Chapter 3: Data Converter Architectures

Chapter 3: Data Converter Architectures
DATA CONVERTER ARCHITECTURES
ANALOG-DIGITAL CONVERSION
1. Data Converter History
2. Fundamentals of Sampled Data Systems
3. Data Converter Architectures
3.1 DAC Architectures
3.2 ADC Architectures
3.3 Sigma-Delta Converters
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 ARCHITECTURES
3.1 DAC ARCHITECTURES
CHAPTER 3
DATA CONVERTER ARCHITECTURES
SECTION 3.1: DAC ARCHITECTURES
James Bryant, Walt Kester
Introduction
Those unfamiliar with DACs regard them simply as devices with digital input and analog
output. But the analog output depends on the presence of that analog input known as the
reference, and the accuracy of the reference is almost always the limiting factor on the
absolute accuracy of a DAC. We shall consider the various architectures of DACs, and
the forms which the reference may take, later in this section.
Some DACs use external references (see Figure 3.1) and have a reference input terminal,
while others have an output from an internal reference. The simplest DACs, of course,
have neither—the reference is on the DAC chip and has no external connections.
VDD
(ANALOG)
REFERENCE
INPUT
VREF
DIGITAL
INPUT
DAC
VSS
ANALOG
OUTPUT
GROUND
(MAY BE INTERNALLY
CONNECTED TO VSS)
Figure 3.1: Basic DAC with External Reference
If a DAC has an internal reference, the overall accuracy of the DAC is specified when
using that reference. If such a DAC is used with a perfectly accurate external reference,
its absolute accuracy may actually be worse than when it is operated with its own internal
reference. This is because it is trimmed for absolute accuracy when working with its own
3.1
ANALOG-DIGITAL CONVERSION
actual reference voltage, not with the nominal value. Twenty years ago it was common
for converter references to have accuracies as poor as ±5% since these references were
trimmed for low temperature coefficient rather than absolute accuracy, and the
inaccuracy of the reference was compensated in the gain trim of the DAC itself. Today
the problem is much less severe, but it is still important to check for possible loss of
absolute accuracy when using an external reference with a DAC which has a built-in one.
DACs which have reference terminals must, of course, specify their behavior and
parameters. If there is a reference input, the first specification will be the reference input
voltage—and of course this has two values, the absolute maximum rating, and the range
of voltages over which the DAC performs correctly.
Most DACs require that their reference voltage be within quite a narrow range whose
maximum value is less than or equal to the DAC's VDD, but some DACs, called
multiplying DACs (or "MDACs"), will work over a wide range of reference voltages that
may go well outside their power supplies. The AD7943 multiplying DAC, for example,
has an absolute maximum rating on its VDD terminal of +6 V but a rating of ±15 V on its
reference input, and it works perfectly well with positive, negative or ac references. (The
generally-accepted definition of an MDAC is that its reference voltage range includes
zero. But some authorities prefer a looser definition, "a DAC with a reference voltage
range greater than 5:1." In this chapter we shall use the term "semi-multiplying DAC" for
devices of this type.) MDACs that work with ac references have a "reference bandwidth"
specification which defines the maximum practical frequency at the reference input.
The reference input terminal of a DAC may be buffered as shown in Figure 3.2, in which
case it has input impedance (usually high) and bias current (usually low) specifications,
or it may connect directly to the DAC. In this case the input impedance specification may
become more complicated since some DAC structures have an input impedance that
varies substantially with the digital code applied to the DAC. In such cases the (usually
simplified) structure of the DAC is shown on the data sheet, and the nominal values of
resistance are given. Where the reference input impedance does not vary with code, the
input impedance should be specified.
Surprisingly for such an accurate circuit, the reference input impedance of a resistive
DAC network is rarely very well-defined. For example, the AD7943 has a nominal input
impedance of 9 kΩ, but the data sheet limits are 6 kΩ and 12 kΩ, a variation of ±33%.
The reasons for this are discussed later in this book (see Chapter 4). In addition, the
reference input impedance is code-dependent for the voltage-mode R-2R architecture.
Where a DAC has a reference output terminal, it will carry a defined reference voltage,
with a specified accuracy. There may also be specifications of temperature coefficient
and long-term stability.
The reference output (if available) may be buffered or unbuffered. If it is buffered the
maximum output current will be specified. In general such a buffer will have a
unidirectional output stage which sources current but does not allow current to flow into
the output terminal. If the buffer does have a push-pull output stage, the output current
will probably be defined as ±(SOME VALUE) mA. If the reference output is unbuffered,
3.2
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
the output impedance may be specified, or the data sheet may simply advise the use of a
high input impedance external buffer.
Any, all or none of these pins may be
connected internally or brought off the chip.
VDD
VREF
BUFFER
AMP
VOLTAGE
REFERENCE
DIGITAL
INPUT
DAC
VSS
ANALOG
OUTPUT
GROUND
(MAY BE INTERNALLY
CONNECTED TO VSS)
Figure 3.2: DAC with Reference and Buffer
DAC Output Considerations
The output of a DAC may be a voltage or a current. In either case it may be important to
know the output impedance. If the voltage output is buffered, the output impedance will
be low. Both current outputs and unbuffered voltage outputs will be high(er) impedance
and may well have a reactive component specified as well as a purely resistive one. Some
DAC architectures have output structures where the output impedance is a function of the
digital code on the DAC—this should be clearly noted on the data sheet.
In theory, current outputs should be connected to zero ohms at ground potential. In real
life they will work with non-zero impedances and voltages. Just how much deviation they
will tolerate is defined under the heading "compliance" and this specification should be
heeded when terminating current-output DACs.
Basic DAC Structures
It is reasonable to consider a changeover switch (a single-pole, double-throw, SPDT
switch), switching an output between a reference and ground or between equal positive
and negative reference voltages, as a 1-bit DAC as shown in Figure 3.3. Such a simple
device is a component of many more complex DAC structures, and is used, with
oversampling, as the basic analog component in many of the sigma-delta DACs we shall
discuss later. Nevertheless it is a little too simple to require detailed discussion, and it is
more rewarding to consider more complex structures.
3.3
ANALOG-DIGITAL CONVERSION
VREF
OUTPUT
Figure 3.3: 1-Bit DAC: Changeover Switch (Single-Pole, Double Throw, SPDT)
The Kelvin Divider (String DAC)
The simplest DAC structure of all, after the changeover switch mentioned above, is the
Kelvin divider or string DAC as shown in Figure 3.4. An N-bit version of this DAC
simply consists of 2N equal resistors in series and 2N switches (usually CMOS), one
between each node of the chain and the output. The output is taken from the appropriate
tap by closing just one of the switches (there is some slight digital complexity involved in
decoding to 1 of 2N switches from N-bit data, but digital circuitry is cheap). The origins
of this DAC date back to Lord Kelvin in the mid-1800s, and it was first implemented
using resistors and relays, and later with vacuum tubes in the 1920s (See References 1, 2,
3).
VREF
R
R
3-TO-8
DECODER
R
8
R
TO
SWITCHES
R
3-BIT
DIGITAL
INPUT
ANALOG
OUTPUT
R
R
R
CIRCA 1920
SWITCHES WERE
RELAYS OR VACUUM TUBES
Figure 3.4: Simplest Voltage-Output Thermometer DAC:
The Kelvin Divider ("String DAC")
This architecture is simple, has a voltage output (but a code-dependent output impedance)
and is inherently monotonic—even if a resistor is accidentally short-circuited, output n
3.4
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
cannot exceed output n + 1. It is linear if all the resistors are equal, but may be made
deliberately nonlinear if a nonlinear DAC is required. Since only two switches operate
during a transition, it is a low-glitch architecture. Also, the switching glitch is not codedependent, making it ideal for low distortion applications. Because the glitch is constant
regardless of the code transition, the frequency content of the glitch is at the DAC update
rate and its harmonics—not at the harmonics of the fundamental DAC output frequency.
The major drawback of the thermometer DAC is the large number of resistors and
switches required for high resolution, and as a result it was not commonly used as a
simple DAC architecture until the recent advent of very small IC feature sizes made it
very practical for low and medium resolution DACs. Today the architecture is quite
widely used in simple DACs, such as digital potentiometers and, as we shall see later, it
is also used as a component in more complex high resolution DAC structures.
As we have already mentioned, the output of a DAC for an all 1s code is 1 LSB below
the reference, so a string DAC intended for use as a general purpose DAC has a resistor
between the reference terminal and the first switch as shown in Figure 3.4.
In an ideal potentiometer, on the other hand, all 0s and all 1s codes should connect the
variable tap to one or other end of the string of resistors. So a digital potentiometer, while
basically the same as a general purpose string DAC, has one fewer resistor, and neither
end of the string has any other internal connection. A simple digital potentiometer is
shown in Figure 3.5.
TERMINAL A
3-TO-8
DECODER
R
R
8
R
TO
SWITCHES
3-BIT
DIGITAL
INPUT
TAP
R
R
R
R
TERMINAL B
Figure 3.5: A Slight Modification to a String DAC Yields a "Digital Potentiometer"
The simplest digital potentiometers are no more complex than this, and none of the
potentiometer terminals may be at a potential outside the 5-V or 3-V logic supply. But
others have more complex decoders with level shifters and additional high voltage supply
terminals, so that while the logic control levels are low (3 V or 5 V), the potentiometer
terminals have a much greater range—up to ±15 V in some cases. Digital potentiometers
3.5
ANALOG-DIGITAL CONVERSION
frequently incorporate nonvolatile logic so that their settings are retained when they are
turned off.
It is evident that string DACs have a large number of resistors (2N for an N-bit DAC as
we have already seen). It is not practical to trim every resistor in a string DAC to obtain
perfect DNL and INL, partly because they are too many, and partly because they are too
small to trim, and mainly because it's too costly.
But if required, it is still possible to trim the INL of a string DAC. The method is shown
in Figure 3.6—a second string of four equal resistors is connected in parallel with the
main string. These resistors are made physically large enough to laser trim. The three
internal nodes of this string are connected by buffer amplifiers to the ¼, ½, and ¾ points
of the main string. The trimmable string is adjusted so that these points on the main string
are at the correct potentials. Typically, INL can be reduced by a factor of four by this
technique, at quite a small cost in complexity. However, in many modern string DACs
trimming is not required, because the resistors are well-matched, and the current drawn
by the CMOS switches is negligible.
INL
00000
11111
MAIN DAC - No trim components present
INL
00000
01000
10000
11000 11111
MAIN DAC - Trimmed
Figure 3.6: Trimming the INL of a String DAC (If Required)
Thermometer (Fully-Decoded) DACs
There is a current-output DAC analogous to a string DAC that consists of 2N–1
switchable current sources (which may be resistors and a voltage reference or may be
active current sources) connected to an output terminal, which must be at, or close to,
ground. Figure 3.7 shows a thermometer DAC which uses resistors connected to a
reference voltage to generate the currents.
3.6
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
VREF
R
R
R
R
R
R
R
3-TO-7
DECODER
7
CURRENT
OUTPUT INTO
VIRTUAL
GROUND
(USUALLY AN
OP-AMP I-V
CONVERTER)
TO
SWITCHES
3-BIT
DIGITAL
INPUT
Figure 3.7: The Simplest Current-Output Thermometer (Fully-Decoded) DAC
If active current sources are used as shown in Figure 3.8, the output may have more
compliance, and a resistive load used to develop an output voltage. The load resistor must
be chosen so that at maximum output current the output terminal remains within its rated
compliance voltage.
I
I
I
I
I
I
I
3-TO-7
DECODER
7
TO
SWITCHES
CURRENT
OUTPUT
MAY HAVE
COMPLIANCE
OF 1 OR 2 V
3-BIT
DIGITAL
INPUT
Figure 3.8: Current Sources Improve the Basic
Current-Output Thermometer DAC
Once a current in a thermometer DAC is switched into the circuit by increasing the
digital code, any further increases do not switch it out again. The structure is thus
inherently monotonic, irrespective of inaccuracies in the currents. Again, like the Kelvin
divider only the advent of high density IC processes has made this architecture practical
for general purpose medium resolution DACs, although a slightly more complex
version—shown in the next diagram—is quite widely used in high speed applications.
3.7
ANALOG-DIGITAL CONVERSION
Unlike the Kelvin divider, this type of current-mode DAC does not have a unique name,
although both types may be referred to as fully-decoded DACs or thermometer DACs.
A DAC where the currents are switched between two output lines—one of which is often
grounded, but may, in the more general case, be used as the inverted output—is more
suitable for high speed applications because switching a current between two outputs is
far less disruptive, and so causes a far lower glitch than simply switching a current on and
off. This architecture is shown in Figure 3.9.
A
C
B
I
I
D
I
E
I
F
I
G
I
I
3-TO-7
DECODER
7
OUTPUT
3-BIT
DIGITAL
INPUT
TO
SWITCHES
CURRENT
OUTPUTS
MAY HAVE
COMPLIANCE
OF 1 OR 2 V
OUTPUT
Figure 3.9: High Speed Thermometer DAC with
Complementary Current Outputs
But the settling time of this DAC still varies with initial and final code, giving rise to
intersymbol interference (ISI). This can be addressed with even more complex switching
where the output current is returned to zero before going to its next value. Note that
although the current in the output is returned to zero it is not "turned off"—the current is
dumped when it is not being used, rather than being switched on and off. The techniques
involved are too complex to discuss in detail here but can be found in Reference 4.
In the normal (linear) version of this DAC, all the currents are nominally equal. Where it
is used for high speed reconstruction, its linearity can be improved by dynamically
changing the order in which the currents are switched by ascending code. Instead of code
001 always turning on current A; code 010 always turning on currents A & B, code 011
always turning on currents A, B & C; etc. the order of turn-on relative to ascending code
changes for each new data point. This can be done quite easily with a little extra logic in
the decoder. The simplest way of achieving it is with a counter which increments with
each clock cycle so that the order advances: ABCDEFG, BCDEFGA, CDEFGAB, etc.,
but this algorithm may give rise to spurious tones in the DAC output. A better approach
is to set a new pseudo-random order on each clock cycle—this requires a little more
logic, but, as we have pointed out, even complex logic is now very cheap and easily
implemented on CMOS processes. There are other, even more complex, techniques
which involve using the data itself to select bits and thus turn current mismatch into
3.8
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
shaped noise. Again they are too complex for a book of this sort. (See References 4 and 5
for a more detailed discussion).
