products (References 7-16).
The IBM-360 mainframe computer and solid state minicomputers (such as the DEC PDPseries starting in 1963) added to the general interest in data analysis applications. Other
driving forces requiring data converters in the 1960s were industrial process control,
measurement, PCM, and military systems.
Efforts continued during the 1960s at Bell Telephone Labs to develop high speed
converters (e.g., 9 bits, 5 MSPS) for PCM applications (Reference 17), and the military
division of Bell Labs began work on the development of hardware and software for an
anti-ballistic missile (ABM) system.
In 1958, the U.S. Army began the development of the Nike-Zeus anti-ballistic missile
system, with Bell Laboratories responsible for much of the hardware design. This
program was replaced by Nike-X in 1963, which was the first program to propose a
1.22
DATA CONVERTER HISTORY
1.2 DATA CONVERTERS OF THE 1950S AND 1960S
digitally controlled phased array radar for guiding the short and long-range interceptor
missiles. The objective of the system was to intercept and destroy incoming Soviet
nuclear warheads above the atmosphere and thereby protect U.S. population centers.
In 1967, President Lyndon Johnson and Secretary of Defense Robert McNamara
redefined the ABM program and changed the name to Sentinel. This system used
basically the same hardware as Nike-X, but the threat definition was changed from the
Soviet Union to China, where work was underway on less sophisticated ICBMs, and
nuclear capability had been demonstrated. This program provoked large scale public
protests when it became clear that nuclear tipped interceptor missiles would be deployed
very close to the cities they were meant to defend.
Richard Nixon became President in 1969, and the ABM program objective and name was
once again changed for political reasons, but still using basically the same hardware. This
time the program was called Safeguard, and the new objective was to protect Minuteman
ICBM fields, Strategic Air Command bases, and Washington DC. The system would be
deployed at up to 12 sites and utilize both short- and long-range missiles.
The Safeguard program became entangled in the politics of the SALT talks with the
Soviet Union, and was eventually scaled back significantly. In the end, the Grand Forks,
North Dakota site was the only site ever built; it became operational on October 1, 1975.
On October 2, 1975, the House of Representatives voted to deactivate the Safeguard
program.
Key to the Nike-X/Sentinel/Safeguard systems was the use of digital techniques to
control the phased array radar and perform other command and control tasks. The logic
was resistor-transistor-logic (RTL), and was mounted in hybrid packages. Also important
to the system were the high speed ADCs used in the phased array radar receiver. Early
prototypes for the required 8-bit 10-MSPS ADC were developed by John M. Eubanks
and Robert C. Bedingfield at Bell Labs between 1963 and 1965. In 1966, these two
pioneers in high speed data conversion left Bell Labs and founded Computer Labs—a
Greensboro, NC based company—and introduced a commercial version of this ADC.
The 8-bit, 10-MSPS converter was rack mounted, contained its own linear power supply,
dissipated nearly 150 watts, and sold for approximately $10,000 (see Figure 1.18). The
same technology was used to produce 9-bit, 5-MSPS and 10-bit 3-MSPS versions.
Although the next generation of Computer Labs' designs would take advantage of
modular op amps (Computer Labs OA-125 and FS-125), ICs such as the Fairchild
µA710/711 comparators, as well as 7400 TTL logic, the first ADCs offered used all
discrete devices.
The early high speed ADCs produced by Computer Labs were primarily used in research
and development projects associated with radar receiver development by companies such
as Raytheon, General Electric, and MIT Lincoln Labs.
In the mid-1960s, development of lower speed instrument, PC-board, and modular ADCs
was pioneered by such companies as Analogic (founded by Bernard M. Gordon) and
Pastoriza Electronics (founded by James Pastoriza). Other companies in data converter
business were Adage, Burr Brown, General Instrument Corp, Radiation, Inc., Redcor
1.23
ANALOG-DIGITAL CONVERSION
Corporation, Beckman Instruments, Reeves Instruments, Texas Instruments, Raytheon
Computer, Preston Scientific, and Zeltex, Inc. Many of the data converters of the 1960s
were in the form of digital voltmeters which used integrating architectures, although
Adage introduced an 8-bit, 1-MSPS sampling ADC, the Voldicon VF7, in the early 1960s
(Reference 5).
19" RACK-MOUNTED, 150W, $10,000.00
INSTALLATION OF 12 ADCs
IN EXPERIMENTAL DIGITAL
RADAR RECEIVER
Figure 1.18: HS-810, 8-bit, 10-MSPS ADC Released by
Computer Labs, Inc. in 1966
In addition to the widespread proliferation of discrete transistor circuits in the 1960s,
various integrated circuit building blocks became available which led to size and power
reductions in data converters. In 1964 and 1965, Fairchild introduced two famous Bob
Widlar IC designs: the µA709 op amp and the µA710/µA711 comparator. These were
quickly followed by a succession of linear ICs from Fairchild and other manufacturers.
The 7400-series of transistor-transistor-logic (TTL) and high-speed emitter-coupled-logic
(ECL) also emerged during this period as well as the 4000-series CMOS logic from RCA
in 1968. In addition to these devices, Schottky diodes, Zener reference diodes, FETs
suitable for switches, and matched dual JFETs also made up some of the building blocks
required in data converter designs of the period.
In 1965, Ray Stata and Matt Lorber founded Analog Devices, Inc. (ADI) in Cambridge,
MA. The initial product offerings were high performance modular op amps, but in 1969
ADI acquired Pastoriza Electronics, a leader in data converter products, thereby making a
solid commitment to both data acquisition and linear products.
Pastoriza had a line of data acquisition products, and Figure 1.19 shows a photograph of a
1969 12-bit, 10-µs general purpose successive approximation ADC, the ADC-12U, that
sold for approximately $800.00. The architecture was successive approximation, and the
ADC-12U utilized a µA710 comparator, a modular 12-bit "Minidac," and 14 7400-series
logic packages to perform the successive approximation conversion algorithm.
1.24
DATA CONVERTER HISTORY
1.2 DATA CONVERTERS OF THE 1950S AND 1960S
2.3W
$800.00
74 and
74H LOGIC
SOON REPLACED
BY 2502, 2503, 2504
SAR LOGIC ICs FROM
AMD AND NATIONAL
"MINIDAC"
µA710
COMPARATOR
Figure 1.19: ADC-12U 12-Bit, 10-µs SAR ADC from Pastoriza Division of
Analog Devices, 1969
The "Minidac" module was actually constructed from "quad switch" ICs (AD550) and a
thin film network (AD850) as shown in Figure 1.20. Figure 1.21 shows details of the
famous quad switch that was patented by James Pastoriza (Reference 18). Chapter 3 of
this book contains more discussion on the quad switch and other DAC architectures.
Notice that in the ADC-12U, the implementation of the successive approximation
algorithm required 14 logic packages. In 1958, Bernard M. Gordon had filed a patent on
the logic to perform the successive approximation algorithm (Reference 19), and in the
early 1970s, Advanced Micro Devices and National Semiconductor introduced
commercial successive approximation register logic ICs: the 2502 (8-bit, serial, not
expandable), 2503 (8-bit, expandable) and 2504 (12-bit, serial, expandable). These were
designed specifically to perform the register and control functions in successive
approximation ADCs. These became standard building blocks in many modular and
hybrid data converters. In fact, the acronym SAR actually stands for successive
approximation register, and hence the term SAR ADC. Earlier converters utilizing the
successive approximation architecture were simply referred to as sequential coders,
feedback coders, or feedback subtractor coders.
1.25
ANALOG-DIGITAL CONVERSION
LOGIC INPUTS
R
i
AD550 "µDAC"
QUAD SWITCH,
QUAD CURRENT
SOURCE
AD550 "µDAC"
QUAD SWITCH,
QUAD CURRENT
SOURCE
AD550 "µDAC"
QUAD SWITCH,
QUAD CURRENT
SOURCE
–
VOUT
+
AD850
CURRENT SETTING
PRECISION THIN FILM RESISTOR NETWORK
REFERENCE
CIRCUIT
Figure 1.20: A 1969 Vintage 12-bit "MiniDAC" Using Quad Current Switches, a
Thin Film Resistor Network, a Voltage Reference, and an Op Amp
CURRENT
OUTPUT
MSB
I
I/4
I/2
I/8
VB
R
2R
4R
8R
–V
Figure 1.21: "Quad Switch" DAC Building Block with External
Thin Film Resistor Network
1.26
DATA CONVERTER HISTORY
1.2 DATA CONVERTERS OF THE 1950S AND 1960S
Data Converter Architectures
Details of the history of the various data converter architectures are contained in Chapter
3 of this book, so for now, we will just summarize the key developments.
The basic algorithm used in the successive approximation (initially called feedback
subtraction) ADC conversion process can be traced back to the 1500s relating to the
solution of a certain mathematical puzzle regarding the determination of an unknown
weight by a minimal sequence of weighing operations (Reference 20). In this problem, as
stated, the object is to determine the least number of weights which would serve to weigh
an integral number of pounds from 1 lb to 40 lb using a balance scale. One solution put
forth by the mathematician Tartaglia in 1556, was to use the series of weights 1 lb, 2 lb,
4 lb, 8 lb, 16 lb, and 32 lb. The proposed weighing algorithm is the same as used in
modern successive approximation ADCs. (It should be noted that this solution will
actually measure unknown weights up to 63 lb rather than 40 lb as stated in the problem).
The algorithm is shown in Figure 1.22 where the unknown weight is 45 lbs. The balance
scale analogy is used to demonstrate the algorithm. The electronic implementation of the
successive approximation ADC is shown in Figure 1.23.
TEST
TOTALS:
ASSUME X = 45
X
IS X ≥ 32 ?
YES
RETAIN 32
1
IS X ≥ (32 +16) ?
NO
REJECT 16
0
IS X ≥ (32 +8) ?
YES
RETAIN 8
1
IS X ≥ (32 +8 + 4) ?
YES
RETAIN 4
1
IS X ≥ (32 +8 + 4 + 2) ?
NO
REJECT 2
0
IS X ≥ (32 +8 + 4 + 2 + 1) ?