Binary-Weighted DACs
One of the earliest reference to an electro-mechanical binary-weighted DAC can be found
in Paul M. Rainey's 1921 (filing date) patent for a PCM-based facsimile transmission
system (Reference 6). This system is discussed in more detail in Chapter 1 of this book,
and the 5-bit reconstruction DAC is shown in Figure 3.10.
R
DIGITAL
INPUTS
INCANDESCENT
LAMP
I
2R
RELAY
LATCHES
4R
–
8R
16R
V
+
Adapted from: Paul M. Rainey, "Facsimile Telegraph System,"
U.S. Patent 1,608,527, Filed July 20, 1921, Issued November 30, 1926
Figure 3.10: Paul M. Rainey's 5-Bit Binary-Weighted DAC
The objective of Rainey's DAC was to control the intensity of the light from an
incandescent lamp located in the receiver. By connecting various combinations of parallel
shorting switches, 32 possible values of series resistance can be obtained ranging from 0
to 31·R, and hence 32 possible levels of light intensity. In the PCM facsimile application,
the lamp output was focused on a photosensitive receiving film designed to reproduce the
image digitized at the transmitter.
Another example of an early vacuum tube binary DAC can be found in John Schelleng's
1946 (filing date) patent for a 6-bit PCM system (Reference 7). Schelleng uses binaryweighted switched voltage sources whose outputs are summed together with a resistor
network as shown in Figure 3.11.
In a truly elegant 1953 paper on successive approximation ADCs (Reference 8), B. D.
Smith of Melpar proposed two classic binary-weighted voltage-mode DAC architectures.
The first, shown in Figure 3.12, uses a binary-weighted resistor network switched
between a reference voltage and ground as the basis of the DAC. A transistorized version
of this approach was later described in 1958 by B. K. Smith (Reference 9). B. D. Smith's
second approach (shown later in Figure 3.16), is one of the first reported uses of an R-2R
ladder network in a voltage-mode switching DAC (see again, Reference 8).
3.9
ANALOG-DIGITAL CONVERSION
MSB
LSB
R
DC AMP
AND LIMITER
0 OR 32V
DC AMP
AND LIMITER
0 OR 32V
ATTENUATOR
1/2
DC AMP
AND LIMITER
0 OR 32V
ATTENUATOR
1/4
DC AMP
AND LIMITER
0 OR 32V
ATTENUATOR
1/8
DC AMP
AND LIMITER
0 OR 32V
ATTENUATOR
1 / 16
DC AMP
AND LIMITER
0 OR 32V
ATTENUATOR
1 / 32
R
R
VOUT
R
RL
R
R
Adapted from: John C. Schelleng, "Code Modulation Communication System,"
U.S. Patent 2,453,461, Filed June 19, 1946, Issued November 9, 1948
Figure 3.11: John Schelleng's 6-Bit Binary-Weighted DAC
V
OUT
R/8
R/4
LSB
R/2
R
MSB
VREF
Adapted from: B. D. Smith, "Coding by Feedback Methods," Proceedings of the
I. R. E., Vol. 41, August 1953, pp. 1053-1058
Figure 3.12: Voltage-mode Binary-Weighted Resistor DAC
The voltage-mode binary-weighted resistor DAC shown in Figure 3.12 is usually the
simplest textbook example of a DAC. However, this DAC is not inherently monotonic
and is actually quite hard to manufacture successfully at high resolutions. In addition, the
output impedance of the voltage-mode binary DAC changes with the input code.
Current-mode binary DACs are shown in Figure 3.13A (resistor-based), and Figure
3.13B (current-source based). An N-bit DAC of this type consists of N weighted current
sources (which may simply be resistors and a voltage reference) in the ratio
3.10
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
1:2:4:8:....:2N–1. The LSB switches the 2N–1 current, the MSB the 1 current, etc. The
theory is simple but the practical problems of manufacturing an IC of an economical size
with current or resistor ratios of even 128:1 for an 8-bit DAC are enormous, especially as
they must have matched temperature coefficients.
(A) RESISTOR
(B) CURRENT SOURCE
VREF
I
R
2R
4R
I/2
I/4
8R
MSB
MSB
I/8
LSB
LSB
T
CURRENT OUTPUTS
INTO VIRTUAL GROUNDS
DIFFICULT TO FABRICATE IN IC
FORM DUE TO LARGE RESISTOR OR CURRENT
RATIOS FOR HIGH RESOLUTIONS
Figure 3.13: Current-Mode Binary-Weighted DACs
If the MSB current is slightly low in value, it will be less than the sum of all the other bit
currents, and the DAC will not be monotonic (the differential non-linearity of most types
of DAC is worst at major bit transitions). This architecture is virtually never used on its
own in integrated circuit DACs, although, again, 3- or 4-bit versions have been used as
components in more complex structures.
However, there is another binary-weighted DAC structure that has recently become
widely used. This uses binary-weighted capacitors as shown in Figure 3.14. The problem
with a DAC using capacitors is that leakage causes it to lose its accuracy within a few
milliseconds of being set. This may make capacitive DACs unsuitable for general
purpose DAC applications, but it is not a problem in successive approximation ADCs,
since the conversion is complete in a few µs or less—long before leakage has any
appreciable effect.
The successive approximation ADC has a very simple structure, low power, and
reasonably fast conversion times. It is probably the most widely used general-purpose
ADC architecture, but in the mid-1990s the subranging ADC was starting to overtake the
successive approximation type in popularity because the R-2R thin-film resistor DAC in
the successive approximation ADC made the chip larger and more expensive than that of
a subranging ADC, even though the subranging types tend to use more power. The
development of sub-micron CMOS processes made possible very small (and therefore
cheap), and very accurate switched capacitor DACs. These enabled a new generation of
successive approximation ADCs to be made small, cheap, low-power and precise, and
thus to regain their popularity (see further discussion in Section 3.2 of this chapter).
3.11
ANALOG-DIGITAL CONVERSION
BIT1
(MSB)
BIT2
SC
BIT3
(LSB)
A
CTOTAL = 2C
C
C/ 2
C/ 4
_
C/ 4
+
S1
S2
S3
S4
AIN
SIN
VREF
SWITCHES SHOWN IN TRACK (SAMPLE) MODE
Figure 3.14: Capacitive Binary-Weighted DAC in
Successive Approximation ADC
The use of capacitive charge redistribution DACs offers another advantage as well—the
DAC itself behaves as a sample-and-hold circuit (SHA), so neither an external SHA nor
allocation of chip area for a separate integral SHA are required.
R-2R DACs
One of the most common DAC building-block structures is the R-2R resistor ladder
network shown in Figure 3.15. It uses resistors of only two different values, and their
ratio is 2:1. An N-bit DAC requires 2N resistors, and they are quite easily trimmed. There
are also relatively few resistors to trim.
R
2R
2R
R
2R
R
2R
2R
Figure 3.15: 4-Bit R-2R Ladder Network
There are two ways in which the R-2R ladder network may be used as a DAC—known
respectively as the voltage mode and the current mode (they are sometimes called
"normal" mode and "inverted" mode, but as there is no consensus on whether the voltage
mode or the current mode is the "normal" mode for a ladder network this nomenclature
can be misleading). Each mode has its advantages and disadvantages.
3.12
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
In the voltage mode R-2R ladder DAC shown in Figure 3.16, the "rungs" or arms of the
ladder are switched between VREF and ground, and the output is taken from the end of the
ladder. The output may be taken as a voltage, but the output impedance is independent of
code, so it may equally well be taken as a current into a virtual ground. As mentioned
earlier, this architecture was proposed by B. D. Smith in 1953 (Reference 8).
R
2R
2R
R
2R
LSB
V
OUT
R
2R
2R
MSB
VREF
Adapted from: B. D. Smith, "Coding by Feedback Methods," Proceedings of the
I. R. E., Vol. 41, August 1953, pp. 1053-1058
Figure 3.16: Voltage-Mode R-2R Ladder Network DAC
The voltage output is an advantage of this mode, as is the constant output impedance,
which eases the stabilization of any amplifier connected to the output node. Additionally,
the switches switch the arms of the ladder between a low impedance VREF connection and
ground, which is also, of course, low impedance, so capacitive glitch currents tend not to
flow in the load. On the other hand, the switches must operate over a wide voltage range
(VREF to ground). This is difficult from a design and manufacturing viewpoint, and the
reference input impedance varies widely with code, so that the reference input must be
driven from a very low impedance. In addition, the gain of the DAC cannot be adjusted
by means of a resistor in series with the VREF terminal.
In the current-mode R-2R ladder DAC shown in Figure 3.17, the gain of the DAC may be
adjusted with a series resistor at the VREF terminal, since in the current mode, the end of
the ladder, with its code-independent impedance, is used as the VREF terminal; and the
ends of the arms are switched between ground (or, sometimes, an "inverted output" at
ground potential) and an output line which must be held at ground potential. The normal
connection of a current-mode ladder network output is to an op amp configured as
current-to-voltage (I/V) converter, but stabilization of this op amp is complicated by the
DAC output impedance variation with digital code.
3.13
ANALOG-DIGITAL CONVERSION
VREF
*
R
R
R
<< R
2R
2R
MSB
2R
2R
2R
LSB
CURRENT
OUTPUT
INTO
VIRTUAL
GROUND
* GAIN TRIM IF REQUIRED
Figure 3.17: Current-Mode R-2R Ladder Network DAC
Current-mode operation has a larger switching glitch than voltage mode since the
switches connect directly to the output line(s). However, since the switches of a currentmode ladder network are always at ground potential, their design is less demanding and,
in particular, their voltage rating does not affect the reference voltage rating. If switches
capable of carrying current in either direction (such as CMOS devices) are used, the
reference voltage may have either polarity, or may even be ac. Such a structure is one of
the most common types used as a multiplying DAC (MDAC). These will be discussed
later in this section.
Since the switches are always at, or very close to, ground potential, the maximum
reference voltage may greatly exceed the logic voltage, provided the switches are makebefore-break—which they are in this type of DAC. It is not uncommon for a CMOS
MDAC to accept a ±30 V reference (or even a 60-V peak-to-peak ac reference) while
working from a single 5-V supply.
Another popular form of R-2R DAC switches equal currents into the R-2R network as
shown in Figure 3.18. This architecture was first implemented by Bernard M. Gordon at
EPSCO (now Analogic, Inc.) in a vacuum tube 11-bit, 50-kSPS successive approximation
ADC. Gordon's 1955 patent application (Reference 10) describes the ADC, which was
the first commercial offering of a complete converter (see Section 3.2 of this chapter and
Chapter 1 for more details). In this architecture the output impedance of the DAC is equal
to R, and this structure is often used in high-speed video DACs. A distinct advantage is
that only a 2:1 resistor ratio is required regardless of the resolution. In some applications,
however, the relatively low output impedance can be a disadvantage.
Figure 3.19 shows a DAC using binary-weighted currents switched into a load. The
output impedance is high, and this architecture generally has a volt or so of output
compliance. The main problem with all of the binary-weighted DACs discussed thus far
is that high resolutions require large resistor ratios, making manufacture very difficult.
3.14
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
I
I
I
I
MSB
LSB
R
OUTPUT
2R
R
2R
R
2R
2R
2R
Adapted from: Bernard M. Gordon and Robert P. Talambiras, "Signal Conversion Apparatus,"
U.S. Patent 3,108,266, filed July 22, 1955, issued October 22, 1963
Figure 3.18: Equal Current Sources Switched into an R-2R Ladder Network
I
I/2
MSB
I/4
I/8
LSB
OUTPUT
R (CAN BE EXTERNAL)
Figure 3.19: Binary-Weighted Current Sources Switched into a Load
In 1970 Analog Devices introduced the AD550 "µDAC" monolithic quad (4-bit) current
switch building block IC shown in Figure 3.20. Notice that the binary-weighted currents
were generated using an external thin film network—on-chip laser trimmed thin film
resistor technology was not developed until several years later. The transistor areas are
scaled (8:4:2:1), thereby ensuring equal current densities in all the transistors for
optimum VBE matching.
An alternative method of developing the binary-weighted currents in the quad switch is
shown in Figure 3.21, where an R-2R ladder network connected to the transistor emitters
accomplishes the binary current division.
3.15
ANALOG-DIGITAL CONVERSION
CURRENT
OUTPUT
MSB
I
I/4
I/2
I/8
VB
R
–V
EXTERNAL
THIN FILM
RESISTOR
NETWORK
2R
8R
4R
Figure 3.20: Binary-Weighted 4-Bit DAC, the AD550 "µDAC" Quad Switch
CURRENT
OUTPUT
MSB
I
I/4
I/2
I/8
I/8
VB
2R
–V
2R
R
2R
2R
R
2R
R
Figure 3.21: Binary-Weighted 4-Bit DAC:
R/2R Ladder Network Current Setting Resistors
Figure 3.22 shows how three AD550 quad switches with 16:1 inter-stage attenuators are
connected to form a 12-bit current-output DAC. Note that the maximum required resistor
ratio of 16:1 is manageable. This monolithic "quad switch" (AD550 µDAC) along with a
thin film resistor network (AD850), voltage reference, and an op amp formed the popular
building blocks for 12-bit DACs in the early 1970s before the complete function was
available in IC form several years later. The concept for the quad switch was patented by
James J. Pastoriza (1970 filing, Reference 11).
3.16
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
MSB
I
I/2
I/4
I/8
CURRENT OUTPUT
I
I/2
I/4
I/8
16:1
ATTENUATOR
I
I/2
I/4
I/8
LSB
16:1
ATTENUATOR
Figure 3.22: 12-Bit Current-Output DAC Using
Cascaded Binary "Quad Switches"
The complete 1970-vintage 12-bit DAC solution, shown in Figure 3.23, consists of three
monolithic quad switches, a thin-film resistor network, an op amp, and a voltage
reference. The matching provided by the monolithic quad switches along with the
accuracy and tracking of the external thin film network provided 12-bit performance
without the need for additional trimming. An interesting and complete analysis of this 12bit DAC based on the quad switches can be found in Reference 12.
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
OP AMP
+
AD850
CURRENT SETTING AND INTERSTAGE ATTENUATOR
PRECISION THIN FILM RESISTOR NETWORK
REFERENCE
CIRCUIT
James J. Pastoriza, "Solid State Digital-to-Analog Converter,"
U.S. Patent 3,747,088, filed December 30, 1970, issued July 17, 1973
Figure 3.23: A 1970 Vintage 12-Bit DAC Using Quad Current Switches,
Thin Film Resistor Network, Op Amp, and Zener Diode Voltage Reference
3.17
ANALOG-DIGITAL CONVERSION
One of the problems in implementing a completely monolithic 12-bit DAC using the
quad switch approach is that each 4-bit DAC requires emitter areas scaled 8:4:2:1. This
requires a total of 15 unit emitter areas, and consumes a fairly large chip area. A few
years after the introduction of the quad switch building block, Paul Brokaw of Analog
Devices invented a technique in which only the first two current sources have an emitter
scaling of 2:1. Subsequent current sources have the same unit emitter area but operate at
different current densities—while still maintaining stable currents over temperature. Paul
Brokaw's classic patent (filed in 1975) describes this technique in detail, and this
particular patent is probably the most referenced and cited patent relating to data
converters (Reference 13).
Segmented DACs
So far we have considered basic DAC architectures. When we are required to design a
DAC with a specific performance, it may well be that no single architecture is ideal. In
such cases, two or more DACs may be combined in a single higher resolution DAC to
give the required performance. These DACs may be of the same type or of different types
and need not each have the same resolution.
In principle, one DAC handles the MSBs, another handles the LSBs, and their outputs are
added in some way. The process is known as "segmentation," and these more complex
structures are called "segmented DACs". There are many different types of segmented
DACs and some, but by no means all, of them will be illustrated in the next few
diagrams.
Figure 3.24 shows two varieties of segmented voltage-output DAC. The architecture in
Figure 3.24A is sometimes called a Kelvin-Varley Divider, or "string DAC." Since there
are buffers between the first and second stages, the second string DAC does not load the
first, and the resistors in this string do not need to have the same value as the resistors in
the other one. All the resistors in each string, however, do need to be equal to each other
or the DAC will not be linear. The examples shown have 3-bit first and second stages but
for the sake of generality, let us refer to the first (MSB) stage resolution as M-bits and the
second (LSB) as K-bits for a total of N = M + K bits. The MSB DAC has a string of 2M
equal resistors, and a string of 2K equal resistors in the LSB DAC.
Buffer amplifiers have offset, of course, and this can cause non-monotonicity in a
buffered segmented string DAC. In the simpler configuration of a buffered Kelvin-Varley
divider buffer (Figure 3.24A), buffer A is always "below" (at a lower potential than)
buffer B, and the extra tap labeled "A" on the LSB string DAC is not necessary. The data
decoding is just two priority encoders. In this configuration, however, buffer offset can
cause non-monotonicity.
3.18
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
(A)
VREF
KELVIN-VARLEY DIVIDER
("STRING DAC")
A
B
(B)
A
VREF
KELVIN DIVIDER AND
R-2R LADDER NETWORK
OUTPUT
B
A
B
OUTPUT
A
B
NOTE:
MSB OF R-2R LADDER
ON RIGHT
A
B
A
A
B
IF THE R-2R LADDER NETWORK
IS MONOTONIC, THE
WHOLE DAC IS
MONOTONIC
Figure 3.24: Segmented Voltage-Output DACs
But if the decoding of the MSB string DAC is made more complex so that buffer A can
only be connected to the taps labeled "A" in the MSB string DAC, and buffer B to the
taps labeled "B," then it is not possible for buffer offsets to cause non-monotonicity. Of
course, the LSB string DAC decoding must change direction each time one buffer
"leapfrogs" the other, and taps A and B on the LSB string DAC are alternately not used—
but this involves a fairly trivial increase in logic complexity and is justified by the
increased performance.
Rather than using a second string of resistors, a binary DAC can be used to generate the
three LSBs as shown in Figure 3.24B. It is quite hard to manufacture very high resolution
R-2R ladder networks—to be more accurate, it is hard to trim them to monotonicity. So it
is quite common to make high resolution DACs with a ladder network for the LSBs, and
some other structure for two to five of the MSBs. This voltage-output DAC (Figure
3.24B) consists of a 3-bit string DAC followed by a 3-bit buffered voltage-mode ladder
network.
An unbuffered version of the segmented string DAC is shown in Figure 3.25. This
version is more clever in concept (and, of course, can be manufactured on CMOS
processes which make resistors and switches but not amplifiers, so it may be cheaper as
well). It is intrinsically monotonic. Here, the resistors in the two strings must be equal,
except that the top resistor in the MSB string must be smaller—1/2K of the value of the
others—and the LSB string has 2K – 1 resistors rather than 2K. Because there are no
buffers, the LSB string appears in parallel with the resistor in the MSB string that it is
switched across and loads it. This drops the voltage across that MSB resistor by 1 LSB of
the LSB DAC—which is exactly what is required. The output impedance of this DAC,
being unbuffered, varies with changing digital code.
3.19
ANALOG-DIGITAL CONVERSION
VREF
R
8
R
63 V
64 REF
55 V
64 REF
48 V
64 REF
R
R
R
R
R
R
R
R
40 V
64 REF
32
64
24
64
16
64
R
R
R
VREF
R
VREF
R
VREF
R
8 V
64 REF
55 V
64 REF
54 V
64 REF
53 V
64 REF
52 V
64 REF
51 V
64 REF
OUTPUT
50 V
64 REF
49 V
64 REF
48 V
64 REF
Dennis Dempsey and Christopher Gorman, "Digital-to-Analog Converter,"
U.S. Patent 5,969,657, filed July 27, 1997, issued October 19, 1999.
Figure 3.25: Segmented Unbuffered String DACs Use Patented Architecture
In order to understand this clever concept better, the actual voltages at each of the taps
has been worked out and labeled for the 6-bit segmented DAC composed of two 3-bit
string DACs shown in Figure 3.25. The reader is urged to go through this simple analysis
with the second string DAC connected across any other resistor in the first string DAC
and verify the numbers. A detailed mathematical analysis of the unbuffered segmented
string DAC can be found in the relevant patent filed by Dennis Dempsey and Christopher
Gorman of Analog Devices in 1997 (Reference 14).
Very high speed DACs for video, communications, and other HF reconstruction
applications are often built with arrays of fully decoded current sources. The two or three
LSBs may use binary-weighted current sources. It is extremely important that such DACs
have low distortion at high frequency, and there are several important issues to be
considered in their design.
First of all, currents are never turned on and off—they are steered to one place or another.
Turning a current off at high speed frequently involves inductive spikes and, in general,
because of capacitance charging, it takes longer than current steering.
Secondly, it is important that the voltage change on the chip required to switch the
current should be kept as small as possible. A voltage change causes more charge to flow
in stray capacitances and a larger charge-coupled glitch.
Finally, the decoding must be done before the new data is applied to the DAC so that all
the data is ready and can be applied simultaneously to all the switches in the DAC. This
is generally implemented by using separate parallel latches for the individual switches in
a fully decoded array. If all switches were to change state instantaneously and
simultaneously there would be no skew glitch—by very careful design of propagation
delays around the chip and time constants of switch resistance and stray capacitance the
3.20
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
update synchronization can be made very good, and hence the glitch-related distortion is
very small.
Two examples of segmented current-output DAC structures are shown in Figure 3.26.
Figure 3.26A shows a resistor-based approach for the 7-bit DAC where the 3 MSBs are
fully decoded, and the 4 LSBs are derived from an R-2R network. Figure 3.26B shows a
similar implementation using current sources. The current source implementation is by
far the most popular for today's high-speed reconstruction DACs.
- MSB THERMOMETER DAC
3-BIT
4- BIT R- 2R DAC
VREF
(A)
R
R
R
R
R
R
R
2R
R
R
R
2R
2R
2R
2R
RESISTOR
BASED
CURRENT
OUTPUT
IOUT
IOUT
- MSB THERMOMETER DAC
3-BIT
(B)
CURRENTSOURCE
BASED
I
I
I
I
I
I
4- BIT BINARY DAC
I
I
2
I
8
I
4
I
16
CURRENT
OUTPUT
IOUT
IOUT
Figure 3.26: Segmented Current-Output DACs:
(A) Resistor-Based, (B) Current-source based
It is also often desirable to utilize more than one fully-decoded thermometer section to
make up the total DAC. Figure 3.27 shows a 6-bit DAC constructed from two fullydecoded 3-bit DACs. As previously discussed, these current switches must be driven
simultaneously from parallel latches in order to minimize the output glitch.
- MSB THERMOMETER DAC
3-BIT
I
I
I
I
I
I
3- BIT LSB THERMOMETER DAC
I
I
8
I
8
I
8
I
8
I
8
I
8
I
8
CURRENT
OUTPUT
IOUT
IOUT
Figure 3.27: 6-Bit Current-Output Segmented DAC
Based on Two 3-Bit Thermometer DACs
3.21
ANALOG-DIGITAL CONVERSION
The AD9775 14-bit, 160-MSPS (input)/400-MSPS (output) TxDAC® uses three sections
of segmentation as shown in Figure 3.28. Other members of the AD977x-family and the
AD985x-family also use this same basic core.
The first 5 bits (MSBs) are fully decoded and drive 31 equally weighted current switches,
each supplying 512 LSBs of current. The next 4 bits are decoded into 15 lines which
drive 15 current switches, each supplying 32 LSBs of current. The 5 LSBs are latched
and drive a traditional binary-weighted DAC which supplies 1 LSB per output level. A
total of 51 current switches and latches are required to implement this ultra low glitch
architecture.
5
31
31
BITS 1-5
DECODE
14
14-BIT
LATCH
5-TO-31
4
15
51-BIT
LATCH
I = 512 LSB
15
BITS 6-9
DECODE
CLOCK
15
CURRENT
SWITCHES
I = 32 LSB
4-TO-15
5
31
CURRENT
SWITCHES
5
5
5 BINARY
CURRENT
SWITCHES
I = 1 LSB
CURRENT
OUTPUT
FS =
2mA20mA
NOTE: Differential Outputs Not Shown
Figure 3.28: AD9775 TxDAC 14-Bit CMOS DAC Core
The basic current switching cell in the TxDAC family is made up of a differential PMOS
transistor pair as shown in Figure 3.29. The differential pairs are driven with low-level
logic to minimize switching transients and time skew. The DAC outputs are symmetrical
differential currents, which help to minimize even-order distortion products (especially
when driving a differential output such as a transformer or an op amp differential currentto-voltage converter).
The overall architecture of the AD977x TxDAC® family and the AD985x-DDS family is
an excellent tradeoff between power/performance, and allows the entire DAC function to
be implemented on a standard CMOS process with no thin-film resistors. Single-supply
operation on +3.3 V or +5 V make the devices extremely attractive for portable and low
power applications.
3.22
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
+VS
RL
RL
Figure 3.29: PMOS Transistor Current Switches
Oversampling Interpolating DACs
The basic concept of an oversampling/interpolating DAC is shown in Figure 3.30. The Nbits of input data are received at a rate of fs. The digital interpolation filter is clocked at
an oversampling frequency of Kfs, and inserts the extra data points. The effects on the
output frequency spectrum are shown in Figure 3.30. In the Nyquist case (A), the
requirements on the analog anti-imaging filter can be quite severe. By oversampling and
interpolating, the requirements on the filter are greatly relaxed as shown in (B). Also,
since the quantization noise is spread over a wider region with respect to the original
signal bandwidth, an improvement in the signal-to-noise ratio is also achieved. By
doubling the original sampling rate (K = 2), an improvement of 3 dB is obtained, and by
making K = 4, an improvement of 6 dB is obtained. Early CD players took advantage of
this, and generally carried the arithmetic in the digital filter to more than N-bits. Today,
most DACs in CD players are sigma-delta types.
One of the earliest publications on the oversampling/interpolating DAC concept was by
Ritchie, Candy, and Ninke in 1974 (Reference 16) and followed by a 1981 patent (filing
date) by Mussman and Korte (Reference 17).
Interpolation filters are not only used in very high speed DACs—they are also found in
the last type of DAC we shall discuss later in Section 3.3 of this chapter—the Σ-∆ DAC.
Where high resolutions are required and the output bandwidth is less than a few hundred
kHz, then Σ-∆ technology is well suited for the application.
3.23
ANALOG-DIGITAL CONVERSION
(FROM PLL CLOCK MULTIPLIER)
Kfs
N-BITS
@ fs
DIGITAL
INTERPOLATION
FILTER
(A):NYQUIST
N-BITS
@ K fs
DAC
Anti-imaging Filter Response
ANALOG
OUTPUT
ANTIIMAGING
FILTER
(B): OVERSAMPLING
WITH INTERPOLATION
f
f
fs
2
fs
Kfs
Kfs
2
Figure 3.30: Oversampling Interpolating DAC
Multiplying DACs
In many DACs, the voltage reference is built in—sometimes it may be varied a little as a
gain adjustment, sometimes it is inaccessible. Other DACs may require an external
reference voltage source, but can accept only a narrow range of reference voltages.
In all DACs, the output is the product of the reference voltage and the digital code, so in
that sense, all DACs are multiplying DACs. But some DACs use an external reference
voltage which may be varied over a wide range. These are "Multiplying DACs" or
MDACs where the analog output is the product of the analog input and the digital code as
shown in Figure 3.31. They are extremely useful in many different applications. A strict
definition of an MDAC is that it will continue to work correctly as its reference is
reduced to zero, but sometimes the term is used less stringently for DACs which work
with a reference range of 10:1 or even 6:1—a better name for devices of this type might
be "semi-multiplying" DACs.
While some types of multiplying DACs will work only with references of one polarity
(two quadrant) others handle bipolar (positive or negative) references, and can work with
an ac signal as a reference as well. A bipolar DAC that will work with bipolar reference
voltages is known as a four-quadrant multiplying DAC. Some types of MDACs are so
configured that they can work with reference voltages substantially greater than their
supply voltage.
3.24
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
VREF
REFERENCE
INPUT
ANALOG OUTPUT
DIGITAL
INPUT
MDAC
N
ANALOG OUPTUT = VREF × DIGITAL INPUT × CONSTANT
DIFFICULT TO MAINTAIN DIFFERENTIAL LINEARITY AS VREF APPROACHES ZERO
Figure 3.31: Multiplying DAC ("MDAC")
Current-mode ladder networks and CMOS switches permit positive, negative, and ac
VREF as shown in Figure 3.32. While this is a simple implementation of an MDAC,
several others are possible.
VREF
*
R
R
R
<< R
2R
2R
MSB
2R
2R
2R
LSB
CURRENT
OUTPUT
INTO
VIRTUAL
GROUND
* GAIN TRIM IF REQUIRED
VREF CAN BE AC, ± ,ALLOWING FOUR-QUADRANT OPERATION
Figure 3.32: Multiplying DAC Using Current-Mode
R-2R Ladder Network and CMOS switches.