YES
RETAIN 1
1
X = 32 + 8 + 4 + 1 = 4510
=
1011012
Figure 1.22: Successive Approximation ADC Algorithm
Analogy Using Binary Weights
1.27
ANALOG-DIGITAL CONVERSION
ANALOG
INPUT
COMPARATOR
+
–
SERIAL DATA
OUTPUT
WEIGHTING
NETWORK
CONVERT
START
CONTROL
LOGIC
Figure 1.23: Basic Successive Approximation ADC (Feedback Subtraction ADC)
It is interesting to note that all the fundamental ADC architectures used today had been
discovered and published in one form or another by the mid-1960s. Figure 1.24 shows a
detailed timeline for the development of the successive approximation ADC. Figure 1.25
shows the timeline high speed ADC architectures, and Figure 1.26 shows the timeline for
counting and integrating ADCs. Even the Σ-∆ ADC architecture had been explored as
shown in Figure 1.27. A much more detailed discussion of each individual architecture
and its history can be found in Chapter 3 of this book.
SAR algorithm dates back to the 1500's
Early SAR ADCs used individually switched binary reference
voltages rather than internal DAC (Schelling: 1946, Goodall: 1947)
Use of internal DAC rather than switched reference voltages to
perform conversion (Kaiser: 1953, B. D. Smith: 1953)
Use of non-uniformly weighted DAC to produce companding
transfer function (B. D. Smith: 1953)
First commercial vacuum tube SAR ADC, 11-bits, 50kSPS
(Bernard M. Gordon - Epsco: 1954)
Design of specific logic function to perform SAR algorithm
(Gordon: 1958)
Led to popular SAR logic ICs: 2503, 2504 from National
Semiconductor and Advanced Micro Devices in the early 1970's
Figure 1.24: SAR ADC Development Summary
1.28
DATA CONVERTER HISTORY
1.2 DATA CONVERTERS OF THE 1950S AND 1960S
Reeve’s counting ADC
1939
Successive approximation
1946
Flash (electron tube coders)
1948
Bit-per-stage (binary and Gray)
1956
Subranging
1956
Subranging with error correction
1964
Pipeline with error correction
1966
Note: Dates are first publications or patent filings
Figure 1.25: High Speed ADC Architecture Timeline
Reeve's counting ADC
1939
Charge run-down:
1946
Ramp run-up
1951
Tracking
1950
Voltage-to-frequency converter (VFC)
1952
Dual Slope
1957
Triple Slope
1967
Quad Slope
1973
Note: Dates are first publications or patent filings
Figure 1.26: Counting and Integrating ADC Architecture Timeline
1.29
ANALOG-DIGITAL CONVERSION
Delta Modulation
1950
Differential PCM
1950
Single and multibit oversampling with noise shaping
1954
First called ∆-Σ, "delta-sigma"
1962
Addition of digital filtering and decimation for Nyquist ADC
1969
Bandpass Sigma-Delta
1988
Note: Dates are first publications or patent filings
Figure 1.27: Sigma-Delta ADC Architecture Timeline
By the end of the 1960s, the key architectures and building blocks were available to allow
for modular and ultimately hybrid converters, and significant work was already underway
to produce the first monolithic converters which were to appear in the early 1970s.
1.30
DATA CONVERTER HISTORY
1.2 DATA CONVERTERS OF THE 1950S AND 1960S
REFERENCES:
1.2 DATA CONVERTERS OF THE 1950s AND 1960s
1.
G.G. Bowers, "Analog to Digital Converters," Control Engineering, April 1957, pp. 107-118.
2.
Bernard M. Gordon and Robert P. Talambiras, "Signal Conversion Apparatus," U.S. Patent 3,108,266,
filed July 22, 1955, issued October 22, 1963. (classic patent describing Gordon's 11-bit, 50-kSPS
vacuum tube successive approximation ADC done at Epsco. The internal DAC represents the first
known use of equal currents switched into an R/2R ladder network.)
3.
G. J. Herring, "High Speed Analogue to Digital Conversion," Journal British IRE, August 1957, pp.
407-420.
4.
G. J. Herring, "Electronic Digitizing Techniques," Journal British IRE, July 1960, pp. 513-517.
5.
F.M. Young, "Factors Limiting A/D Conversion State of the Art," Data Systems Engineering, May
1964, pp. 35-39.
6.
George Flynn, "Analog to Digital Converters," Electronic Products, Vol. 10, No. 5, October 1967, pp.
18-48.
7.
Bernard M. Gordon, "Definition of Accuracy of Voltage to Digital Converters," Instruments and
Control Systems, May 1959, pp.710.
8.
W. M. Gaines, "Terminology for Functional Characteristics of Analog to Digital Converters," Control
Engineering, February 1961.
9.
Bernard M. Gordon, "Designing Sampled Data Systems," Control Engineering, April 1961, pp. 127132.
10. Bernard M. Gordon, "How to Specify Analog-to-Digital Converters," Electronic Design, May 10
1961, pp. 36-39.
11. Bernard M. Gordon, "How Much Do Components Limit Converter Performance?" Electronic Design,
June 21, 1961, pp.52-53.
12. P. Barr, "Influence of Aperture Time and Conversion Rate on the Accuracy of A/D Converters," Data
Systems Engineering, May 1964, pp. 30-34.
13. A. Van Doren, "Solving Error Problems in Digital Conversion Systems," Electromechanical Design,
April 1966, pp.44-46.
14. J. Freeman, "Specifying Analog to Digital Converters," The Electronic Engineer, June 1968, pp. 4448.
15. Bernard M. Gordon, "Speaking Out on Analog to Digital Converters," EEE Magazine, December
1968.
16. Bernard M. Gordon, "Bernard Gordon of Analogic Speaks Out on What's Wrong with A/D Converter
Specs," EEE Magazine, February 1969, pp. 54-61.
17. J. O. Edson and H. H. Henning, "Broadband Codecs for an Experimental 224Mb/s PCM Terminal,"
Bell System Technical Journal, Vol. 44, pp. 1887-1940, Nov. 1965.
18. James J. Pastoriza, "Solid State Digital-to-Analog Converter," U.S. Patent 3,747,088, filed December
30, 1970, issued July 17, 1973. (the first patent on the quad switch approach to building high
resolution DACs).
1.31
ANALOG-DIGITAL CONVERSION
19. Bernard M. Gordon and Evan T. Colton, "Signal Conversion Apparatus," U.S. Patent 2,997,704, filed
February 24, 1958, issued August 22, 1961. (classic patent describes the logic to perform the
successive approximation algorithm in a SAR ADC).
20. W. W. Rouse Ball and H. S. M. Coxeter, Mathematical Recreations and Essays, Thirteenth Edition,
Dover Publications, 1987, pp. 50, 51. (describes a mathematical puzzle for measuring unknown
weights using the minimum number of weighing operations. The solution proposed in the 1500's is the
same basic successive approximation algorithm used today).
Note on Anti-Ballistic Missile (ABM) system history:
Although most of the information regarding the various ABM systems was classified at the time, today the
information is in the public domain. There are several excellent websites which can provide further
information and references:
http://www.ucsusa.org is a site maintained by the Union of Concerned Scientists, 2 Brattle Square,
Cambridge, MA 02238. There is a section on the site under the security section titled "From Nike-Zeus to
Safeguard: US Defenses Against ICBMs, 1958-1976.
http://www.paineless.id.au/missiles is a site maintained by an individual with an excellent collection of
references about the ABM programs. A good whitepaper on the site is titled "Nuclear ABM Defense of the
USA."
In addition to the websites, the following book published by the Bell System provides a good discussion of
the development of the ABM system:
M. D. Fagen, A History of Engineering and Science in the Bell System: National Service in War and
Peace (1925-1975), Bell Telephone Laboratories, 1978, ISBN 0-932764-00-2.
1.32
DATA CONVERTER HISTORY
1.3 DATA CONVERTERS OF THE 1970S
SECTION 1.3: DATA CONVERTERS OF THE 1970s
Introduction
The year 1970 began one of the most exciting decades in the history of data converters.
The ADC/DAC market was driven by a number of applications, including high resolution
digital voltmeters, industrial process control, digital video, military phased array radar,
medical imaging, vector scan displays, and raster scan displays. Most of these systems
had formerly utilized conventional analog signal processing techniques, and the increased
availability of low cost computing technology generated a desire to take advantage of the
increased performance and flexibility offered by digital signal processing and analysis—
and of course, the need for compatible data converters.
As a result, a number of companies entered the data converter field, including Analog
Devices, Analogic Corporation (initially Epsco and later Gordon Engineering), Burr
Brown, Computer Labs, Datel, Hybrid Systems, ILC/Data Device Corporation,
Micronetworks, National Semiconductor, Teledyne Philbrick, and Zeltex.
Integrated circuit building blocks, as well as complete IC data converters of the 1970s,
came from Analog Devices, Advanced Micro Devices, Fairchild, Signetics, Intersil,
Micro Power Systems, Motorola, National Semiconductor, TRW (LSI Division), and
Precision Monolithics.
The data converters of the 1970s made maximum utilization of all the technologies
available: monolithic, modular, and hybrid, with the modular and hybrid products
typically offering higher resolution and faster speed than the existing monolithic parts.
An often overlooked fact is that both customer education and high quality application
support is required in order to take full advantage of an emerging technology such as data
conversion. From the beginning, Analog Devices has always realized the importance of
excellent application material, starting with a set of tutorial articles on op amps by its
founder, Ray Stata (Reference 1). These articles were published in 1965—the same year
Analog Devices was founded—and are still recognized as classic tutorials on basic op
amp theory and applications.
The ADI continuing thread of customer support through applications information was
enhanced considerably in 1967, when the Analog Dialogue magazine was launched (see
Reference 2). The initial charter for the magazine was stated as "A Journal for the
Exchange of Operational Amplifier Technology," later on this was broadened to "A
Journal for the Exchange of Analog Technology."
Dissemination of analog circuit information is what the early Analog Dialogue did, and
did it well. The premier issue featured an op amp article by Ray Stata that is still
available as an application note (see Reference 3). A similar comment can also be made
for a subsequent Ray Stata article (see Reference 4).
A milestone in the life of the young magazine was the arrival of Dan Sheingold as editor,
in 1969 (see Reference 5). Already highly experienced as a skilled op amp expert and
1.33
ANALOG-DIGITAL CONVERSION
editorial writer from vacuum tube and early solid-state years at George A. Philbrick
Researches (GAP/R), Dan Sheingold brought a unique set of skills to the task of editorial
guidance for Analog Dialogue. Dan's leadership as editor continues today, in 2004. For
more than 35 years his high technical communication standards have been an industry
benchmark.