Intentionally Nonlinear DACs
Thus far, we have emphasized the importance of maintaining good differential and
integral linearity. However, there are situations where ADCs and DACs which have been
made intentionally nonlinear (but maintaining good differential linearity) are useful,
3.25
ANALOG-DIGITAL CONVERSION
especially when processing signals having a wide dynamic range. One of the earliest uses
of nonlinear data converters was in the digitization of voiceband signals for pulse code
modulation (PCM) systems. Major contributions were made at Bell Labs during the
development of the T1 carrier system. The motive for the nonlinear ADCs and DACs was
to reduce the total number of bits (and therefore the serial transmission rate) required to
digitize voice channels. Straight linear encoding of a voice channel required 11- or 12bits and a sampling rate of 8 kSPS per. In the 1960s Bell Labs determined that 7-bit
nonlinear encoding was sufficient; and later in the 1970s they went to 8-bit nonlinear
encoding for better performance (References 18-23).
The nonlinear transfer function allocates more quantization levels out of the total range
for small signals and fewer for large amplitude signals. In effect, this reduces the
quantization noise associated with small signals (where it is most noticeable) and
increases the quantization noise for larger signals (where it is less noticeable). The term
companding is generally used to describe this form of encoding.
The logarithmic transfer function chosen is referred to as the "Bell µ-255" standard, or
simply "µ-law." A similar standard developed in Europe is referred to as "A-law." The
Bell µ-law allows a dynamic range of about 4000:1 using 8 bits, whereas an 8-bit linear
data converter provides a range of only 256:1.
The first generation channel bank (D1) generated the logarithmic transfer function using
temperature controlled resistor-diode networks for "compressors" ahead of a 7-bit linear
ADC in the transmitter. Corresponding resistor-diode "expandors" having an inverse
transfer function followed the 7-bit linear DAC in the receiver. The next generation D2
channel banks used nonlinear ADCs and DACs to accomplish the compression/expansion
functions in a much more reliable and cost-effective manner, and eliminated the need for
the temperature-controlled diode networks.
In his 1953 classic paper, B. D. Smith proposed that the transfer function of a successive
approximation ADC utilizing a nonlinear internal DAC in the feedback path would be the
inverse transfer function of the DAC (Reference 8). The same basic DAC could therefore
be used in the ADC and also for the reconstruction DAC. Later in the 1960s and early
1970s, nonlinear ADC and DAC technology using piecewise linear approximations of the
desired transfer function allowed low cost, high volume implementations (References 1823). These nonlinear 8-bit, 8-kSPS data converters became popular telecommunications
building blocks.
The nonlinear transfer function of the 8-bit DAC is first divided into 16 segments
(chords) of different slopes—the slopes are determined by the desired nonlinear transfer
function. The 4 MSBs determine the segment containing the desired data point, and the
individual segment is further subdivided into 16 equal quantization levels by the 4 LSBs
of the 8-bit word. This is shown in Figure 3.33 for a 6-bit DAC, where the first 3 bits
identify one of the 8 possible chords, and each chord is further subdivided into 8 equal
levels defined by the 3 LSBs. The 3 MSBs are generated using a nonlinear string DAC,
and the 3 LSBs are generated using a 3-bit binary R-2R DAC.
3.26
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
V
REF
128R
NONLINEAR DIVIDER
R-2R DAC
LSB
128mV
V
REF = 2.048V
OUTPUT
LSB
64mV
LSB
32mV
64R
32R
LSB
16mV
16R
LSB
8mV
LSB
4mV
8R
LSB
2mV
4R
2R
R
LSB
1mV
DYNAMIC RANGE = 2048:1
001000
011000
101000
111000
010000
100000
110000
111111
(LINEARLY CODED 6-BIT
DYNAMIC RANGE = 64:1)
DIGITAL CODE
Figure 3.33: Nonlinear 6-Bit Segmented DAC
In 1982, Analog Devices introduced the LOGDAC® AD7111 monolithic multiplying
DAC featuring wide dynamic range using a logarithmic transfer function. The basic DAC
in the LOGDAC is a linear 17-bit voltage-mode R-2R DAC preceded by an 8-bit input
decoder (see Figure 3.34). The LOGDAC can attenuate an analog input signal, VIN, over
the range 0 dB to 88.5 dB in 0.375 dB steps. The degree of attenuation across the DAC is
determined by an nonlinear-coded 8-bit word applied to the onboard decode logic. This
8-bit word is mapped into the appropriate 17-bit word, which is then applied to a 17-bit
R-2R ladder. A functional diagram of the LOGDAC is shown in Figure 3.34. In addition
to providing the logarithmic transfer function, the LOGDAC also acts as a full fourquadrant multiplying DAC.
VIN
R
R
2R
2R
R
2R
2R
MSB
2R
LSB
RFB
IOUT
AGND
Figure 3.34: AD7111 LOGDAC® (Released 1982)
3.27
ANALOG-DIGITAL CONVERSION
With the introduction of high resolution linear ADCs and DACs, the method used in the
LOGDAC® is widely used today to implement various nonlinear transfer functions such
as the µ-law and A-law companding functions required for telecommunications and other
applications. Figure 3.35 shows a general block diagram of the modern approach. The µlaw or A-law companded input data is mapped into data points on the transfer function of
a high resolution DAC. This mapping can be easily accomplished by a simple lookup
table in either hardware, software, or firmware. A similar nonlinear ADC can be
constructed by digitizing the analog input signal using a high resolution ADC and
mapping the data points into a shorter word using the appropriate transfer function. A big
advantage of this method is that the transfer curve does not have to be approximated with
straight line segments as in the earlier method, thereby providing more accuracy.
INPUT
DATA
N-BITS
LOW RESOLUTION
NONLINEARLY
CODED DATA
M>N
M-BITS
DECODE M-BITS
LOGIC
M-BIT
(LOOKUP
LATCH
TABLE)
DECODE LOGIC
CAN BE HARDWARE,
SOFTWARE,
OR FIRMWARE
M-BIT
LINEAR
DAC
HIGH RESOLUTION
LINEAR DAC
VOUT
NONLINEAR
ANALOG OUTPUT
Figure 3.35: General Nonlinear DAC
Counting, Pulse-Width-Modulated (PWM) DACs
Although much less popular than the parallel input DACs previously described, various
types of DACs can be constructed using counters to generate an output voltage
proportional to a digital input word. A. H. Reeves' classic 1939 PCM patent (Reference
24) describes a 5-bit counting ADC and DAC, and the DAC circuit is shown in Figure
3.36. The operation is simple. A sampling clock starts the counter, which is loaded with
the digital word and simultaneously sets an R/S flip-flop. The counter counts upward at a
fast rate, and when it reaches all "ones", the R/S flip-flop is reset. The output of the R/S
flip flop is therefore a pulse-width-modulated (PWM) pulse whose width is proportional
to the complement of the binary word. In a variation on the method, the sampling clock
starts a ramp generator, and the reset pulse activates a sample-and-hold which stores the
output of the ramp generator.
3.28
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
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
5-BIT
COUNTER
VOICE
OUTPUT
R
FF
LPF
S
CLOCK
600kHz
÷ 100
Figure 3.36: A.H. Reeves' 5-Bit Counting DAC
Resolution must be traded for update rate in a counting DAC, because the counter must
cycle through all possible outputs in the sampling interval. Counting DACs do have the
advantage, however, that they are inherently monotonic.
Cyclic Serial DACs
Cyclic serial DACs are rarely used today, but in the early days of PCM they were
somewhat attractive because they took advantage of the serial nature of the PCM pulse
stream. An example of a 4-bit implementation is shown in Figure 3.37 and is based on
1948 patent filing (Reference 25). Proper operation of this DAC depends on receiving the
PCM data in the proper order: the LSB is first, and the MSB is last.
Assume that the initial charge on the capacitor is zero and that the serial PCM data
represents the digital code 1011. The receipt of a pulse in Position 1 (n = 1) closes S1 and
connects S2 to the output of the G = 0.5 amplifier. The voltage VR/2 is stored on the
capacitor, and S2 is then connected to the input of the summer. The receipt of a pulse in
Position 2 (n = 2) closes S2 and connects VR to the summer, whose other input is VR/2.
S2 is then connected to the amplifier output, and the charge on the capacitor is now VR/4
+ VR/2. The receipt of no pulse in Position 3 (n = 3) simply causes the capacitor output to
be divided by two, leaving VR/8 + VR/4 on the amplifier output. This voltage is
transferred to the capacitor by S2. In the final cycle, the receipt of a pulse in Position 4 (n
= 4) adds VR to VR/8 + VR/4 which is then divided by two, leaving a final voltage on the
capacitor of VR/16 + VR/8 + VR/2 = 11 VR/16. The final output voltage is then sampled
by a sample-and-hold which holds the output voltage until the completion of the next
cycle.
3.29
ANALOG-DIGITAL CONVERSION
S1
VR
+
Vn
∑
GAIN = 0.5
+
SAMPLE
AND HOLD
VOUT
S2
C
LSB
1
MSB
1
0
n=1 n=2 n=3 n=4
SERIAL PCM
CODE = 1 0 1 1
n = 0, V0 = 0
1
t
n = 1, V1 = VR/2
n = 2, V2 = VR/4 + VR/2
n = 3, V3 = VR/8 + VR/4
n = 4, V4 = VR/16 + VR/8 + VR/2 = 11VR/16
Figure 3.37: 4-Bit Cyclic Serial DAC
It should be noted that this architecture can be made to handle PCM data which has the
MSB first, by using a G = 2 amplifier and making a few other minor modifications
relating to signal scaling.
A truly elegant serial PCM DAC architecture—for its time—was developed by C. E.
Shannon and A. J. Rack in 1948 (References 26, 27). The original concept was
Shannon's, but Rack added an improvement that made the DAC less sensitive to timing
jitter in the PCM pulse stream. The concept circuits are shown in Figure 3.38.
In Figure 3.38A, the serial PCM pulses (LSB first, MSB last) control a switch which is
closed for a small amount of time if a pulse is present (representing a logic "1"), thereby
injecting a fixed charge into the capacitor. If no pulse is present in a given pulse position
(representing a logic "0"), the switch remains open, and no additional charge is injected.
The RC time constant is chosen such that the capacitor voltage discharges to exactly onehalf its initial value in the time interval between PCM pulses, T. The equation which
must be satisfied is simply RC = T/ln2.
The diagram shows the capacitor voltage for the binary code 1011. The vertical axis is
normalized so that unity represents the voltage change produced by a single switch
closure. At the end of the fourth pulse position, the voltage on the capacitor is 11/16,
which corresponds to the binary code of 1011 with an LSB weight of 1/16. The sampleand-hold is activated at the end of the fourth pulse position to acquire and hold the
capacitor voltage until the next PCM word is completed.
3.30
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
v(t)
v(t)
SAMPLE
AND HOLD
R1
C1
C
R
(A)
SHANNON
LSB
1
k•v(t)
SAMPLE
AND HOLD
C2
LSB
MSB
1
0
3
2
1
11
8
1
1
2
R2
CODE = 1 0 1 1
1
k•v(t)
1
3
8
0
3
2
T
1
2
t
1
11
8
HOLD
11
16
(B)
SHANNON-RACK
MSB
1
3
4
L
3
4
3
8
T
HOLD
11
16
t
Figure 3.38: Shannon and Shannon-Rack Decoder
Notice that any jitter in the PCM pulses or the sample-and-hold clock will produce an
error in the final held output voltage. A. J. Rack devised an elegant solution to this
problem, as shown in Figure 3.38B. Rack added a second capacitor, shunted by both a
resistor and an inductor, in series with the original R1-C1 network. The values of the
second capacitor, C2, and the inductor, L, used with it are such as to make the circuit
resonant at the PCM pulse frequency, 1/T. The second resistor, R2, is adjusted so that the
oscillation developed across the resonant circuit is reduced to exactly one-half amplitude
between each pulse period. The resulting composite waveform has regions of zero-slope
spaced one code period, T, apart, thereby making the circuit much less sensitive to timing
jitter in either the PCM pulse train or the sample-and-hold clock. The Shannon-Rack
encoder was implemented in a experimental late-1940s Bell Labs PCM system described
in Reference 27. The resolution was 7 bits, the sampling rate was 8 kSPS, and the
frequency of the PCM pulses was 672 kHz.
Other Low-Distortion Architectures
Modern low-glitch segmented DACs are capable of very low levels of distortion.
However, in some cases, further distortion improvements can be obtained using a
technique called deglitching. The concept requires a track-and-hold and is illustrated in
Figure 3.39.
3.31
ANALOG-DIGITAL CONVERSION
T/H
DAC
OUTPUT
MODE CONTROL
DAC
INPUT
DAC
OUTPUT
T/H
MODE
1000 . . . 0
0111 . . . 1
T
H
H
T
H
T
T
T/H
OUTPUT
Figure 3.39: Deglitching DAC Outputs Using a Track-and-Hold
Just prior to latching new data into the DAC, the track-and-hold is put into the hold mode
so that the DAC switching glitches are isolated from the output. The switching transients
produced by the SHA are code-independent and occur at the clock frequency, and hence
are easily filtered. However, great care must be taken so that the relative timing between
the track-and-hold clock and the DAC update clock is optimum. In addition, the
distortion performance of the track-and-hold must be at least 6- to 10-dB better than the
DAC, or no improvement in SFDR will be realized. Achieving good results using an
external track-and-hold deglitcher becomes increasingly more difficult as clock
frequencies approach 100 MSPS. In most cases, designers should try and use selfcontained devices that are fully specified for low distortion without the requirement of
additional external circuitry, but the technique may still be of use in some applications.
Digital audio applications require DACs with resolutions of over 16-bits and extremely
low total harmonic distortion (THD). One way to make them is to use the segmented
architecture with several decoded MSBs as previously described, but this is likely to have
comparatively large DNL at the MSB transition, which is just where low DNL is needed
for low-level audio distortion. This problem can be avoided by using a digital adder to
put a digital offset in the DAC code, so that the MSB transition of the input code is well
offset from the mid-point of the DAC transfer characteristic, and then using an analog
offset on the DAC output to restore the dc level at the crossover. This technique, of
course, renders part of the DAC's range unusable, but it does minimize mid-scale
distortion.
Figure 3.40 shows the AD1862 20-bit DAC (introduced in 1990) where a digital offset of
1/16th full-scale is added to the incoming 20-bit binary word. The DAC fully decodes the
3 MSBs and generates the 17 LSBs using a binary R-2R DAC section. In order to prevent
clipping at the positive end of the range, the carry output of the adder drives an additional
current switch having a weight equal to bit 4. Finally, an offset current equal to 1/16th
3.32
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
full-scale is subtracted from the DAC output to compensate for the constant digital offset.
It should be noted that since the introduction of the AD1862 in 1990, most modern lowdistortion audio DACs today utilize the sigma-delta architecture almost exclusively.
FROM SERIAL
INPUT
REGISTER
CARRY
I/2
BIT 1
BIT 2
ADDER
+0001
BIT 3
3 TO 7
DECODE
25-BIT
LATCH
I
I
I
I
I
I
I
EQUAL
CURRENT
SWITCHES
I/2
BIT 4
BIT 5
I/4
R-2R
BIT 20
I
AD1862: Released in 1990
2 20
CURRENT
OUTPUT
Figure 3.40: Digital Offset Minimizes Mid-Scale Distortion of Small Signals
DAC Logic Considerations
The earliest monolithic DACs contained little, if any, logic circuitry, and parallel data had
to be maintained on the digital input to maintain the digital output. Today almost all
DACs are latched and data need only be written to them, not maintained. Some even have
nonvolatile latches and remember settings while turned off.
There are innumerable variations of DAC input structure, which will not be discussed
here, but nearly all are described as "double-buffered." A double-buffered DAC has two
sets of latches. Data is initially latched in the first rank and subsequently transferred to
the second as shown in Figure 3.41. There are two reasons why this arrangement is
useful.
DIGITAL
INPUT
INPUT STRUCTURE:
MAY BE SERIAL,
PARALLEL, BYTE-WIDE,
ETC.
OUTPUT LATCH
TRANSFERS DATA
TO DAC TIMING IS
INDEPENDENT OF
INPUT
OUTPUT
DAC
fc = SAMPLING FREQUENCY
OUTPUT STROBE MAY GO TO MANY DACs
Figure 3.41: Double-Buffered DAC Permits Complex
Input Structures and Simultaneous Update
3.33
ANALOG-DIGITAL CONVERSION
The first is that it allows data to enter the DAC in many different ways. A DAC without a
latch, or with a single latch, must be loaded in parallel with all bits at once, since
otherwise its output during loading may be totally different from what it was, or what it is
to become. A double-buffered DAC, on the other hand, may be loaded with parallel data,
or with serial data, with 4-bit or 8-bit words, or whatever, and the output will be
unaffected until the new data is completely loaded and the DAC receives its update
instruction.
The other convenience of the double-buffered structure is that many DACs may be
updated simultaneously: data is loaded into the first rank of each DAC in turn, and when
all is ready, the output buffers of all the DACs are updated at once. There are many DAC
applications where the output of a number of DACs must change simultaneously, and the
double-buffered structure allows this to be done very easily.
Most early monolithic high resolution DACs had parallel or byte-wide data ports and
tended to be connected to parallel data buses and address decoders and addressed by
microprocessors as if they were very small write-only memories. (Some parallel DACs
are not write-only, but can have their contents read as well—this is convenient for some
applications, but is not very common.) A DAC connected to a data bus is vulnerable to
capacitive coupling of logic noise from the bus to the analog output, and therefore many
DACs today have serial data structures. These are less vulnerable to such noise (since
fewer noisy pins are involved), use fewer pins and therefore take less board space, and
are frequently more convenient for use with modern microprocessors, most of which
have serial data ports. Some, but not all, of such serial DACs have both data outputs and
data inputs so that several DACs may be connected in series, with data clocked to all of
them from a single serial port. This arrangement is often referred to as "daisy-chaining."
Of course, serial DACs cannot be used where high update rates are involved, since the
clock rate of the serial data would be too high. Some very high speed DACs actually have
two parallel data ports, and use them alternately in a multiplexed fashion (sometimes this
is called a "ping-pong" input) to reduce the data rate on each port as shown in Figure
3.42. The alternate loading (ping-pong) DAC in the diagram loads from port A and port
B alternately on the rising and falling edges of the clock, which must have a mark-space
ratio close to 50:50. The internal clock multiplier ensures that the DAC itself is updated
with data A and data B alternately at exactly 50:50 time ratio, even if the external clock is
not so precise.
3.34
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
DATA PORT A
N
N
LATCH
A
N
×2 CLOCK
MULTIPLIER
CLOCK
(50% DUTY
CYCLE)
DATA
MULTIPLEXER
N
OUTPUT
DAC
DATA PORT B
N
DAC
LATCH
LATCH
B
N
OUTPUT
Figure 3.42: Alternate Loading (Ping-Pong) High Speed DAC
3.35
ANALOG-DIGITAL CONVERSION
REFERENCES:
3.1 DAC ARCHITECTURES
1.
Peter I. Wold, "Signal-Receiving System," U.S. Patent 1,514,753, filed November 19, 1920, issued
November 11, 1924. (thermometer DAC using relays and vacuum tubes).
2.
Clarence A. Sprague, "Selective System," U.S. Patent 1,593,993, filed November 10, 1921, issued
July 27, 1926. (thermometer DAC using relays and vacuum tubes).
3.
Leland K. Swart, "Gas-Filled Tube and Circuit Therefor," U.S. Patent 2,032,514, filed June 1, 1935,
issued March 3, 1936. (a thermometer DAC based on vacuum tube switches).
4.
Robert Adams, Khiem Nguyen, and Karl Sweetland, "A 113 dB SNR Oversampling DAC with
Segmented Noise-Shaped Scrambling, " ISSCC Digest of Technical Papers, vol. 41, 1998, pp. 62,
63, 413. (describes a segmented audio DAC with data scrambling).
5.
Robert W. Adams and Tom W. Kwan, "Data-directed Scrambler for Multi-bit Noise-shaping D/A
Converters," U.S. Patent 5,404,142, filed August 5, 1993, issued April 4, 1995. (describes a
segmented audio DAC with data scrambling).
6.
Paul M. Rainey, "Facimile Telegraph System," U.S. Patent 1,608,527, filed July 20, 1921, issued
November 30, 1926. (the first PCM patent. Also shows a relay-based 5-bit electro-mechanical flash
converter and a binary DAC using relays and multiple resistors).
7.
John C. Schelleng, "Code Modulation Communication System," U.S. Patent 2,453,461, Filed June 19,
1946, Issued November 9, 1948. (vacuum tube binary DAC using binary weighted voltages summed
into load resistor with equal resistor weights).
8.
B. D. Smith, "Coding by Feedback Methods," Proceedings of the I. R. E., Vol. 41, August 1953, pp.
1053-1058. (Smith uses an internal binary weighted DAC and also points out that a non-linear transfer
function can be achieved by using a DAC with non-uniform bit weights, a technique which is widely
used in today's voiceband ADCs with built-in companding. He was also one of the first to propose
using an R/2R ladder network within the DAC core).
9.
Bruce K. Smith, "Digital Attenuator," U.S. Patent 1,976,527, filed July 17, 1958, issued March 21,
1961. (describes a transistorized voltage output DAC similar to B. D. Smith above).
10. 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, 50kSPS
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.)
11. 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).
12. Eugene R. Hnatek, A User's Handbook of D/A and A/D Converters, John Wiley, New York, 1976,
ISBN 0-471-40109-9, pp. 282-295. (contains an excellent description of the Analog Devices' AD550
monolithic µDAC quad current switch, and AD850 thin film network—building blocks for 12-bit DACs
introduced in 1970 ).
13. 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. (the most referenced data converter patent ever issued).
3.36
DATA CONVERTER ARCHITECTURES
3.1 DAC ARCHITECTURES
14. Dennis Dempsey and Christopher Gorman, "Digital-to-Analog Converter," U.S. Patent 5,969,657,
filed July 27, 1997, issued October 19, 1999. (describes an elegant solution for segmented unbuffered
string DACs).
15. John A. Schoeff, "An Inherently Monotonic 12 Bit DAC," IEEE Journal of Solid State Circuits,
Vol. SC-14, No. 6, December 1979, pp. 904-911. (describes one of the first monolithic DACs to use
segmentation).
16. G. R. Ritchie, J. C. Candy, and W. H. Ninke, "Interpolative Digital-to-Analog Converters," IEEE
Transactions on Communications, Vol. COM-22, November 1974, pp. 1797-1806. (one of the
earliest papers written on oversampling interpolative DACs).
17. H. G. Musmann and W. W. Korte, "Generalized Interpolative Method for Digital/Analog Conversion
of PCM Signals," U.S. Patent 4,467,316, filed June 3, 1981, issued August 21, 1984. (a description of
interpolating DACs).
18. B. Smith, "Instantaneous Companding of Quantized Signals, Bell System Technical Journal, Vol. 36,
May 1957, pp. 653-709. (one of the first papers written about using nonlinear coding techniques for
speech signals in PCM).
19. H. Kaneko and T. Sekimoto, "Logarithmic PCM Encoding Without Diode Compandor," IEEE
Transactions on Communications Systems, Vol. 11, No. 3, September 1963, pp. 296-307. (describes
several methods for nonlinear encoding speech directly without the need for diode compandors).
20. C. L. Dammann, "An Approach to Logarithmic Coders and Decoders," NEREM Record, Boston MA,
November 2-4, 1966, pp. 196-197. (more discussions on nonlinear coders and decoders for PCM).
21. H. Kaneko, "A Unified Formulation of Segment Companding Laws and Synthesis of Codecs and
Digital Compandors," Bell System Technical Journal, Vol. 49, September 1970, pp. 1555-1558.
(discusses the piecewise linear approximation to the logarithmic transfer companding function).
22. M. R. Aaron and H. Kaneko, "Synthesis of Digital Attenuators for Segment Companded PCM Codes,"
Transactions on Communications Technology, COM-19, December 1971, pp. 1076-1087. (more on
nonlinear coding).
23. C. L. Dammann, L. D. McDaniel, and C. L. Maddox, "D2 Channel Bank: Multiplexing and Coding,"
Bell System Technical Journal, Vol. 51, October 1972, pp. 1675-1699. (still more on nonlinear
coding).
24. A.H. Reeves, "Electric Signaling System," U.S. Patent 2,272,070, filed November 22, 1939, issued
February 3, 1942. (Reeves' classic PCM patent which describes a 5-bit PWM counting ADC and
DAC).
25. R. L. Carbrey, "Decoder for Pulse Code Modulation," U.S. Patent 2,579,302, filed January 17, 1948,
issued December 18, 1951. (describes cyclic or sequential attenuation DAC).
26. R. L. Carbrey, "Decoding in PCM,", Bell Labs Record, November 1948, pp. 451-456. (describes the
Shannon-Rack PCM DAC).
27. 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. (further details
of the Shannon-Rack decoder as part of the experimental PCM system).
3.37
ANALOG-DIGITAL CONVERSION
NOTES:
3.38
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
SECTION 3.2: ADC ARCHITECTURES
Walt Kester, James Bryant
Introduction
As in the case of DACs, the relationship between the digital output and the analog input
of an ADC depends upon the value of the reference, and the accuracy of the reference is
almost always the limiting factor on the absolute accuracy of a ADC. We shall consider
the various architectures of ADCs, and the forms which the reference may take, later in
this section.
Similar to DACs, many ADCs use external references (see Figure 3.43) and have a
reference input terminal, while others have an output from an internal reference. The
simplest ADCs, of course, have neither—the reference is on the ADC chip and has no
external connections.
VDD
SAMPLING
CLOCK
VREF
ANALOG
INPUT
ADC
DIGITAL
OUTPUT
EOC, DATA READY, ETC.
VSS
GROUND
(MAY BE INTERNALLY
CONNECTED TO VSS)
Figure 3.43: Basic ADC with External Reference
If an ADC has an internal reference, its overall accuracy is specified when using that
reference. If such an ADC is used with a perfectly accurate external reference, its
absolute accuracy may actually be worse than when it is operated with its own internal
reference. This is because it is trimmed for absolute accuracy when working with its own
actual reference voltage, not with the nominal value. Twenty years ago it was common
for converter references to have accuracies as poor as ±5% since these references were
trimmed for low temperature coefficient rather than absolute accuracy, and the
inaccuracy of the reference was compensated in the gain trim of the ADC itself. Today
the problem is much less severe, but it is still important to check for possible loss of
absolute accuracy when using an external reference with an ADC which has a built-in
one.
3.39
ANALOG-DIGITAL CONVERSION
ADCs which have reference terminals must, of course, specify their behavior and
parameters. If there is a reference input the first specification will be the reference input
voltage—and of course this has two values, the absolute maximum rating, and the range
of voltages over which the ADC performs correctly.
Most ADCs require that their reference voltage is within quite a narrow range whose
maximum value is less than or equal to the ADC's VDD. Notice that this is unlike DACs,
where many allow the reference to be varied over a wide range (as in the MDACs or
semi-multiplying DACs previously discussed in Section 3.1 of this chapter).
The reference input terminal of an ADC may be buffered as shown in Figure 3.44, in
which case it has input impedance (usually high) and bias current (usually low)
specifications, or it may connect directly to the ADC. In either case, the transient currents
developed on the reference input due to the internal conversion process need good
decoupling with external low-inductance capacitors. Most ADC data sheets recommend
appropriate decoupling networks.
Any, all or none of these pins may be
connected internally or brought off the chip.
VDD
SAMPLING
CLOCK
VREF
BUFFER
AMP
ANALOG
INPUT
ADC
VOLTAGE
REFERENCE
DATA
OUTPUT
EOC, DATA READY, ETC.
VSS
GROUND
(MAY BE INTERNALLY
CONNECTED TO VSS)
Figure 3.44: ADC with Reference and Buffer
Where an ADC has an internal reference, it will carry a defined reference voltage, with a
specified accuracy. There may also be specifications of temperature coefficient and longterm stability.
The reference input may be buffered or unbuffered. If it is buffered, the maximum output
current will probably be specified. In general such a buffer will have a unidirectional
output stage which sources current but does not allow current to flow into the output
terminal. If the buffer does have a push-pull output stage, the output current will probably
be defined as ±(SOME VALUE) mA. If the reference output is unbuffered, the output
3.40
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
impedance may be specified, or the data sheet may simply advise the use of a high input
impedance external buffer.
The sampling clock input (sometimes called convert-start or encode command) is a
critical function in an ADC and a source of some confusion. Many of the early integrated
circuit ADCs (such as the industry-standard AD574) did not have a built-in sample-andhold function, and were known simply as encoders. These converters required an external
clock to begin the conversion process. In the case of the AD574, the application of the
external clock initiated an internal high-speed clock oscillator which in turn controlled
the actual conversion process.
Most modern ADCs have the sample-and-hold function on-chip and require an external
sampling clock to initiate the conversion. In some ADCs, only a single sampling clock is
required—in others, both a high frequency clock as well as a lower speed sampling clock
are required. Regardless of the ADC, it is extremely important to read the data sheet and
determine exactly what the external clock requirements are because they can vary widely
from one ADC to another, since there is no standard.