Realizing the need for a comprehensive book on the newly emerging field of data
conversion, Analog Devices published the first edition of the Analog-Digital Conversion
Handbook in 1972, under the excellent editorship of Dan Sheingold (Reference 6). An
interim revision, Analog-Digital Conversion Notes, was published in 1977 (Reference 7).
In 1986, in conjunction with Prentice-Hall, Sheingold published a third revision, again
titled Analog-Digital Conversion Handbook (Reference 8). All of these books provided
detailed information on data converter architectures, specifications, design, and
applications, and helped in the adoption of uniform terminology and performance metrics
throughout the industry.
Monolithic Data Converters of the 1970s
Bipolar Process IC DACs of the 1970s
The earliest monolithic DACs were made using bipolar process technology; they
included only the basic core of a complete DAC—the array of switches and resistors to
set the weight of each bit. An example is the 1408 and a later higher speed derivative, the
DAC08, introduced in 1975 and shown in Figure 1.28.
Figure 1.28: DAC08 8-Bit 85-ns IC DAC, 1975
These converters were produced by several manufacturers and were available at low cost.
However, they required many additional external components in order to be usable in a
system design. These external components included several resistors, a voltage reference,
a latch, an output op amp, possibly a compensation capacitor, and usually one or more
trimming potentiometers.
Converters like the 1408 and DAC08 were limited to 8-bit accuracy by the matching and
tracking limitations of the diffused resistors. When higher accuracy is required, lower
tempco resistors are needed, and some means of post-fabrication adjustment is desirable.
1.34
DATA CONVERTER HISTORY
1.3 DATA CONVERTERS OF THE 1970S
Thin film resistors exhibit low tempcos and can be trimmed with a laser—they are well
suited for use in data converters. By the mid 1970s, Analog Devices had developed
considerable expertise, not only the deposition of thin film resistors, but also trimming at
the wafer level.
The Analog Devices' AD562, designed by Bob Craven and introduced in 1974, was
originally a "compound monolithic" manufactured using two IC chips mounted in the
same package, without the traditional hybrid substrate for mounting and interconnection.
Instead, the two chips were designed so that a set of wire bonds between the two chips (in
addition to the usual ones to the package pins) were all that were necessary to assemble a
12-bit accurate DAC in an IC package. In the original AD562, one chip contained the
thin film resistor network (including the bit weight-setting resistors and output gainsetting resistors), and the other chip contained the reference control amplifier and the
current switches for the 12 bits. As the processing matured, the manufacture of larger
chips became more practical. The two chips of the original AD562 were later merged into
a single-chip version. It became the first 12-bit DAC qualified by the U.S. Department of
Defense under MIL-M-38510.
While the AD562 was the first 12-bit IC DAC and embodied the solution to some
extremely difficult design problems; it was still really only a building block, since it
lacked buffer latches, a voltage reference, and an output amplifier. Shortly after the twochip AD562 was introduced (Reference 9, 10), a version with a third chip was developed.
The third chip was a 2.5-V bandgap reference (designed by Paul Brokaw and described in
detail in Reference 11 and 12). This made the DAC function much more complete. The
resulting product, known as the AD563, also became quite popular and eventually made
the transition to a completely monolithic single-chip device.
Another problem with the AD562 was that, while reasonably fast, it lacked sufficient
speed for many applications—its settling time was approximately 1 µs. Later advances in
switch design and Zener diode fabrication led to a higher speed DAC, the AD565,
introduced in 1978 (later followed by the AD565A in 1981). A simplified diagram of the
AD565 is shown in Figure 1.29.
The bit switches used in this design were much smaller than those used in the AD562,
allowing a substantial reduction in chip area and increasing the yield of good chips per
wafer. The new switches yielded a settling time of 200 ns to ½ LSB. The AD565 used a
buried Zener reference which had less noise than the bandgap reference used in the
AD563.
The AD565 retained the same pinout as the earlier AD563, allowing drop-in replacement
with improved performance and lower price.
It should be noted that the first monolithic single-chip 10-bit DAC with thin film laser
wafer trimmed (LWT) resistors and internal reference was the AD561, designed by Peter
Holloway, which was introduced by Analog Devices in 1976 (Reference 13). This DAC
utilized a method of compensating for errors produced by operating the internal current
source transistors at different current densities. This idea is patented by Paul Brokaw and
is one of the most widely referenced patents in data conversion (Reference 14).
1.35
ANALOG-DIGITAL CONVERSION
Figure 1.29: AD565, 12-Bit, 200-ns DAC, 1978
CMOS IC DACs of the 1970s
As we have seen, the early commercially available monolithic DACs were principally
processed by conventional bipolar linear processing techniques. Before 1974, when the
AD7520 CMOS DAC was introduced, 10-bit conversion had been difficult to obtain with
good yields (and low cost) because of the finite β of switching devices, the VBE-matching
requirement, the matching and tracking requirements on the diffused resistor ladders, and
the tracking limitations caused by the thermal gradients produced by high internal power
dissipation.
Most of these problems were solved or avoided with CMOS devices. The CMOS
transistors have nearly infinite current gain, eliminating β problems. There is no
equivalent in CMOS circuitry to a bipolar transistor's VBE drop; instead, a CMOS switch
in the ON condition is almost purely resistive, with the resistance value controllable by
device geometry. The temperature problems of diffused resistors were eliminated by
using thin film resistors instead. A simplified diagram of the AD7520 10-bit, 500-ns
CMOS multiplying DAC introduced in 1974 is shown in Figure 1.30.
The AD7520 architecture is a standard current-mode R-2R (also referred to as an
"inverted R-2R") and is described further in Chapter 3 of this book. The output drives the
inverting input of an external op amp connected as an I/V converter. The 10-kΩ feedback
resistor for the op amp is internal to the AD7520 to provide good tracking. The key to the
linearity of the AD7520 is that the geometries of the switches corresponding to the first
six bits are tapered so as to obtain ON resistances that are related in binary fashion.
The AD7520 architecture was extended to 12-bit resolution in the AD7541 by merely
adding additional switch cells and resistors. However, in order to achieve 12-bit linearity,
laser trimming at the wafer level was required. The AD7541, introduced in 1978, was the
first 12-bit CMOS multiplying DAC. Settling time to ½ LSB was 1 µs.
1.36
DATA CONVERTER HISTORY
1.3 DATA CONVERTERS OF THE 1970S
S10
VDD
BIT 10 (LSB)
GND
Figure 1.30: AD7520, the First Monolithic CMOS Multiplying 10-Bit DAC, 1974
The AD7520 and AD7541 were the beginning of an entire product line of general
purpose multiplying CMOS DACs from Analog Devices. Some of these products will be
discussed in Section 1.3, Data Converters of the 1980s.
Another useful feature for a DAC is the addition of an on-chip latch (generally referred to
as a "buffered" DAC). The latch allows the DAC to be connected to a microprocessor
data bus. The AD7524, shown in Figure 1.31, was an 8-bit multiplying CMOS DAC
which had an on-chip latch. Data is loaded into the DAC by first asserting the
CHIP SELECT pin. Data is then written into the latch by asserting the WRITE pin.
Returning the WRITE pin to 0 disconnects the latch from the data bus, and the bus can
then be connected to another device if desired.
S8
VDD
CHIP SELECT
DATA LATCHES
WRITE
DB7 (MSB)
DB6
DB5
GND
DB0 (LSB)
DATA INPUTS
Figure 1.31: AD7524 8-Bit Buffered µP Compatible DAC, 1978
1.37
ANALOG-DIGITAL CONVERSION
Future DAC products added a second latch, and are referred to as a "double-buffered"
DAC. The input latch is used to load the data (either serial, parallel, or in bytes), and
when the second parallel "DAC latch" is clocked, the DAC output is updated.
Data converter designers soon realized the importance of making them easy to interface
to microprocessors and DSPs, and this trend added additional functionality to the devices
and continues to this day.
A summary of the monolithic DAC developments during the 1970s is shown in Figure
1.32.
Bipolar:
AD550 "µDAC" Building Block Quad Switch, 1970
AD562 12-Bit, 1.5µs (2-chip, compound monolithic) DAC, 1974
1408 8-Bit 250ns DAC, 1975
DAC08 8-Bit, 80ns DAC, 1976
AD561 10-Bit, 250ns, LWT Current-Output DAC with Reference, 1976
AD565 12-Bit, 200ns, LWT Current-Output DAC with Reference, 1978
CMOS:
AD7520 10-Bit, 500ns, Multiplying DAC, 1974
AD7541 12-Bit, 1µs, LWT Multiplying DAC, 1978
AD7524, 8-Bit, 150ns LWT Multiplying DAC with DAC buffer latch,
µP interface, 1978
Figure 1.32: Summary: Monolithic DACs of the 1970s
Monolithic ADCs of the 1970s
Although most of the ADCs of the early 1970s were modular or hybrid, there was
considerable effort by data converter manufacturers to produce an all-monolithic ADC.
An early attempt was the AD7570 10-bit, 20-µs CMOS SAR ADC introduced in 1975.
However, due to the difficulty of designing good comparators, amplifiers, and references
on the early CMOS process, the AD7570 required an external LM311 comparator as well
as a voltage reference.
The integrating ADC architecture was suitable for the early CMOS processes, and in
1976, Analog Devices introduced the 13-bit AD7550 which utilized a unique architecture
called "quad slope." The architecture was patented by Ivar Wold (Reference 15).
The first complete monolithic ADC was the 10-bit, 25-µs AD571 SAR ADC introduced
in 1978 and designed by Paul Brokaw (Reference 16). The AD571 was designed on a
bipolar process with LWT thin film resistors. In order to implement the logic functions
required in the SAR ADC, integrated-injection logic (I2L) was added to the bipolar
process. This process allowed reasonably dense low voltage logic to be included on the
same chip as high-breakdown precision linear circuitry.
1.38
DATA CONVERTER HISTORY
1.3 DATA CONVERTERS OF THE 1970S
The I2L process was particularly useful in manufacturing ADCs, because only a single
additional diffusion step was required beyond those used in the standard linear process.
Furthermore, this diffusion did not significantly interfere with the other steps in the
process, so the analog circuitry was relatively unaffected by the addition of the logic.
The 10-bit AD571 (and the 8-bit AD570) were completely self-contained monolithic
ADCs with internal clock, buried Zener voltage reference, laser-trimmed DAC (based on
the design in Reference 14), and three-state output buffers. A simplified diagram of the
AD571 is shown in Figure 1.33.
Figure 1.33: AD571 Complete 10-Bit, 25-µs IC ADC, 1978
Probably the most significant SAR ADC ever introduced was the 12-bit, 35-µs AD574 in
1978. The AD574 represents a complete solution, including buried Zener reference,
timing circuits, and three-state output buffers for direct interfacing to an 8-, 12-, or 16-bit
microprocessor bus. In its introductory form, the AD574 was manufactured using
compound monolithic construction, based on two chips—one an AD565 12-bit currentoutput DAC, including reference and thin film scaling resistors; and the other containing
the successive approximation register (SAR) and microprocessor interface logic functions
as well as a precision latching comparator. The AD574 soon emerged as the industrystandard 12-bit ADC in the early 1980s. In 1985, the device became available in singlechip monolithic form for the first time; thereby making low-cost commercial plastic
packaging possible. A simplified block diagram of the AD574 is shown in Figure 1.34.
1.39
ANALOG-DIGITAL CONVERSION
Figure 1.34: The Industry-Standard AD574 12-Bit, 35-µs IC ADC, 1978
As the 1970s came to a close, the first high-speed video flash ADCs emerged, starting
with the TDC-1007J 8-bit, 30-MSPS from the LSI division of TRW in 1979 (Reference
17). TRW also introduced a lower power 6-bit version, the TDC-1014J. Also in 1979,
Advanced Micro Devices introduced the AM6688 4-bit, 100-MSPS flash ADC, designed
by Jim Giles, who had previously designed the AM685 and AM687 fast ECL
comparators. There is more discussion on the history of flash converters in Section 1.4,
Data Converters of the 1980s. A summary of key monolithic ADC developments in the
1970s is given in Figure 1.35.
AD7570, 10-bit, 20µs CMOS SAR ADC, (required external comparator,
reference) 1975
AD7550, 13-bit integrating ADC based on "quad slope" technique, 1976
AD571, Complete 10-bit, 25µs monolithic SAR ADC, with reference, using
I2L and LWT thin film resistors, designed by Paul Brokaw, 1978
AD574, 12-bit, 35µs two-chip, compound monolithic complete SAR ADC:
AD565 DAC + logic chip), 1978; single-chip version 1985
Flash ADCs:
TRW TDC-1007J / TDC-1016J, 8-bit / 6-bit 30MSPS ADCs, 1979
AM6688 4-bit, 100MSPS ADC, 1979
Figure 1.35: Summary: Monolithic ADCs of the 1970s
1.40
DATA CONVERTER HISTORY
1.3 DATA CONVERTERS OF THE 1970S
Hybrid Data Converters of the 1970s
Although a few more instrument-type rack-mounted data converters were developed in
the early 1970s (such as the VHS-series and 7000-series from Computer Labs, Inc.), the
demand for lower cost more compact high performance data converters led
manufacturers to turn toward hybrid and modular techniques—as the monolithic
technology of the period was not yet capable of supporting the high-end converter
functions in single-chip form.
Hybrid and modular data converter designers of the 1970s had a virtual smorgasbord of
components from which to choose, including IC op amps, IC DACs, comparators,
discrete transistors, various logic chips, etc. Figure 1.36 lists some of the more popular
hybrid and modular building blocks for the 1970s.
Quad switches (AD550 µDAC)
Precision thin film resistor networks (AD850)
IC DACs: AD562, AD563, AD565, 1408, DAC08
IC Comparators: µA710, µA711, NE521, LM311, LM361, MC1650, AM685,
AM687
Successive approximation registers (SARs): 2502, 2503, 2504
IC and hybrid op amps
IC voltage references, Zener references
Fast PNP and fast NPN discrete transistors
Matched monolithic dual FETs
Monolithic transistor arrays (RCA CA-series)
Schottky diodes
CMOS and DMOS switches
TTL, CMOS, ECL logic
4, 6, 8-bit monolithic flash ADCs (starting in 1979)
Figure 1.36: Building Block Components for 1970s Hybrid and
Modular Data Converters
Hybrids generally utilize ceramic substrates with either thick or thin film conductors.
Individual die are bonded to the substrate (usually with epoxy), and wire bonds make the
connections between the bond pads and the conductors. The hybrid is usually
hermetically sealed in some sort of ceramic or metal package. Accuracy was achieved by
trimming thick or thin film resistors after assembly and interconnection, but before
sealing. Manufacturers used thin film networks, discrete thin film resistors, deposited
thick or thin film resistors, or some combination of the above.
Although the chip-and-wire hybrid was certainly more expensive to manufacture than an
IC, it allowed performance levels that could not be achieved with existing monolithic
technology of the period. The popular hybrid circuits followed an evolutionary path to
monolithic form, generally over a 5- to 10-year period, depending upon the particular
device. One of the most popular 12-bit DACs of the 1970s was the DAC80, originally
1.41
ANALOG-DIGITAL CONVERSION
introduced in the mid 1970s as a hybrid device consisting of 11 chips: three quad switch
arrays, two op amps, two resistor networks, a Zener diode, two clamp diodes, and a chip
capacitor (see Figure 1.37A). In 1978, when monolithic technology had progressed to the
point where it was possible to combine the switch and resistor network into a single chip,
a three-chip DAC80 was introduced as shown in Figure 1.37B. The three chips in this
design included a voltage reference, an output op amp, and the switch/resistor/controlamplfier chip.
1975
1978
(A) EARLY INDUSTRYSTANDARD VERSION
11-CHIPS
(B) 3-CHIP VERSION
(C) SINGLE-CHIP
CERAMIC
PACKAGE
(C) SINGLE-CHIP
PLASTIC
PACKAGE
1983
1984
Figure 1.37: DAC-80 12-Bit DAC Evolution
The newer design offered performance identical to that of the original DAC80, but with a
tremendous improvement in reliability and at a much lower cost. Then, in 1983, the first
single-chip DAC80 became available (Figure 1.37C). It, of course, provided further cost
reduction and reliability improvement, compared to the 3- and 11-chip versions. Finally,
in 1984, this popular device was offered in a low-cost plastic DIP package. Thus, in
approximately 10 years, the DAC80 evolved from a relatively high-cost hybrid to a highvolume commodity IC.
Another excellent example of hybrid technology was the AD572 12-bit, 25-µs SAR ADC
introduced in 1977. The AD572 was complete with internal clock, voltage reference,
comparator, and input buffer amplifier. The SAR register was the popular 2504. The
internal DAC was comprised of a 12-bit switch chip and an actively trimmed thin film
ladder network (separately packaged as the two-chip AD562 DAC). The AD572 was the
first military-approved 12-bit ADC processed to MIL-STD-883B, and specified over the
full operating temperature range of –55°C to +125°C. A photograph of the AD572 is
shown in Figure 1.38.
1.42
DATA CONVERTER HISTORY
1.3 DATA CONVERTERS OF THE 1970S
1.7" × 1.1" × 0.2", 0.9W
Figure 1.38: AD572 12-Bit, 25-µs Mil-Approved Hybrid ADC, 1977
Many hybrid circuits were introduced in the 1970s, and a few of the key ones are listed in
Figure 1.39. Toward the end of the 1970s, Computer Labs, Inc., introduced a number of
very fast hybrid data converters based on thick film laser trimmed resistor technology
(Computer Labs was acquired by Analog Devices in 1978). The thick film resistor
technology developed at Computer Labs during the 1970s was capable of 12-bit
accuracy, which was quite remarkable, as most hybrid manufacturers used more
expensive thin film resistors for 12-bit devices.
DAC80, 12-bit DAC, 1975
ADC80, 12-bit, 25µs SAR ADC, 1975
AD572, 12-bit, 25µs Mil-Approved ADC, 1977
HDS-1250 12-bit, 35ns DAC (also 8-, 10-bit versions), 1979
HAS-1202, 12-bit, 2.2µs SAR ADC (also 8-, 10-bit versions), 1979
HTC-0300, 300ns SHA; HTS-0025, 25ns SHA, 1979
Figure 1.39: Hybrid ADC and DAC Milestones of the 1970s
Both the HDS-series DACs and the HAS-series ADCs utilized actively trimmed thick
film resistors and discrete PNP transistor switches for the internal DACs.
It should be noted at this point that none of the monolithic or hybrid ADCs of the 1970s
were sampling ADCs with internal sample-and-holds (SHAs). In order to process ac
signals, a separate SHA had to be connected to the ADC (with the appropriate interface
1.43
ANALOG-DIGITAL CONVERSION
and timing circuits). This generated the need for hybrid SHAs, such as the HTC-0300 and
the faster HTS-0025.
Modular Data Converters of the 1970s
Designers of modular data converters in the 1970s had even more flexibility than hybrid
designers. In fact, modular technology began in the late 1960s before hybrids became
popular, and still exists to this day for certain products. Figure 1.40 shows two of the
early popular modular ADCs: the ADC-12QZ and the MAS-1202. The modular
technology was quite straightforward—components were mounted on a small PC board
and encapsulated in a potted module after trimming (usually done with manually selected
resistors). The potting compound served to distribute the heat throughout the module,
provided some degree of thermal tracking between critical components, and made it
somewhat more difficult for competitors to reverse engineer the circuits.
ADC-12QZ GENERAL PURPOSE 12-BIT,
40µs SAR ADC INTRODUCED IN 1972
2" × 4" × 0.4", 1.8W, $130.00
Quad-Switch, Thin-Film Network
Internal DAC
MAS-1202 12-BIT, 2µs
SAR ADC INTRODUCED IN 1975
2" × 4" × 0.4", 2.4W, $270.00
Discrete PNP-Transistor Internal DAC,
Manually Selected Precision Resistors
Figure 1.40: Early Modular ADCs of the 1970s
The ADC-12QZ 12-bit, 40-µs SAR ADC was introduced by Analog Devices in 1972, an
outgrowth of the Pastoriza Electronics' early converter product line acquired in 1969. The
ADC-12QZ utilized a DAC made up of three "quad switch" building blocks previously
discussed and a matching pre-trimmed thin film resistor network. A SAR logic chip,
comparator, a reference, and miscellaneous other components completed the parts list.
The MAS-1202 12-bit, 2-µs SAR ADC designed at Computer Labs, Inc., was released in
1975 and utilized an internal DAC based on fast PNP transistor switches and manuallyselected precision resistors.
The ADC-12QZ and the MAS-1202 each dissipated approximately 2 W, and commanded
premium prices of $130.00 and $270.00, respectively.