At some point after the assertion of the sampling clock, the output data is valid. This data
may be in parallel or serial format depending upon the ADC. Early successive
approximation ADCs such as the AD574 simply provided a STATUS output (STS)
which went high during the conversion, and returned to the low state when the output
data was valid. In other ADCs, this line is variously called busy, end-of-conversion
(EOC), data ready, etc. Regardless of the ADC, there must be some method of knowing
when the output data is valid—and again, the data sheet is where this information can
always be found.
There are one or two other practical points which are worth remembering about the logic
of ADCs. On power-up, many ADCs do not have logic reset circuitry and may enter an
anomalous logical state. One or two conversions may be necessary to restore their logic
to proper operation so: (a) the first and second conversions after power-up should never
be trusted, and (b) control outputs (EOC, data ready, etc.) may behave in unexpected
ways at this time (and not necessarily in the same way at each power-up), and (c) care
should be taken to ensure that such anomalous behavior cannot cause system latch-up.
For example, EOC (end-of-conversion) should not be used to initiate conversion if there
is any possibility that EOC will not occur until the first conversion has taken place, as
otherwise initiation will never occur.
Some low-power ADCs now have power-saving modes of operation variously called
standby, power-down, sleep, etc. When an ADC comes out of one of these low-power
modes, there is a certain recovery time required before the ADC can operate at its full
specified performance. The data sheet should therefore be carefully studied when using
these modes of operation, and there may be several different levels of power-down.
Another detail which can cause trouble is the difference between EOC and DRDY (data
ready). EOC indicates that conversion has finished, DRDY that data is available at the
output. In some ADCs, EOC functions as DRDY—in others, data is not valid until
several tens of nanoseconds after the EOC has become valid, and if EOC is used as a data
strobe, the results will be unreliable.
3.41
ANALOG-DIGITAL CONVERSION
As a final example, some ADCs use CS (Chip Select) edges to reset internal logic, and it
may not be possible to perform another conversion without asserting or reasserting CS
(or it may not be possible to read the same data twice, or both).
For more detail, it is important to read the whole data sheet before using an ADC since
there are innumerable small logic variations from type to type. Unfortunately, many data
sheets are not as clear as one might wish, so it is also important to understand the general
principles of ADCs in order to interpret data sheets correctly. That is one of the purposes
of this section.
We are now ready to discuss the various architectures for ADCs. Because it is a
fundamental building block used in all ADCs, the comparator (a 1-bit ADC) is treated
first. It is logical to follow this with the flash converter architecture because it is
somewhat analogous to the fully-decoded (thermometer) DAC architecture previously
discussed. The successive approximation ADC architecture is treated next followed by
subranging and pipelined architectures. The folding (Gray-code) architecture completes
the primary architectures used in so-called high-speed ADCs.
The last part of this sections discusses the various counting and integrating architectures
which are generally more suited to high resolution lower speed ADCs. Sigma-Delta (Σ-∆)
ADCs and DACs are treated in a separate section which concludes this chapter.
The Comparator: A 1-Bit ADC
As a changeover switch is a 1-bit DAC, so a comparator is a 1-bit ADC (see Figure 3.45).
If the input is above a threshold, the output has one logic value, below it has another.
Moreover, there is no ADC architecture which does not use at least one comparator of
some sort.
The most common comparator has some resemblance to an operational amplifier in that it
uses a differential pair of transistors or FETs as its input stage, but unlike an op amp, it
does not use external negative feedback, and its output is a logic level indicating which of
the two inputs is at the higher potential. Op amps are not designed for use as
comparators—they may saturate if overdriven and recover slowly. Many op amps have
input stages which behave in unexpected ways when used with large differential voltages,
and their outputs are rarely compatible with standard logic levels. There are cases,
however, when it may be desirable to use an op amp as a comparator, and an excellent
treatment of this subject can be found in Reference 1.
3.42
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
LATCH
ENABLE
+
DIFFERENTIAL
ANALOG INPUT
LOGIC
OUTPUT
–
COMPARATOR
OUTPUT
"1"
VHYSTERESIS
"0"
0
DIFFERENTIAL ANALOG INPUT
Figure 3.45: The Comparator: A 1-Bit ADC
Comparators used as building blocks in ADCs need good resolution which implies high
gain. This can lead to uncontrolled oscillation when the differential input approaches
zero. In order to prevent this, hysteresis is often added to comparators using a small
amount of positive feedback. Figure 3.45 shows the effects of hysteresis on the overall
transfer function. Many comparators have a millivolt or two of hysteresis to encourage
"snap" action and to prevent local feedback from causing instability in the transition
region. Note that the resolution of the comparator can be no less than the hysteresis, so
large values of hysteresis are generally not useful.
Early comparators were designed with vacuum tubes and were often used in radio
receivers—where they were called discriminators, not comparators. Most modern
comparators used in ADCs include a built-in latch which makes them sampling devices
suitable for data converters. A typical structure is shown in Figure 3.46 for the AM685
ECL (emitter-coupled-logic) latched comparator introduced in 1972 by Advanced Micro
Devices, Inc. (see Reference 2). The input stage preamplifier drives a cross-coupled latch.
The latch locks the output in the logic state it was in at the instant when the latch was
enabled. The latch thus performs a track-and-hold function, allowing short input signals
to be detected and held for further processing. Because the latch operates directly on the
input stage, the signal suffers no additional delays—signals only a few nanoseconds wide
can be acquired and held. The latched comparator is also less sensitive to instability
caused by local feedback than an unlatched one.
3.43
ANALOG-DIGITAL CONVERSION
LATCH
ENABLE
Q
+
PREAMP
–
LATCH
Q
From James N. Giles, "High Speed Transistor
Difference Amplifier," U.S. Patent 3,843,934,
filed January 31 1973, issued October 22, 1974
Figure 3.46: The AM685 ECL Comparator (1972)
Where comparators are incorporated into IC ADCs, their design must consider resolution,
speed, overload recovery, power dissipation, offset voltage, bias current, and the chip
area occupied by the architecture which is chosen. There is another subtle but
troublesome characteristic of comparators which can cause large errors in ADCs if not
understood and dealt with effectively. This error mechanism is the occasional inability of
a comparator to resolve a small differential input into a valid output logic level. This
phenomenon is known as metastability—the ability of a comparator to balance right at its
threshold for a short period of time.
The metastable state problem is illustrated in Figure 3.47. Three conditions of differential
input voltage are illustrated: (1) large differential input voltage, (2) small differential
input voltage, and (3) zero differential input voltage. The approximate equation which
describes the output voltage, VO(t) is given by:
VO ( t ) = ∆VIN Ae t / τ ,
Eq. 3.1
Where ∆VIN = the differential input voltage at the time of latching, A = the gain of the
preamp at the time of latching, τ = regeneration time constant of the latch, and t = the
time that has elapsed after the comparator output is latched (see References 3 and 4).
For small differential input voltages, the output takes longer to reach a valid logic level.
If the output data is read when it lies between the "valid logic 1" and the "valid logic 0"
region, the data can be in error. If the differential input voltage is exactly zero, and the
comparator is perfectly balanced at the time of latching, the time required to reach a valid
logic level can be quite long (theoretically infinite). However, hysteresis and noise on the
input makes this condition highly unlikely. The effects of invalid logic levels out of the
comparator are different depending upon how the comparator is used in the actual ADC.
3.44
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
vo(t) = ∆VinA e
COMPARATOR
OUTPUT
LARGE ∆VIN
t/τ
VALID LOGIC "1"
1
3
2
SMALL ∆VIN
∆VO
≈ ZERO ∆VIN
UNDEFINED
VALID LOGIC "0"
t
DATA 1
VALID
DATA 2
VALID
DATA 3
VALID
LATCHED MODE (HOLD)
LATCH
ENABLE
TRANSPARENT MODE (TRACK)
t
t=0
Figure 3.47: Comparator Metastable State Errors
From a design standpoint, comparator metastability can be minimized by making the
gain, A, high, minimizing the regeneration time constant, τ, by increasing the gainbandwidth of the latch, and allowing sufficient time, t, for the output of the comparator to
settle to a valid logic level. It is not the purpose of this discussion to analyze the complex
tradeoffs between speed, power, and circuit complexity when optimizing comparator
designs, but an excellent treatment of the subject can be found in References 3 and 4.
From a user standpoint, the effect of comparator metastability (if it affects the ADC
performance at all) is in the bit error rate (BER)—which is not usually specified on most
ADC data sheets. A discussion of this specification can be found in Chapter 2 of this
book. The resulting errors are often referred to as sparkle codes, rabbits, or flyers.
Bit error rate should not be a problem in a properly designed ADC in most applications,
however the system designer should be aware that the phenomenon exists. An application
example where it can be a problem is when the ADC is used in a digital oscilloscope to
detect small-amplitude single-shot randomly occurring events. The ADC can give false
indications if its BER is not sufficiently small.
3.45
ANALOG-DIGITAL CONVERSION
High Speed ADC Architectures
Flash Converters
Flash ADCs (sometimes called parallel ADCs) are the fastest type of ADC and use large
numbers of comparators. An N-bit flash ADC consists of 2N resistors and 2N–1
comparators arranged as in Figure 3.48. Each comparator has a reference voltage from
the resistor string which is 1 LSB higher than that of the one below it in the chain. For a
given input voltage, all the comparators below a certain point will have their input
voltage larger than their reference voltage and a "1" logic output, and all the comparators
above that point will have a reference voltage larger than the input voltage and a "0" logic
output. The 2N–1 comparator outputs therefore behave in a way analogous to a mercury
thermometer, and the output code at this point is sometimes called a thermometer code.
Since 2N–1 data outputs are not really practical, they are processed by a decoder to
generate an N-bit binary output.
STROBE
+
ANALOG
INPUT
–
+VREF
1.5R
R
+
–
R
+
–
R
+
PRIORITY
ENCODER
AND LATCH
N
DIGITAL
OUTPUT
–
R
+
–
R
+
–
R
+
–
0.5R
Figure 3.48: 3-bit All-Parallel (Flash) Converter
The input signal is applied to all the comparators at once, so the thermometer output is
delayed by only one comparator delay from the input, and the encoder N-bit output by
only a few gate delays on top of that, so the process is very fast. However, the
architecture uses large numbers of resistors and comparators and is limited to low
resolutions, and if it is to be fast, each comparator must run at relatively high power
levels. Hence, the problems of flash ADCs include limited resolution, high power
dissipation because of the large number of high speed comparators (especially at
sampling rates greater than 50 MSPS), and relatively large (and therefore expensive) chip
sizes. In addition, the resistance of the reference resistor chain must be kept low to supply
adequate bias current to the fast comparators, so the voltage reference has to source quite
large currents (typically > 10 mA).
3.46
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
The first documented flash converter was part of Paul M. Rainey's electro-mechanical
PCM facsimile system described in a relatively ignored patent filed in 1921 (Reference
5—see further discussions in Chapter 1 of this book). In the ADC, a current proportional
to the intensity of light 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 (see Figure 3.49). Each individual photocell output activates part
of a relay network which generates the 5-bit binary code.
SERIAL DATA TO RECEIVER
ROTATING COMMUTATOR
STATIONARY
ELECTRICAL CONTACTS
PARALLEL BINARY
OUTPUT DATA
LIGHT
SOURCE
RELAY DECODING
LOGIC
TRANSPARENCY
(NEGATIVE)
RECEIVING
PHOTOCELL
GALVANOMETER
DEFLECTED
LIGHT BEAM
PHOTOCELL BANK (32)
Figure 3.49: A 5-Bit Flash ADC Proposed by Paul Rainey
Adapted from Paul M. Rainey, "Facsimile Telegraph System," U.S. Patent
1,608,527, Filed July 20, 1921, Issued November 30, 1926
A significant development in ADC technology during the period was the electron beam
coding tube developed at Bell Labs and shown in Figure 3.50. The tube described by R.
W. Sears in Reference 6 was capable of sampling at 96 kSPS with 7-bit resolution. The
basic electron beam coder concepts are shown in Figure 3.50 for a 4-bit device. The tube
used a fan-shaped beam creating a "flash" converter delivering a parallel output word.
Early electron tube coders used a binary-coded shadow mask (Figure 3.50A), and large
errors can occur if the beam straddles two adjacent codes and illuminates both of them.
The errors associated with binary shadow masks were later eliminated by using a Gray
code shadow mask as shown in Figure 3.50B. 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 7). 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
3.47
ANALOG-DIGITAL CONVERSION
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.
Collector
Y Deflectors
Shadow Mask
Collector
Electron gun
(A) BINARY CODED
SHADOW MASK
(B) GRAY CODED
SHADOW MASK
Figure 3.50: The Electron Beam Coder from Bell Labs (1948)
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 8). Shortly thereafter,
however, advances in all solid-state ADC techniques made the electron tube technology
obsolete.
It was soon recognized that the flash converter offered the fastest sampling rates
compared to other architectures, but the problem with this approach is that the
comparator circuit itself is quite bulky using discrete transistor circuits and very
cumbersome using vacuum tubes. Constructing a single latched comparator cell using
either technology is quite a task, and extending it to even 4-bits of resolution (15
comparators required) makes it somewhat unreasonable. Nevertheless, work was done in
the mid 1950s and early 1960s as shown in Robert Staffin and Robert D. Lohman's patent
which describes a subranging architecture using both tube and transistor technology
(Reference 9). The patent discusses the problem of the all-parallel approach and points
out the savings by dividing the conversion process into a coarse conversion followed by a
fine conversion.
Tunnel (Esaki) diodes were used as comparators in several experimental early flash
converters in the 1960s as an alternative to a latched comparator based solely on tubes or
transistors (see References 10-13).
3.48
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
In 1964 Fairchild introduced the first IC comparators, the µA711/712, designed by Bob
Widlar. The same year, Fairchild also introduced the first IC op amp, the µA709—
another Widlar design. Other IC comparators soon followed including the Signetics 521,
National LM361, Motorola MC1650 (1968), AM685/687 (1972/1975). With the
introduction of these building block comparators and the availability of TTL and ECL
logic ICs, 6-bit rack-mounted discrete flash converters were introduced by Computer
Labs, Inc., including the VHS-630 (6-bit, 30 MSPS in 1970) and the VHS-675 (6-bit, 75
MSPS in 1975). The VHS-675 shown in Figure 3.51 used 63 AM685 ECL comparators
preceded by a high-speed track-and-hold, ECL decoding logic, contained a built-in linear
power supply (ac line powered), and dissipated a total of 130 W (sale price was about
$10,000 in 1975). Instruments such as these found application in early high speed data
acquisition applications including military radar receivers.