One of the first complete modular sampling ADCs of the 1970s utilized an open-card
construction with a combination of hybrid, IC, and discrete building blocks. The
MOD-815 8-bit, 15-MSPS ADC was introduced by Computer Labs in 1976. The design
1.44
DATA CONVERTER HISTORY
1.3 DATA CONVERTERS OF THE 1970S
utilized two 4-bit flash converters in a subranging architecture (see Chapter 3 of this
book). Each 4-bit flash converter was constructed of 8 dual AM687 ECL comparators.
The MOD-815 was one of the first commercial ADCs used in the rapidly emerging field
of digital television.
Other popular card-level modules included the MOD-1205 12-bit, 5-MSPS ADC and the
MOD-1020 10-bit, 20-MSPS ADC introduced by Analog Devices/Computer Labs in
1979. These products made use of the first flash converters on the market: the Advanced
Micro Devices' AM6688 4-bit, 100-MSPS flash and the TRW TDC-1007J (8-bit) and
TDC-1014J (6-bit) flash converters. Figure 1.41 shows a photo of the MOD-1020 with
the key devices labeled. The architecture was subranging, with two "stacked" AM6688s
providing the first 5-bit conversion, and the 6-bit TDC-1014J the second 6-bit
conversion. The extra bit was used for error correction. The MOD-1020 utilized quite a
bit of ECL logic and dissipated a total of 21 W. Due to the high level of performance and
the large number of costly hybrid and IC building blocks, the MOD-1020 commanded a
premium price of $3,500.00.
TRW TDC1014J
6-BIT, 30MSPS
FLASH
ADC
ANALOG
DELAY
LINE
AMP
AM6688
4-BIT,
100MSPS
FLASH
ADCs
SHA
DAC
DIFFERENTIAL
ECL OUTPUTS
TIMING
TIMING
5" × 7" × 0.5", 21W, $3,500.00
Figure 1.41: MOD-1020, 10-Bit, 20-MSPS Sampling ADC Introduced in 1979
Starting in the mid-1970s, most of the modular sampling converters were tested using
FFT techniques to measure SNR, ENOB, and distortion. (See Chapter 5 of this book,
Testing Data Converters).
A summary of some of the popular modular ADCs and DACs of the 1970s is given in
Figure 1.42.
1.45
ANALOG-DIGITAL CONVERSION
DAC-12QZ, 12-bit DAC, 1970
ADC-12QZ, 12-bit, 40µs SAR ADC, 1972
MAS-1202, 12-bit, 2µs SAR ADC (also 8- 10-bit versions) 1975
ADC1130, 14-bit, 12µs SAR ADC, 1975
MDS-1250, 12-bit, 50ns DAC, (also 8-, 10-bit versions) 1975
THS-0300, 300ns SHA; THS-0025, 25ns SHA, 1975
MOD-815, 8-bit, 15MSPS Video, sampling ADC, 1976
SDC1700, Synchro-to-Digital Converter, 1977
DAC1138, 18-bit DAC, (most accurate DAC for 10 years), 1977
MOD-1205, 12-bit, 5MSPS, sampling ADC, 1979
MOD-1020, 10-bit, 20MSPS, sampling ADC, 1979
Figure 1.42: Modular ADC and DAC Milestones of the 1970s
1.46
DATA CONVERTER HISTORY
1.3 DATA CONVERTERS OF THE 1970S
REFERENCES:
1.3 DATA CONVERTERS OF THE 1970s
1.
Ray Stata, "Operational Amplifiers - Parts I and II," Electromechanical Design, September,
November, 1965.
2.
Analog Dialogue, Vol. 1, No. 1, April 1967 (The premier issue of Analog Dialogue— 'A Journal for
the Exchange of Operational Amplifier Technology').
3.
Ray Stata, "Operational Integrators," Analog Dialogue, Vol. 1, No. 1, April, 1967 (Reprinted as ADI
AN357).
4.
Ray Stata, "User's Guide to Applying and Measuring Operational Amplifier Specifications," Analog
Dialogue, Vol. 1, No. 3 (Reprinted as ADI AN356).
5.
Analog Dialogue, Vol. 3, No. 1, March 1969 (The first issue of Analog Dialogue under the editorial
guidance of Dan Sheingold).
6.
Dan Sheingold, Analog-Digital Conversion Handbook, First Edition, Analog Devices, 1972.
7.
Dan Sheingold, Analog-Digital Conversion Notes, Analog Devices, 1977.
8.
Dan Sheingold, Analog-Digital Conversion Handbook, Analog Devices/Prentice-Hall, 1986, ISBN0-13-032848-0.
9.
Robert B. Craven, "An Integrated Circuit 12-bit D/A Converter, IEEE International Solid-State
Circuits Conference Digest of Technical Papers, February 1975, pp. 40-41.
10. Robert B. Craven, "Solid-State Digital-to-Analog Converter, U.S. Patent 3,961,326, filed September
12, 1974, issued June 1, 1976.
11. Paul Brokaw, "A Simple Three-Terminal IC Bandgap Voltage Reference," IEEE Journal of Solid
State Circuits, Vol. SC-9, December, 1974.
12. Paul Brokaw, "More About the AD580 Monolithic IC Voltage Regulator," Analog Dialogue, 9-1,
1975.
13. Peter Holloway and Mark Norton, "A High-Yield Second-Generation 10-Bit Monolithic DAC,"
ISSCC Digest of Technical Papers, February 1976, pp. 106-107.
14. Adrian Paul Brokaw, "Digital-to-Analog Converter with Current Source Transistors Operated
Accurately at Different Current Densities," U.S. Patent 3,940,760, filed March 21, 1975, issued
February 24, 1976.
15. Ivar Wold, "Integrating Analog-to-Digital Converter Having Digitally Derived Offset Error
Compensation and Bipolar Operation without Zero Discontinuity," U.S. Patent 3,872,466, filed July
19, 1973, issued March 18, 1975.
16. A. Paul Brokaw, "A Monolithic 10-Bit A/D Using I2L and LWT Thin-Film Resistors," IEEE Journal
of Solid-State Circuits, Vol. SC-13, pp. 736-745, December 1978.
17. J. Peterson, "A Monolithic Video A/D Converter," IEEE Journal of Solid-State Circuits, Vol. SC14, No. 6, December 1979, pp. 932-937.
1.47
ANALOG-DIGITAL CONVERSION
NOTES:
1.48
DATA CONVERTER HISTORY
1.4 DATA CONVERTERS OF THE 1980S
SECTION 1.4: DATA CONVERTERS OF THE 1980s
Walt Kester
Introduction
The 1980s represented high growth years for both IC, hybrid, and modular data
converters. The driving market forces were instrumentation, data acquisition, medical
imaging, professional and consumer audio/video, computer graphics, and a host of others.
The increased availability of relatively low-cost microprocessors, high-speed memory,
DSPs, and the emergence of the IBM-compatible PC increased the interest in all areas of
signal processing. The emphasis in ADCs began to rapidly shift to include acperformance and wide dynamic range; and hence, a great demand for sampling ADCs at
all frequencies. Specifications such as signal-to-noise ratio (SNR), signal-to-noise and
distortion (SINAD), effective number of bits (ENOB), noise power ratio (NPR), spurious
free dynamic range (SFDR), aperture time jitter, etc., began to appear on most ADC data
sheets; and glitch impulse area, SFDR, etc., on DAC data sheets.
There was a proliferation of high-speed bipolar and CMOS flash ADCs in the 1980s,
with 4-, 6-, 8-, 9-, and 10-bit at sampling rates from 20 MSPS to 100 MSPS. Digital
video was a chief driving force for the 8-, 9-, and 10-bit devices. In the graphics display
area, high-speed video RAM-DACs emerged, with CMOS being the ideal process for
these memory-intensive devices.
Voiceband and audio signal processing led to the demand for 16- and 18-bit ADCs and
DACs, and the emergence of the compact disk (CD) player fueled the need for low-cost
audio DACs.
In the 1980s, general purpose ADCs and DACs began to offer more resolution, more
functionality, and more complete solutions to the problems of data acquisition and
distribution, including multichannel ADCs and DACs. The development of linearcompatible CMOS processes (such as the Analog Devices' LC2MOS and BiCMOS II in
the mid-1980s) allowed data converter designers to provide more functionality by the
addition of features such as on-chip voltage references and buffer amplifiers, for
example.
Another important process development in the mid-1980s was the introduction of Analog
Devices' first-generation complementary bipolar (CB) process, which offered high-speed,
high-performance matching PNP and NPN transistors. The high-speed op amps produced
on the CB process made excellent drivers for many of the new ADCs, and the CB process
eventually yielded some extremely high performance IF-sampling ADCs in the 1990s.
Refer to Chapter 4 of this book, Data Converter Process Technology, for more discussion
regarding this topic.
Monolithic DACs of the 1980s
A listing of key IC DAC product introductions during the 1980s is shown in Figure 1.43.
Rather than discuss each one individually, we will simply point out examples which
illustrate the general trends in the product line.
1.49
ANALOG-DIGITAL CONVERSION
AD558, 8-bit, 1µs, µP compatible, voltage output, Bipolar/I2L DAC, 1980
AD7528, Dual 8-bit, buffered,CMOS MDAC, 1981
AD7546, 16-bit, segmented, CMOS voltage mode DAC (required external amps), 1982
AD7545, Buffered,12-bit CMOS MDAC, 1982
AD390, Quad 12-bit voltage output DAC (compound monolithic), 1982
AD7240, 12-bit, voltage mode, CMOS DAC, 1983
AD7226, Quad 8-bit, double-buffered, voltage output, LC2MOS DAC, 1984
AD9700, 125MSPS, 8-bit Video ECL DAC, 1984
AD7535, 14-bit, double-buffered LC2MOS MDAC, 1985
AD569, 16-bit, segmented, double-buffered voltage output, BiCMOS DAC, 1986
AD7245, 12-bit, double-buffered, voltage output, LC2MOS DAC, with reference, 1987
AD1856/AD1860, 16-/18-bit audio BiCMOS DACs for CD players, 1988
ADV453/ADV471/ADV476/ADV478 CMOS Video RAM-DACs, 1988
AD7840, 14-bit, double-buffered, voltage output, LC2MOS DAC, with reference, 1989
AD7846, 16-bit, segmented, voltage output, LC2MOS DAC, 1989
Figure 1.43: Monolithic DACs of the 1980s
The AD558 8-bit, 1-µs CMOS DAC introduced in 1980 illustrates the trend toward
microprocessor compatible interfaces, which are almost universal today for generalpurpose DACs and ADCs. Later products would utilize double-buffered digital inputs,
where an input register accepted parallel, serial, or byte-wide data, and a second parallel
latch was used to actually update the DAC switches.
The AD7546, although requiring two external op amps to perform the complete 16-bit
DAC function, illustrates the trend toward higher resolution. The segmented architecture
used in the AD7546 was later utilized in the 16-bit AD569, which integrated the entire
function onto one chip.
The introduction of multiple DACs in the same package is illustrated by the dual
AD7528, the quad AD390 (compound monolithic), and the quad AD7226 (single-chip).
The initial CMOS DACs were current-output, requiring an external op amp to perform
the current-to-voltage conversion; but with the advent of LC2MOS and BiCMOS
processes, a number of DACs provided voltage-output capability. These same processes
also allowed the integration of the voltage reference on-chip, thereby providing a more
complete solution.
Audio and video monolithic DACs began to appear in the mid 1980s. The
AD1856/AD1860 16-/18-bit audio DACs were targeted toward the emerging compact
disk (CD) player market. The AD9700 DAC, introduced in 1984, was the first monolithic
IC DAC designed for raster scan graphics applications which allowed the sync, blanking,
10% white, and reference white levels to be set with separate internal switches. This
allowed the full 8-bit range to be dedicated to the active video region. The AD9700 was
based on the HDG-series of hybrid DACs previously introduced in 1980. The AD9700
1.50
DATA CONVERTER HISTORY
1.4 DATA CONVERTERS OF THE 1980S
would later be replaced by the ADV-series of CMOS video RAM-DACs which included
on-chip color palette memory as well as the basic DAC function.
Monolithic ADCs of the 1980s
The two-chip AD574, introduced in 1978, was well on its way to becoming an industry
standard converter by the time of the introduction of the single-chip AD574 12-bit,
35-µs SAR ADC in 1985. Since that time, this converter did become an industry
standard, and is still in production today.
Figure 1.44 lists some of the important monolithic ADCs introduced during the 1980s. It
is interesting to note the emergence of the sampling monolithic ADC starting in the mid1980s. The addition of sample-and-holds, references, and buffer amplifiers was made
considerably easier with the addition of bipolar capability to the CMOS process
(LC2MOS and BiCMOS). Another trend which emerged was the addition of front-end
multiplexers to the basic ADC, as in the case of the 4-channel AD7582 in 1984, thereby
providing a more complete data acquisition solution.
Although the basic Σ-∆ ADC architecture had been well known since the 1950s and
1960s, the first actual commercial offering of a monolithic Σ-∆ ADC was in 1988 by
Crystal Semiconductor (CSZ5316). The device had 16-bit resolution and an effective
throughput rate of 20 kSPS, making it suitable for voiceband digitization. The digital
audio market (both professional and consumer) generated a demand for Σ-∆ ADCs with
higher throughput rates and greater resolution; and the precision measurement market
required 20+ bit resolution, although at much lower throughput rates. Both these needs
would be addressed during the 1990s by an explosion of ADCs and DACs using the Σ-∆
architecture.
AD574, 12-bit, 35µs, industry-standard, single-chip ADC, 1985
AD673, 8-bit, complete ADC, 1983
AD7582, 4-channel muxed input 12-bit CMOS ADC, 1984
AD670, 8-bit, 10µs ADCPORT, 1984
AD7820, 8-bit, 1.36µs half-flash, sampling ADC, 1985
AD7572, 12-bit, 5µs SAR LC2MOS ADC with reference, 1986
AD7575, 8-bit, 5µs SAR LC2MOS sampling ADC, 1986
AD7579, 10-bit, 50kSPS, LC2MOS SAR sampling ADC with AC specs, 1987
AD7821, 8-bit, 1MSPS half-flash sampling ADC with AC specs, 1988
AD674, 12-bit,15µs ADC, 1988
AD7870, 12-bit, 100kSPS LC2MOS SAR sampling ADC with AC specs, 1989
AD7871, 14-bit, 83kSPS LC2MOS SAR sampling ADC with AC specs, 1989
First commercial 16-bit sigma-delta ADC, Crystal Semiconductor, 1988
Figure 1.44: Monolithic ADCs of the 1980s
1.51
ANALOG-DIGITAL CONVERSION
Monolithic Flash ADCs of the 1980s
As previously mentioned, the rapidly growing digital video market, coupled with the
introduction of the TRW TDC-1007J 8-bit, 30-MSPS flash ADC in 1979, spurred many
IC manufacturers to develop similar, but lower powered flash ADCs with resolutions
ranging from 4- to 10-bits, and sampling rates as high as high as 500 MSPS. Most, but
certainly not all, are listed in Figure 1.45 which covers the period from approximately
1979 to 1990, the peak years for flash converters.
TDC1007J, 8-bit, 30MSPS, (TRW, LSI), 1979
TDC1016J, 6-bit, 30MSPS, (TRW, LSI), 1979
AM6688, 4-bit, 100MSPS, (AMD), 1979
SDA6020, 6-bit, 50MSPS, (Siemens) 1980
TLM1070, 7-bit, 20MSPS, CMOS (Telmos), 1982
MP7684, 8-bit, 20MSPS, CMOS (Micro Power), 1983
TDC1048, 8-bit, 30MSPS, (TRW, LSI), 1983
AD9000, 6-bit, 75MSPS, 1984
AD9002, 8-bit, 150MSPS, 1987
AD770, 8-bit, 200MSPS, 1988
AD9048, 8-bit, 35MSPS, 1988
AD9006/AD9016, 6-bit, 500MSPS, 1989
AD9012, 8-bit, 100MSPS, TTL, 1988
AD9028/AD9038 8-bit, 300MSPS, 1989
AD9020, 10-bit, 60MSPS, 1990
AD9058, Dual 8-bit, 50MSPS, 1990
AD9060, 10-bit, 75MSPS, 1990
Figure 1.45: Monolithic Flash ADCs of the 1980s
Both bipolar and CMOS technology was used to produce these devices, with the CMOS
converters offering lower power, but at the expense of somewhat inferior performance—
especially in the early offerings. The chief problem with the early CMOS flash converters
were error codes known as "sparkle codes" produced by comparator metastability (see
Data Converter Architectures, Chapter 3 for details). The bipolar process comparator
designs were less susceptible because they typically had much higher regenerative gain.
Today, these metastability problems have largely been overcome in CMOS devices
designed on submicron processes; however, they still may occur if care is not taken in the
design.
Although bipolar and CMOS flash converter designs were used in most of the 6- to 10-bit
video ADCs of the 1980s, the lower power subranging and pipelined architectures
became prevalent in the 1990s, as ADC designers gained experience with faster CMOS
and BiCMOS processes. Today, the flash converter architecture is widely used as a
building block within pipelined ADCs. There are, however, a few GaAs flash converters
1.52
DATA CONVERTER HISTORY
1.4 DATA CONVERTERS OF THE 1980S
of 6- or 8-bits resolutions which serve the relatively niche markets requiring sampling
rates of 1 GSPS or greater.
Hybrid and Modular DACs and ADCs of the 1980s
The demand for hybrid and modular DACs and ADCs peaked in the 1980s, primarily
because of the 3- to 5-year lead time hybrids and modules held compared to single-chip
monolithic converters with equivalent performance. In addition, the large number of flash
converters and other components served as building blocks for higher resolution
subranging ADCs. Some examples of important hybrid and modular data converters
introduced in the 1980s are shown in Figure 1.46.
Hybrids:
HDS-1240E, 12-bit, 40ns ECL DAC, 1980
HDG-Series, 4-, 6-, 8-bit, 5ns video ECL DACs, 1980
HAS-1409, 14-bit, 1.25MSPS sampling ADC, 1983
HAS-1201, 12-bit, 1MSPS sampling ADC, 1984
AD376, 16-bit, 20µs SAR ADC, 1985
AD1332, 12-bit, 125kSPS sampling ADC with 32-word FIFO, 1988
AD9003, 12-bit, 1MSPS sampling ADC, 1988
AD9005, 12-bit, 10MSPS sampling ADC, 1988
AD1377, 16-bit, 10µs SAR ADC, 1989
Modules:
ADC1140, 16-bit, 35µs SAR ADC, 1982
CAV-1220, 12-bit, 20MSPS sampling ADC, 1986
CAV-1040, 10-bit, 40MSPS sampling ADC, 1986
AD1175, 22-bit integrating ADC, 1987
Figure 1.46: High Performance Hybrid and Modular
DACs and ADCs of the 1980s
An interesting thick film hybrid family, introduced by Analog Devices in 1980, was the
HDG-series 4-, 6-, and 8-bit video ECL DACs. Designed for raster scan RGB graphics
displays, these DACs had an 8-bit settling time of approximately 5 ns. In addition to fast
settling, these were among the first video DACs to allow the sync, blanking, 10% white,
and reference white levels to be set with separate internal switches. This allowed the full
8-bit DAC range to be dedicated to the active video region. The HDG-series was a
precursor to the fully monolithic CMOS video DACs and RAM-DACs which were
introduced later in the 1980s (ADV-series in Figure 1.43).
High performance hybrid subranging ADCs also appeared during the 1980s, most
utilizing high speed flash converters as internal building blocks. Most were sampling
1.53
ANALOG-DIGITAL CONVERSION
devices, complete with ac specifications, culminating in the 12-bit 10-MSPS AD9005
introduced in 1988.
Hybrids of the 1980s also achieved resolutions not yet attainable in monolithic
technology, such as the AD1377 16-bit, 10-µs SAR ADC introduced in 1989.
Also worthy of notice were several modules introduced in the 1980s which also pushed
the speed and resolution envelope. The CAV-1220 12-bit, 20-MSPS ADC and the
CAV-1040 10-bit, 40-MSPS ADC introduced in 1986 set new standards in high-speed
dynamic range performance, while the AD1175 22-bit integrating ADC set the standard
for high resolution when it was introduced in 1987.
1.54
DATA CONVERTER HISTORY
1.5 DATA CONVERTERS OF THE 1990S
SECTION 1.5: DATA CONVERTERS OF THE 1990s
Walt Kester
Introduction
The markets influencing data converters in the 1990s were even more diverse and
demanding than those of the 1980s. Some of the major applications were industrial
process control, measurement, instrumentation, medical imaging, audio, video, and
computer graphics. In addition, communications became an even bigger driving force for
low-cost, low-power, high-performance data converters in modems, cell phone handsets,
and wireless infrastructure (basestations).