19" × 17" × 7"
VHS-630
6-Bits, 30 MSPS
32 dual MC1650 MECL III
Comparators
100 watts (linear power
supplies included)
VHS-675
6-Bits, 75 MSPS
64 AM685 Comparators
130 watts (linear power
supplies included)
Figure 3.51: VHS-Series ADCs from Computer Labs, Inc.
VHS-630 (1970), VHS-675 (1975)
The AM685 comparator was also used as a building block in the 4-bit 100-MSPS boardlevel flash ADC, the MOD-4100, introduced in 1975 and shown in Figure 3.52.
The first integrated circuit 8-bit video-speed 30-MSPS flash converter, the TDC1007J,
was introduced by TRW LSI division in 1979 (References 14 and 15). A 6-bit version of
the same design, the TDC1014J followed shortly. Also in 1979, Advanced Micro
Devices, Inc. introduced the AM6688, a 4-bit 100-MSPS IC flash converter.
Flash converters became very popular in the 1980s for high speed 8-bit video
applications as well as building blocks for higher resolution subranging card-level,
modular, and hybrid ADCs. Many were fabricated on CMOS processes for lower power
dissipation. Recently, however, the subranging pipeline architecture has become popular
for 8-bit ADCs up to about 250 MSPS. For instance, the AD9480 8-bit 250-MSPS ADC
is fabricated on a high speed BiCMOS process and dissipates less than 400mW compared
to the several watts required for a full flash implementation on a similar process.
3.49
ANALOG-DIGITAL CONVERSION
14 watts total
AM685 ECL COMPARATORS
(16 TOTAL)
9" × 6" × 2"
Figure 3.52: MOD-4100 4-Bit, 100-MSPS Flash Converter,
Computer Labs, 1975
In practice, IC flash converters are currently available up to 10-bits, but more commonly
they have 8-bits of resolution. Their maximum sampling rate can be as high as 1 GHz
(these are generally made on Gallium Arsenide processes with several watts of power
dissipation), with input full-power bandwidths in excess of 300 MHz.
But as mentioned earlier, full-power bandwidths are not necessarily full-resolution
bandwidths. Ideally, the comparators in a flash converter are well matched both for dc
and ac characteristics. Because the sampling clock is applied to all the comparators
simultaneously, the flash converter is inherently a sampling converter. In practice, there
are delay variations between the comparators and other ac mismatches which cause a
degradation in the effective number of bits (ENOBs) at high input frequencies. This is
because the inputs are slewing at a rate comparable to the comparator conversion time.
For this reason, track-and-holds are often required ahead of flash converters to achieve
high SFDR on high frequency input signals.
The input to a flash ADC is applied in parallel to a large number of comparators. Each
has a voltage-variable junction capacitance, and this signal-dependent capacitance results
in most flash ADCs having reduced ENOB and higher distortion at high input
frequencies. For this reason, most flash converters must be driven with a wideband op
amp which is tolerant to the capacitive load presented by the converter as well as high
speed transients developed on the input.
Comparator metastability in a flash converter can severely impact the bit error rate
(BER). Figure 3.53 shows a simple flash converter with one stage of binary decoding
logic. The two-input AND gates convert the thermometer code output of the parallel
comparators into a "one-hot out of 7" code. The decoding logic is simply a "wired-or"
3.50
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
array, a technique popular with ECL logic. Assume that the comparator labeled "X" has
metastable outputs labeled "X". The desired output code should be either 011 or 100, but
note that the 000 code (both gate outputs high) and the 111 code (both gate outputs low)
are also possible due to the metastable states, representing a ½ FS error.
ANALOG
INPUT
Figure 3.53: Metastable Comparator Output States May Cause
Error Codes in Data Converters
Metastable state errors in flash converters can be reduced by several techniques, one of
which involves decoding the comparator outputs in Gray code followed by a Gray-tobinary conversion as in the Bell Labs electron beam encoder previously described. The
advantage of Gray code decoding is that a metastable state in any of the comparators can
produce only a 1-LSB error in the Gray code output. The Gray code is latched and then
converted into a binary code which, in turn, will only have a maximum of 1-LSB error as
shown in Figure 3.54. (This is described in more detail in Chapter 1 of this book in the
section on coding).
The same principles have been applied to several modern IC flash converters to minimize
the effects of metastable state errors as described in References 3, 16, 17, for example.
Power dissipation is always a big consideration in flash converters, especially at
resolutions above 8 bits. A clever technique was used in the AD9410 10-bit, 210-MSPS
ADC called interpolation to minimize the number of preamplifiers in the flash converter
comparators and also reduce the power (2.1 W). The method is shown in Figure 3.55 (see
Reference 18).
3.51
ANALOG-DIGITAL CONVERSION
(A) 4-BIT GRAY CODE
1000
1001
1011
1010
1110
1111
1101
1100
0100
0101
0111
0110
0010
0011
0001
0000
ONLY ONE BIT
CHANGES
BETWEEN
ANY TWO
ADJACENT
CODES
(B) AFTER CONVERSION
TO BINARY CODE
METASTABLE STATE
ERROR PRODUCES
ONLY ONE OF TWO
POSSIBLE GRAY CODES
1100
0100
1111
1110
1101
1100
1011
1010
1001
1000
0111
0110
0101
0100
0011
0010
0001
0000
Figure 3.54: Gray Code Decoding Reduces
Amplitude of Metastable State Errors
ANALOG
INPUT
B
+
A2
V2
B
-
V2
LATCH
2
V1A
DECODE
V1A = V1 + V2
2
A1
V1
-
LATCH
1A
A
+
A
ANALOG
INPUT
V1
B
A
A
B
A
B
B
A
LATCH
1
LATCH
STROBE
AD9410: 10-Bits, 210MSPS
Figure 3.55: "Interpolating" Flash Reduces the Number
of Preamplifiers by Factor of Two
3.52
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
The preamplifiers (labeled "A1", "A2", etc.) are low-gain gm stages whose bandwidth is
proportional to the tail currents of the differential pairs. Consider the case for a positivegoing ramp input which is initially below the reference to AMP A1, V1. As the input
signal approaches V1, the differential output of A1 approaches zero (i.e., A = A ), and the
decision point is reached. The output of A1 drives the differential input of LATCH 1. As
the input signals continues to go positive, A continues to go positive, and B begins to go
negative. The interpolated decision point is determined when A = B . As the input
continues positive, the third decision point is reached when B = B . This novel
architecture reduces the ADC input capacitance and thereby minimizes its change with
signal level and the associated distortion. The AD9410 also uses an input sample-andhold circuit for improved ac linearity.
Successive Approximation ADCs
The successive approximation ADC has been the mainstay of data acquisition for many
years. Recent design improvements have extended the sampling frequency of these ADCs
into the megahertz region. The Analog Devices PulSAR® family of SAR ADCs uses
internal switched capacitor techniques along with auto calibration techniques to extend
the resolution of these ADCs to 18-bits on CMOS processes without the need for
expensive thin-film laser trimming.
The basic successive approximation ADC is shown in Figure 3.56. It performs
conversions on command. On the assertion of the CONVERT START command, the
sample-and-hold (SHA) is placed in the hold mode, and all the bits of the successive
approximation register (SAR) are reset to "0" except the MSB which is set to "1". The
SAR output drives the internal DAC. If the DAC output is greater than the analog input,
this bit in the SAR is reset, otherwise it is left set. The next most significant bit is then set
to "1". If the DAC output is greater than the analog input, this bit in the SAR is reset,
otherwise it is left set. The process is repeated with each bit in turn. When all the bits
have been set, tested, and reset or not as appropriate, the contents of the SAR correspond
to the value of the analog input, and the conversion is complete. These bit "tests" can
form the basis of a serial output version SAR-based ADC.
The fundamental timing diagram for a typical SAR ADC is shown in Figure 3.57. The
end of conversion is generally indicated by an end-of-convert (EOC), data-ready
(DRDY), or a busy signal (actually, not-BUSY indicates end of conversion). The
polarities and name of this signal may be different for different SAR ADCs, but the
fundamental concept is the same. At the beginning of the conversion interval, the signal
goes high (or low) and remains in that state until the conversion is completed, at which
time it goes low (or high). The trailing edge is generally an indication of valid output
data, but the data sheet should be carefully studied—in some ADCs additional delay is
required before the output data is valid.
3.53
ANALOG-DIGITAL CONVERSION
CONVERT
START
TIMING
ANALOG
INPUT
COMPARATOR
EOC,
DRDY,
OR BUSY
SHA
CONTROL
LOGIC:
SUCCESSIVE
APPROXIMATION
REGISTER
(SAR)
DAC
OUTPUT
Figure 3.56: Basic Successive Approximation ADC
(Feedback Subtraction ADC)
SAMPLE X
SAMPLE X+1
SAMPLE X+2
CONVST
CONVERSION
TIME
TRACK/
ACQUIRE
CONVERSION
TIME
TRACK/
ACQUIRE
EOC,
BUSY
OUTPUT
DATA
DATA
X
DATA
X+1
Figure 3.57: Typical SAR ADC Timing
An N-bit conversion takes N steps. It would seem on superficial examination that a 16-bit
converter would have twice the conversion time of an 8-bit one, but this is not the case.
In an 8-bit converter, the DAC must settle to 8-bit accuracy before the bit decision is
made, whereas in a 16-bit converter, it must settle to 16-bit accuracy, which takes a lot
longer. In practice, 8-bit successive approximation ADCs can convert in a few hundred
nanoseconds, while 16-bit ones will generally take several microseconds.
3.54
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
While there are some variations, the fundamental timing of most SAR ADCs is similar
and relatively straightforward. The conversion process is generally initiated by asserting a
CONVERT START signal. The CONVST signal is a negative-going pulse whose
positive-going edge actually initiates the conversion. The internal sample-and-hold
(SHA) amplifier is placed in the hold mode on this edge, and the various bits are
determined using the SAR algorithm. The negative-going edge of the CONVST pulse
causes the EOC or BUSY line to go high. When the conversion is complete, the BUSY
line goes low, indicating the completion of the conversion process. In most cases the
trailing edge of the BUSY line can be used as an indication that the output data is valid
and can be used to strobe the output data into an external register. However, because of
the many variations in terminology and design, the individual data sheet should always be
consulted when using a specific ADC.
It should also be noted that some SAR ADCs require an external high frequency clock in
addition to the CONVERT START command. In most cases, there is no need to
synchronize the two. The frequency of the external clock, if required, generally falls in
the range of 1 MHz to 30 MHz depending on the conversion time and resolution of the
ADC. Other SAR ADCs have an internal oscillator which is used to perform the
conversions and only require the CONVERT START command. Because of their
architecture, SAR ADCs generally allow single-shot conversion at any repetition rate
from dc to the converter's maximum conversion rate.
Notice that the overall accuracy and linearity of the SAR ADC is determined primarily by
the internal DAC. Until recently, most precision SAR ADCs used laser-trimmed thin-film
DACs to achieve the desired accuracy and linearity. The thin-film resistor trimming
process adds cost, and the thin-film resistor values may be affected when subjected to the
mechanical stresses of packaging.
For these reasons, switched capacitor (or charge-redistribution) DACs have become
popular in newer SAR ADCs. The advantage of the switched capacitor DAC is that the
accuracy and linearity is primarily determined by high-accuracy photolithography, which
in turn controls the capacitor plate area and the capacitance as well as matching. In
addition, small capacitors can be placed in parallel with the main capacitors which can be
switched in and out under control of autocalibration routines to achieve high accuracy
and linearity without the need for thin-film laser trimming. Temperature tracking between
the switched capacitors can be better than 1 ppm/ºC, thereby offering a high degree of
temperature stability.
A simple 3-bit capacitor DAC is shown in Figure 3.58. The switches are shown in the
track, or sample mode where the analog input voltage, AIN, is constantly charging and
discharging the parallel combination of all the capacitors. The hold mode is initiated by
opening SIN, leaving the sampled analog input voltage on the capacitor array. Switch SC
is then opened allowing the voltage at node A to move as the bit switches are
manipulated. If S1, S2, S3, and S4 are all connected to ground, a voltage equal to –AIN
appears at node A. Connecting S1 to VREF adds a voltage equal to VREF/2 to –AIN. The
comparator then makes the MSB bit decision, and the SAR either leaves S1 connected to
VREF or connects it to ground depending on the comparator output (which is high or low
depending on whether the voltage at node A is negative or positive, respectively). A
similar process is followed for the remaining two bits. At the end of the conversion
3.55
ANALOG-DIGITAL CONVERSION
interval, S1, S2, S3, S4, and SIN are connected to AIN, SC is connected to ground, and the
converter is ready for another cycle.
BIT1
(MSB)
BIT2
SC
BIT3
(LSB)
A
CTOTAL = 2C
C
C/ 2
C/ 4
_
C/ 4
+
S1
S2
S3
S4
AIN
SIN
VREF
SWITCHES SHOWN IN TRACK (SAMPLE) MODE
Figure 3.58: 3-Bit Switched Capacitor DAC
Note that the extra LSB capacitor (C/4 in the case of the 3-bit DAC) is required to make
the total value of the capacitor array equal to 2C so that binary division is accomplished
when the individual bit capacitors are manipulated.
The operation of the capacitor DAC (cap DAC) is similar to an R-2R resistive DAC.
When a particular bit capacitor is switched to VREF, the voltage divider created by the bit
capacitor and the total array capacitance (2C) adds a voltage to node A equal to the
weight of that bit. When the bit capacitor is switched to ground, the same voltage is
subtracted from node A.
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 1). 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 3.59 where the unknown weight is 45 lbs. The balance scale
analogy is used to demonstrate the algorithm.
3.56
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
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 3.59: Successive Approximation ADC Algorithm
Early implementations of the successive approximation ADC did not use either DACs or
SARs and implemented similar functions in a variety of ways. In fact, early SAR ADCs
were referred to as sequential coders, feedback coders, or feedback subtractor coders.
The term SAR ADC came about in the 1970s when commercial successive approximation
register logic ICs such as the 2503 and 2504 became available from National
Semiconductor and Advanced Micro Devices. These devices were designed specifically
to perform the register and control functions in successive approximation ADCs and were
standard building blocks in many modular and hybrid data converters.
From a data conversion standpoint, the successive approximation ADC architecture
formed the building block for the T1 PCM carrier system and is still a popular
architecture today, but the exact origin of this architecture is not clear. It is interesting
that it did not appear in Reeves' otherwise comprehensive patent (Reference 2). Although
countless patents have been granted relating to refinements and variations on the
successive approximation architecture, they do not claim the fundamental principle.