Other trends were the emphasis on lower power and single-supply voltages for portable
battery-powered applications. While the reduced supply voltages were compatible with
the higher speed, lower voltage processes, the reduced signal range and headroom made
the converter designs more sensitive to noise. Packaging trends also changed in the 1990s
from the traditional DIP to smaller surface mount packages suitable for high-volume
automatic mass-assembly manufacturing techniques. These included both leaded types
and non-leaded types such as ball grid array (BGA) and chip-scale packages (CSP).
Even in the general purpose data converter market, there was a demand for more analog
and digital functionality, such as putting an entire data acquisition system on a chip
including the input multiplexer, programmable gain amplifier (PGA), sample-and-hold,
and the ADC function. Many applications required both the ADC and DAC function; and
this led to the integration of both on a single chip, called a coder-decoder, or CODEC.
Specially designed analog front ends (AFEs) and mixed-signal front ends (MxFE™) were
included with the basic ADC function in applications such as CCD image processors and
IF sampling receivers.
In the TxDAC® family, digital functions such as interpolation filters and digital
modulators were integrated with a high-speed, low-distortion CMOS DAC core. Data
converter designers made more use of "core" designs to yield several products with
various options, such as serial or parallel output parts, etc. The TxDAC-series is a good
illustration of the application of this concept, where a variety of resolutions, update rates,
and internal digital processing are offered across a broad number of individual products,
all of which use essentially the same DAC core.
Because of the increase in frequency-domain signal processing applications, there was
even greater emphasis on dynamic range and ac performance in practically all data
converters, and a large number of monolithic sampling ADCs were introduced to meet
the demand. The pipelined subranging architecture virtually replaced the higher power
flash ADCs of the 1980s, and some of the key ac specifications were SNR, SINAD,
ENOB, and SFDR.
In the 1990s, CMOS became the process-of-choice for general-purpose data converters,
with BiCMOS reserved for the high-end devices. In a few cases, high-speed
complementary bipolar processes were utilized for ultra high performance data
converters. CMOS is also the ideal process for the Σ-∆ architecture, which became the
1.55
ANALOG-DIGITAL CONVERSION
topology of choice for ADCs and DACs used in voiceband and audio applications, as
well as in higher resolution, low frequency measurement converters.
A major process technology shift occurred in the 1990s, when parasitics ultimately
became the performance-limiting factor for high-speed chip-and-wire hybrid data
converters. The newer ICs, with their smaller feature size and reduced parasitics, allowed
them to achieve higher levels of performance than attainable in a chip-and-wire hybrid or
a module—a reversal of the situation that existed throughout the 1970s and most of the
1980s.
In the following sections, we will examine the data converter trends of the 1990s and
2000s, using a few example products as illustrations. It would be impossible to cover the
individual products in as much detail as we did for the 1970s and 1980s, simply because
of the large number of individual data converters introductions in the 1990s. Many of
these products are discussed in Chapter 8 of this book, Data Converter Applications.
Monolithic DACs of the 1990s
A significant trend in general purpose DACs of the 1990s was toward more functionality
in all areas, especially with respect to the input structure. DACs were designed
specifically to handle parallel, serial, and byte-wide loading, and the inputs were
generally double-buffered. The serial interface became popular for interfacing to
microprocessors and DSPs. In many cases, the same core DAC design was utilized to
provide these various options as separate products in the appropriate packages.
Obviously, this required the introduction of many individual DAC products to cover all
the desired options.
A variety of options existed with respect to the output structure also. Audio and video
DACs tended to use current outputs, while some of the more general-purpose DACs
offered either current or voltage output options.
The trend toward multiple DACs is illustrated by an early example of an octal DAC, the
AD7568 12-bit LC2MOS DAC shown in Figure 1.47, introduced in 1991. This DAC
utilized the popular multiplying architecture and provided current outputs designed to
drive an op amp connected as an I/V converter. Notice that the DAC is double buffered—
the input shift register accepts the serial input data and loads it into the appropriate input
latch, and asserting LDAC simultaneously latches the data into the eight parallel
individual DAC latches.
Consumer audio compact disk players spurred the market for low distortion 16+ bit
DACs in the late 1980s, and the first audio DACs were parallel linear DACs capable of
oversampling at rates of 8 times or 16 times the basic CD update rate of 44.1 kSPS.
Resolutions ranged from 16-bits with the early audio DACs to 18 and 20 bits with later
versions. For example, the AD1865 dual 18-bit stereo DAC was introduced in 1991 and
was capable of 16-times oversampling.
1.56
DATA CONVERTER HISTORY
1.5 DATA CONVERTERS OF THE 1990S
By the mid-1990s, the Σ-∆ architecture began to replace the parallel DACs in audio
applications. Sigma-delta offered much higher oversampling ratios, thereby relaxing
output filter requirements, as well as provided higher dynamic range with lower
distortion. Some early offerings were the AD1857, AD1858, and AD1859 introduced in
1996. These DACs offered resolutions ranging from 16 bits to 20 bits, used serial
interfaces, and were single-supply devices.
Figure 1.47: AD7568 Octal 12-Bit LC2MOS MDAC, 1991
Although there were a few bipolar process high speed ECL DACs introduced in the early
1990s, such as the AD9712 12-bit, 100-MSPS DAC and the AD9720 10-bit, 400-MSPS
DAC, the majority of video and communications DACs were low-power, low-glitch,
low-distortion CMOS devices. The ADV-series of video CMOS RAM-DACs continued
to expand during the 1990s, and an entire family of 8-, 10-, 12-, 14-, and 16-bit transmitDACs (TxDAC®) for communications was started in 1996, with new introductions
continuing to this day (AD976x, AD977x, and AD978x-series).
Direct digital synthesis (DDS) systems on a single chip emerged in the 1990s, largely
because of the relative ease with which digital logic could be added to a high
performance CMOS DAC core. The first of these was the AD7008, 10-bit, 50-MSPS
DDS introduced in 1993 (see Figure 1.48). This was soon followed by additional
offerings, such as the 10-bit, 125-MSPS AD9850 in 1996. Later DDS systems added
phase and frequency modulation capability, on-chip clock multipliers, more resolution,
and update rates as high as 1 GHz.
1.57
ANALOG-DIGITAL CONVERSION
Figure 1.48: AD7008 10-Bit, 50-MSPS Complete CMOS DDS, 1993
The digital potentiometer, another highly popular component today, had its origins in
1989 with the release of the AD8800 TrimDAC®, the first in a series. The TrimDACs
were basically 8-bit voltage-output DACs designed and optimized to replace mechanical
potentiometers. The TrimDAC family became popular, and in 1995, the first DigiPOTs®
were introduced. The basic concept behind the digital pot was simply to use a CMOS
"string DAC" as a variable resistor. The AD8402 2-channel (8-bits), AD8403 4-channel
(8-bits) were the first offerings in 1995. Since then the product line has been expanded to
include many more products including some with non-volatile memory (AD51xx and
AD52xx series).
Figure 1.49 summarizes some of the key DAC developments during the 1990s.
Multiple DACs: AD7568, 12-bit octal single +5V supply CMOS MDAC, 1991
Audio DACs
Parallel 8×, 16× oversampling, early 1990s
Sigma-delta, starting with the AD1857, AD1858, AD1859 in 1996
Video RAM-DACs-continued expansion of product line
Transmit DACs (TxDACs) for communications, 1996
Direct Digital Synthesis (DDS) Systems, AD9008, 1993
TrimDACs, 1989
Digital potentiometers, 1995
Figure 1.49: Summary: Monolithic DACs of the 1990s
1.58
DATA CONVERTER HISTORY
1.5 DATA CONVERTERS OF THE 1990S
Monolithic ADCs of the 1990s
During the decade of the 1990s, monolithic ADC performance overtook that of modular
and hybrid converters, primarily because of the greatly reduced parasitics in the new IC
processes. The AD1674 12-bit, 100-kSPS sampling SAR ADC introduced in 1990 was
pin-compatible with the industry-standard AD574 introduced a decade earlier. A
simplified block diagram of the AD1674 is shown in Figure 1.50. This represented a
trend toward sampling ADCs that was to continue throughout the 1990s, because of the
increased interest in the digital processing of ac signals.
Figure 1.50: AD1674, 12-Bit, 100-kSPS Sampling ADC
(AD574 Pin-Compatible), 1990
Progress was also being made in CMOS sampling ADCs, as illustrated by the
introduction of the AD7880 12-bit, 66-kSPS ADC in 1990. A simplified block diagram is
shown in Figure 1.51. Although the AD7880 required an external reference, it still
represented a breakthrough because of its low power (25 mW) and single +5-V operation.
1.59
ANALOG-DIGITAL CONVERSION
Figure 1.51: AD7880 12-Bit, 66-kSPS, Single +5 V, LC2MOS
Sampling SAR ADC, 1990
In 1992, the 12-bit, 1.25-MSPS AD1671 BiCMOS sampling ADC was introduced. The
basic subranging pipelined architecture utilized in the AD1671 was based on a previous
non-sampling version, the 2-MSPS AD671, which had been introduced in 1990. A
simplified block diagram of the AD1671 is shown in Figure 1.52.
Figure 1.52: AD1671 12-Bit, 1.25-MSPS Sampling BiCMOS ADC, 1992
1.60
DATA CONVERTER HISTORY
1.5 DATA CONVERTERS OF THE 1990S
A significant breakthrough in speed and performance was also achieved in 1992 with the
release of the AD872 12-bit, 10-MSPS BiCMOS sampling ADC, shown in Figure 1.53.
This ADC also used the pipelined architecture with error correction.
Figure 1.53: AD872 12-Bit 10-MSPS BiCMOS Sampling ADC, 1992
In 1996, three single supply (+5 V) CMOS ADCs were released using an architecture
similar to the 12-bit AD872: the AD9220 (10 MSPS), the AD9221 (1 MSPS), and
AD9223 (3 MSPS). All three parts utilized the same basic design, with the operating
current scaled to yield the three sampling rate options. Power dissipation for the three
devices is 250 mW (AD9220, 10 MSPS), 60 mW (AD9221, 1 MSPS), and 100 mW
(AD9223, 3 MSPS).
A true breakthrough in wide dynamic range IF-sampling ADCs occurred in 1995 with the
release of the 12-bit, 41-MSPS AD9042. A functional diagram of the AD9042 is shown
in Figure 1.54. This converter was the first to achieve greater than 80-dB SFDR for
signals over a 20-MHz Nyquist bandwidth. It was fabricated on the Analog Devices'
XFCB high-speed complementary bipolar process.
Figure 1.54: AD9042 12-Bit 41-MSPS XFCB, Sampling ADC,1995
1.61
ANALOG-DIGITAL CONVERSION
The concept of a complete data acquisition system on a chip saw fruition in the 1990s
with the introduction of the AD789x-series of LC2MOS SAR single-supply (+5 V) ADCs
in 1993. These parts offer up to 8 channels of multiplexed inputs and sampling rates of
100 kSPS to 600 kSPS. In order to accommodate the more traditional industry-standard
bipolar inputs of ±10 V and ±5 V, the series provides input thin film resistor
attenuators/level shifters to match the input range of the internal SAR ADCs.
A significant technology change occurred in the 1990s, when the high-power 8- , 9-, and
10-bit flash converters of the 1980s were gradually replaced by lower power pipelined
and folding architectures. This was typified by the introduction of the AD9054 8-bit,
200-MSPS ADC in 1997. The AD9054 utilized a unique architecture consisting of five
folding stages followed by a 3-bit parallel flash stage.
CMOS Σ-∆ ADCs became the architecture of choice for measurement, voiceband, and
audio ADCs beginning in the early 1990s. The AD7001 was the first GSM baseband
converter and was introduced in 1990. Sigma-delta was utilized in many other voiceband
and audio converters as well as high-resolution measurement ADCs. The AD771x-family
of 24-bit measurement converters was introduced in 1992.
These converters include on-chip multiplexers and PGAs and are designed for direct
interfacing to many sensors, such as thermocouples, bridges, RTDs, etc. A significant
product in the family is the AD7730, introduced in 1997, which allows a load cell output
with a 10-mV fullscale voltage to be digitized to over 80,000 noise-free codes (16.5 bits).
A simplified block diagram of the AD7730 is shown in Figure 1.55.
Figure 1.55: AD7730 24-Bit Signal-Conditioning Σ-∆ ADC, 1997
1.62
DATA CONVERTER HISTORY
1.5 DATA CONVERTERS OF THE 1990S
Another application of Σ-∆ technology appeared in the late 1990s, with the introduction
of the ADE775x-series of energy metering ICs. These parts measure the instantaneous
current and voltage of the power mains and calculate power usage, thereby replacing
mechanical devices.
A summary of key monolithic ADCs of the 1990s is given in Figure 1.56.
AD1674, 12-bit, 100kSPS, sampling ADC, AD574A pin-compatible, 1990
AD7880, 12-bit, 66kSPS LC2MOS sampling ADC, 1990
AD7001, CMOS GSM baseband converter, 1990
AD771x-series 24-bit Σ-∆ measurement ADCs, 1992
AD1671, 12-bit, 1.25MSPS BiCMOS sampling ADC, 1992
AD872, 12-bit, 10MSPS BiCMOS sampling ADC, 1992
AD9220/AD9221/AD9223, 12-bit, 10/1/3 MSPS, CMOS sampling ADCs, 1996
AD9042, 12-bit, 41MSPS sampling ADC, 80dB SFDR, 1995
AD7730, 24-bit bridge transducer measurement ADC, 1997
AD9054, 8-bit, 200MSPS sampling ADC, 1997
ADuC812 MicroConverter® (precision ADCs, DACs, 8051-core, flash
memory, 1999
Figure 1.56: Summary: Monolithic ADCs of the 1990s
Because of the ease with which digital functionality can be added to BiCMOS or CMOS
ADCs and DACs, there have been an increasing number of highly integrated applicationspecific integrated circuits during the 1990s and continuing to this day. A few of the
important areas served by these chips are listed in Figure 1.57. Most have already been
mentioned in this chapter, and are covered throughout this book, especially in Chapter 8.
An important development occurred in 1999 with the introduction of the first Analog
Devices' precision analog microcontroller, the ADuC812 MicroConverter®. The
MicroConverter includes not only precision signal conditioning circuitry (ADCs, DACs,
multiplexers, etc.) but also flash memory and an 8051-based microprocessor core. Later
MicroConverter products have included higher resolution Σ-∆ ADCs (see Chapter 8 of
this book). This level of integration represents an optimum solution for many generalpurpose sensor conditioning and signal processing applications.
1.63
ANALOG-DIGITAL CONVERSION
Voiceband codecs
Audio codecs
AC'97 SoundMAX® computer audio codecs
I/O Ports
Mixed signal front ends: modems, communications, CCD imaging, flat
panel displays
Transmit and receive signal processors
Direct conversion (Othello® radio) chipsets
Direct digital synthesis
TxDACs® with interpolation, filters, digital quadrature modulation, etc.
Cell phone chipsets
Energy metering
Video RAM-DACs
Video encoders/decoders, codecs
Touchscreen digitizers
MicroConverter products (high performance ADCs, DACs, + 8051 µP
core and flash memory)
Figure 1.57: Summary: Integrated Functions of the 1990s
Hybrid and Modular DACs and ADCs of the 1990s
Although monolithic data converter solutions largely replaced the chip-and-wire hybrids
and modules of the 1970s and 1980s, there were a few significant introductions early in
the 1990s. The AD9014 14-bit, 10-MSPS modular ADC introduced in 1990 represented a
major breakthrough in dynamic performance. The part achieved 90-dB SFDR over the
Nyquist bandwidth, and utilized a number of proprietary monolithic building blocks
which were later used in the fully integrated design of the AD9042 12-bit, 41-MSPS
ADC introduced in 1995.
The AD1382 and AD1385 hybrid 16-bit, 500-kSPS sampling ADCs, introduced in 1992,
represented state-of-the-art performance at the time, and the AD1385 was one of the first
ADCs utilizing autocalibration to maintain its linearity.
Later in the 1990s, multichip module (MCM) technology became an excellent alternative
to the expensive modules and chip-and-wire hybrids. Lower-cost packaging techniques
allow high performance monolithic ADCs, such as the AD9042, to be packaged as duals
along with front-end conditioning circuits. For instance the AD10242 introduced in 1996
(dual 12-bit, 41-MSPS AD9042s) offered an attractive cost-effective solution in
applications requiring dual high-performance 12-bit ADCs and more functionality in the
analog front end.
1.64
DATA CONVERTER HISTORY
1.6 DATA CONVERTERS OF THE 2000S
SECTION 1.6: DATA CONVERTERS OF THE 2000s
Walt Kester
The data converter trends started in the 1990s shown in Figures 1.56 and 1.57 have
continued into the 2000s. Power dissipation has dropped, and along with it, power supply
voltages. Supplies of 5-V, 3.3-V, 2.5-V, and 1.8-V parts have followed as CMOS line
spacings shrank to 0.6 µm, 0.35 µm, 0.25µm, and 0.18 µm. Smaller surface mount and
chip-scale packages have also emerged as the modern replacement for the nearly obsolete
DIP packages of the 1970s and 1980s.
Although the trend toward more highly integrated functions continues, data converter
manufacturers are realizing that "smart partitioning" can offer a higher performance and
more cost-effective solution than simply always adopting a "system-on-a-chip"
philosophy. This topic is explored in much more detail in Chapter 4 of this book, Data
Converter Process Technology.
The Analog Devices' portfolio of 16- and 18-bit SAR sampling ADCs has grown to over
30 models, including the latest offerings in the breakthrough high-resolution Pulsar®
series. For example, the AD7664 16-bit, 570-kSPS ADC was introduced in 2000, the
AD7677 16-bit, 1-MSPS ADC in 2001, and the AD7674 18-bit, 800-kSPS ADC in 2003,
and the AD7621 18-bit, 3-MSPS ADC in 2003.
In the 2000s, general purpose multiple DAC offerings expanded to include 16-channels
(AD5390, AD5391), 32-channels (AD5382, AD5383) and 40-channels (AD5380,
AD5381). High speed DACs reached 1-GSPS update rates with the AD9858 10-bit DDS
system.
Turning to IF sampling data converters, the AD6645 14-bit 80-/105-MSPS ADC was
introduced in 2000, and the AD9430 12-bit, 210-MSPS ADC in 2002. Both converters
represent breakthroughs in sampling rate and dynamic range.
Significant multichip module (MCM) introductions were the AD10678 16-bit,
65-/80-/105-MSPS ADC introduced in 2003, and the AD12400 12-bit, 400-MSPS ADC
also introduced in 2003. Both devices use high performance IC sampling ADCs as
building blocks followed by proprietary digital post processing.
In 2002 and 2003, the ADuC-series of MicroConverter® products was expanded to
include 16- and 24-bit on-chip Σ-∆ ADCs. In addition, the SAR-based MicroConverter
product line was expanded. Some of the future MicroConverter products to be introduced
starting in 2004 will utilize the highly popular ARM7®-microcontroller core.
These data converter highlights of the 2000s are summarized in Figure 1.58.
There are, of course, many other things that could be said about the history of data
converters. Many other examples of modern ADCs and DACs are given throughout the
remaining chapters of this book. Looking to the future, we can expect many new
breakthroughs, not only in sheer performance, but in levels of integration.
1.65
ANALOG-DIGITAL CONVERSION
Continued expansion of Analog Front Ends (AFEs) and Multiplexed
Front Ends (MxFEs®)
16-, 18-bit Pulsar® series of switched capacitor SAR sampling ADCs
AD7674, 18-bit, 800kSPS ADC, 2003
AD7621, 16-bit, 3MSPS ADC, 2003
Multiple DACs: 16-channel (AD5390, AD5391), 32-channel
(AD5382/AD5383), 40-channel (AD5380/AD5381)
IF-Sampling ADCs
AD6645 14-bit, 105MSPS ADC, 2000
AD9430 12-bit, 210MSPS ADC, 2002
Multichip modules (MCMs):
AD12400, 12-bit, 400MSPS ADC, 2003
AD10678, 16-bit, 65/80/105 MSPS ADC, 2003
AD9858 10-bit, 1GSPS DDS system, 2003
16-/24-bit Σ-∆ MicroConverter products, 2002, 2003
ARM7-based MicroConverter products, 2004
Figure 1.58: Data Converter Highlights of the 2000s
1.66
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