The first mention of the successive approximation ADC architecture in the context of
PCM was by J. C. Schelleng of Bell Telephone Laboratories in a patent filed in 1946
(Reference 21). A block diagram of the 6-bit transmitting ADC reproduced from the
patent is shown in Figure 3.60. The blocks labeled selectors are key to understanding its
operation. If the differential input to a selector is greater than its designated voltage, then
the differential output of the selector is connected to its designated reference voltage, and
the corresponding binary digit is recorded as a "1". If the differential input to a selector is
less than its designated voltage, the differential output of the selector is zero, and the
corresponding binary digit is recorded as a "0". Notice that all the selectors are floating
except for the 32-V selector. For that reason, the digit grounders are required to level
shift the outputs of the floating selectors and reference them to system ground.
3.57
ANALOG-DIGITAL CONVERSION
Extracted from: John C. Schelleng, "Code Modulation Communication System,"
U.S. Patent 2,453,461, Filed June 19, 1946, Issued November 9, 1948
Figure 3.60: J. C. Schelleng's 1946 Successive Approximation ADC
The operation of the ADC follows the fundamental successive approximation algorithm.
The input signal is first tested by the 32-V selector. If it is greater than 32 V, the selector
output is set to 32 V, and a "1" is recorded for the MSB. If it is less than 32 V, the
selector output is set to 0 V, and a "0" is recorded for the MSB. The process is continued
sequentially for the remaining bits. The selectors are "stacked," i.e. the output of a given
selector is connected to one input of the following selector. Therefore the output of the
LSB selector represents a 5-bit analog approximation to the input signal.
Details of the rather cumbersome and somewhat impractical vacuum tube design of the
selectors and digit grounders are shown in Figure 3.61, also reproduced from the patent.
The battery with the label 98 is 32 V, and the battery with the label 130 is 16 V, etc.,
thereby constituting the binary weighted voltage set required to perform the conversion
algorithm.
A much more elegant implementation of the successive approximation ADC is described
by Goodall of Bell Telephone Labs in a 1947 article (Reference 22). This ADC has 5-bit
resolution and samples the voice channel at a rate of 8 kSPS. The voice signal is first
sampled, and the corresponding voltage stored on a capacitor. It is then compared to a
reference voltage which is equal to ½ the full-scale voltage. If it is greater than the
reference voltage, the MSB is registered as a "1," and an amount of charge equal to ½
scale is subtracted from the storage capacitor. If the voltage on the capacitor is less than
½ scale, then no charge is removed, and the bit is registered as a "0". After the MSB
decision is completed, the cycle continues for the second bit, but with the reference
voltage now equal to ¼ scale.
3.58
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
SELECTOR
16V
DIGIT
GROUNDER
SELECTOR, 32V
Extracted from: John C. Schelleng, "Code Modulation Communication System,"
U.S. Patent 2,453,461, Filed June 19, 1946, Issued November 9, 1948
Figure 3.61: Details of the Selectors and Digit Grounders
Both the Schelleng and the Goodall ADCs use a process of addition/subtraction of binary
weighted reference voltages to perform the SAR algorithm. Although the DAC function
is there, it is not performed using a traditional binary weighted DAC. The ADCs
described by H. R. Kaiser et. al. (Reference 23) and B. D. Smith (Reference 24) in 1953
use an actual binary weighted DAC to generate the analog approximation to the input
signal, similar to modern SAR ADCs. Smith also points out that non-linear ADC transfer
functions can be achieved by using a non-uniformly weighted DAC. This technique has
become the basis of modern companding voiceband codecs. Before this non-linear ADC
technique was developed, linear ADCs were used, and the compression and expansion
functions were performed by diode/resistor networks which had to be individually
calibrated and held at a constant temperature to prevent drift errors (Reference 25).
Of course, no discussion on ADC history would be complete without crediting the truly
groundbreaking work of Bernard M. Gordon at EPSCO (now Analogic, Incorporated).
Gordon's 1955 patent application (Reference 26) describes an all-vacuum tube 11-bit, 50kSPS successive approximation ADC—representing the first commercial offering of a
complete converter (see Figure 3.62). The DATRAC was offered in a 19" × 26" × 15"
housing, dissipated several hundred watts, and sold for approximately $8000.00.
In a later patent (Reference 27), Gordon describes the details of the logic block required
to perform the successive approximation algorithm. The SAR logic function was later
implemented in the 1970s by National Semiconductor and Advanced Micro Devices—the
popular 2502/2503/2504 family of IC logic chips. These chips were to become an integral
building block of practically all modular and hybrid successive approximation ADCs of
the 1970s and 1980s.
3.59
ANALOG-DIGITAL CONVERSION
Courtesy,
Analogic Corporation
8 Centennial Drive
Peabody, MA 01960
19" × 15" × 26"
150 lbs
http://www.analogic.com
$8,500.00
Figure 3.62: 1954 "DATRAC" 11-Bit, 50-kSPS SAR ADC
Designed by Bernard M. Gordon at EPSCO
Because of their popularity, successive approximation ADCs are available in a wide
variety of resolutions, sampling rates, input and output options, package styles, and costs.
It would be impossible to attempt to list all types, but Figure 3.63 shows a number of
recent Analog Devices' SAR ADCs which are representative. Note that many devices are
complete data acquisition systems with input multiplexers which allow a single ADC
core to process multiple analog channels.
RESOLUTION
SAMPLING
RATE
POWER
CHANNELS
AD7482
12-BITS
3.0MSPS
80mW
1
AD7484
14-BITS
3.0MSPS
80mW
1
AD7490
12-BITS
1.0MSPS
6mW
16
AD7928
12-BITS
1.0MSPS
5.4mW
8
AD974
16-BITS
0.2MSPS
120mW
4
AD7677*
16-BITS
1.0MSPS
130mW
1
AD7621*
16-BITS
3.0MSPS
100mW
1
AD7674*
18-BITS
0.8MSPS
120mW
1
* PulSAR® SERIES
Figure 3.63: Resolution / Conversion Time Comparison
for Representative Single-Supply SAR ADCs
3.60
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
An example of modern charge redistribution successive approximation ADCs is Analog
Devices' PulSAR series. The AD7677 is a 16-bit, 1-MSPS, PulSAR, fully differential,
ADC that operates from a single 5 V power supply (see Figure 3.64). The part contains a
high-speed 16-bit sampling ADC, an internal conversion clock, error correction circuits,
and both serial and parallel system interface ports. The AD7677 is hardware factory
calibrated and comprehensively tested to ensure such ac parameters as signal-to-noise
ratio (SNR) and total harmonic distortion (THD), in addition to the more traditional dc
parameters of gain, offset, and linearity. It features a very high sampling rate mode
(Warp) and, for asynchronous conversion rate applications, a fast mode (Normal) and, for
low power applications, a reduced power mode (Impulse) where the power is scaled with
the throughput. There are three speed versions of the device, the AD7675 (100 kSPS),
AD7676 (500 kSPS), and the AD7677 (1 MSPS). The latest addition to the 16-bit family
is 3-MSPS AD7621. An 18-bit family of PulSARs is also available: the AD7678 (100
kSPS), AD7679 (570 kSPS) and the AD7674 (800 kSPS).
Figure 3.64: AD7677 16-Bit 1-MSPS Switched Capacitor PulSAR® ADC
Subranging, Error Corrected, and Pipelined ADCs
Because of the sheer complexity of constructing an all-parallel flash converter using
either vacuum tubes, transistors, or tunnel diodes, the early work such as in the Staffin
and Lohman 1956 (filed) patent in Reference 9 used subranging to simplify the
conversion process. However, in order to make the subranging ADC practical, a suitable
fast sample-and-hold was required. Early subranging ADCs using vacuum tube
technology were limited by the sample-and-hold performance, but by 1964 transistors
were widely available and Gray and Kitsopoulos of Bell Labs describe pioneering work
on the classic diode-bridge sample-and-hold in their 1964 paper (Reference 28).
A basic two-stage N-bit subranging ADC is shown in Figure 3.65. The ADC is based on
two separate conversions—a coarse conversion (N1 bits) in the MSB sub-ADC (SADC)
followed by a fine conversion (N2 bits) in the LSB sub-ADC. Early subranging ADCs
nearly always used flash converters as building blocks, but a number of recent ADCs
utilize other architectures for the individual ADCs.
3.61
ANALOG-DIGITAL CONVERSION
ANALOG
INPUT
RESIDUE
SIGNAL
–
SAMPLE
AND HOLD
N1-BIT
SADC
N1-BIT
SDAC
+
Σ
G
N2-BIT
SADC
SAMPLING
CLOCK
CONTROL
OUTPUT REGISTER
N1 MSBs
DATA OUTPUT,
N2 LSBs
N-BITS = N1 + N2
See: R. Staffin and R. Lohman, "Signal Amplitude Quantizer,"
U.S. Patent 2,869,079, Filed December 19, 1956, Issued January 13, 1959
Figure 3.65: N-bit Two-Stage Subranging ADC
The conversion process begins placing the sample-and-hold in the hold mode followed by
a coarse N1-bit sub-ADC (SADC) conversion of the MSBs. The digital outputs of the
MSB converter drive an N1-bit sub-DAC (SDAC) which generates a coarsely quantized
version of the analog input signal. The N1-bit SDAC output is subtracted from the held
analog signal, amplified, and applied to the N2-bit LSB SADC. The amplifier provides
gain, G, sufficient to make the "residue" signal exactly fill the input range of the N2
SADC. The output data from the N1 SADC and the N2 SADC are latched into the output
registers yielding the N-bit digital output code, where N = N1 + N2.
In order for this simple subranging architecture to work satisfactorily, both the N1 SADC
and SDAC (although they only have N1 bits of resolution) must be better than N-bits
accurate. The residue signal offset and gain must be adjusted such that it precisely fills
the range of the N2 SADC as shown in Figure 3.66A. If the residue signal drifts by more
than 1 LSB (referenced to the N2 SADC), then there will be missing codes as shown in
Figure 3.66B where the residue signal enters the out-of-range regions labeled "X" and
"Y". Any nonlinearity or drift in the N1 SADC will also cause missing codes if it exceeds
1 LSB referenced to N-bits. In practice, an 8-bit subranging ADC with N1 = 4 bits and
N2 = 4 bits represents a realistic limit to this architecture in order to maintain no missing
codes over a reasonable operating temperature range.
When the interstage alignment is not correct, missing codes will appear in the overall
ADC transfer function as shown in Figure 3.67. If the residue signal goes into positive
overrange (the "X" region), the output first "sticks" on a code and then "jumps" over a
region leaving missing codes. The reverse occurs if the residue signal is negative
overrange.
3.62
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
0
1
2
2N1–2 2N1–1
3
(A)
IDEAL
N1 SADC
R = RANGE
OF N2 SADC
MISSING CODES
0
1
2
(B)
NONLINEAR
N1 SADC
X
R
3
Y
MISSING CODES
Figure 3.66: Residue Waveforms at Input of N2 Sub-ADC
"STICKS"
"JUMPS"
DIGITAL
OUTPUT
MISSING CODES
"JUMPS"
MISSING CODES
"STICKS"
ANALOG INPUT
Figure 3.67: Missing Codes Due to MSB SADC Nonlinearity or Interstage
Misalignment
Figure 3.68 shows a popular 8-bit 15-MSPS subranging ADC manufactured by Computer
Labs, Inc. in the mid-1970s. This converter was a basic two-stage subranging ADC with
two 4-bit flash converters—each composed of 8 dual AM687 high speed comparators.
The interstage offset adjustment potentiometer allowed the transfer function to be
optimized in the field. This ADC was popular in early digital video products such as
frame stores and time base correctors.
3.63
ANALOG-DIGITAL CONVERSION
AM687 Dual Comparators (16 Total)
RESIDUE WAVEFORM
OFFSET ADJUST
25 watts
7" × 6" × 2.5"
Figure 3.68: MOD-815, 8-Bit, 15 MSPS 4×4 Subranging ADC, 1976,
Computer Labs, Inc.
In order to reliably achieve higher than 8-bit resolution using the subranging approach, a
technique generally referred to as digital corrected subranging, digital error correction,
overlap bits, redundant bits, etc. is utilized. This method was referred to in literature as
early as 1964 by T. C. Verster (Reference 29) and quickly became widely known and
utilized (References 30-33). The fundamental concept is illustrated in Figure 3.69.
CORRECTED MSBs
–FS
000
0
001
010
011
+FS
100
101
110
111
X
+ 001 TO
N1 MSBs
R
Y
000
001
010
011
100 101
110
– 001 TO
N1 MSBs
111
UNCORRECTED MSBs
Figure 3.69: Error Correction Using Added Quantization Levels for N1 = 3
A residue waveform is shown for the specific case where N1 = 3 bits. In a standard
subranging ADC, the residue waveform must exactly fill the input range of the N2
SADC—it must stay within the region designated R. The missing code problem is solved
by adding extra quantization levels in the positive overrange region X and the negative
overrange region Y. These additional levels require additional comparators in the basic
N2 flash SADC. The scheme works as follows. As soon as the residue enters the X
3.64
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
region, the N2 SADC should return to all-zeros and start counting up again. In addition,
the code 001 is added to the output of the N1 SADC to make the MSBs read the correct
code. The figure labels the uncorrected MSB regions on the lower part of the waveform
and the corrected MSB regions on the upper part of the waveform. A similar situation
occurs when the residue waveform enters the negative overrange region Y. Here, the first
quantization level in the Y region should generate the all-ones code, and the additional
overrange comparators should cause the count to decrease. In the Y region, the code 001
must be subtracted from the MSBs to produce the corrected MSB code. It is key to
understand that in order for this correction method to work properly, the N1 SDAC must
be more accurate than the total resolution of the ADC. Nonlinearity or gain errors in the
N1 SDAC affect the amplitude of the vertical portions of the residue waveform and
therefore can produce missing codes in the output.
Horna in a 1972 paper (Reference 32) describes an experimental 8-bit 15-MSPS error
corrected subranging ADC using Motorola MC1650 dual ECL comparators as the flash
converter building blocks. Horna adds additional comparators in the second flash
converter and describes this procedure in more detail. He points out that the correction
logic can be greatly simplified by adding an appropriate offset to the residue waveform so
that there is never a negative overrange condition. This eliminates the need for the
subtraction function—only an adder is required. The MSBs are either passed through
unmodified, or 1 LSB (relative to the N1 SADC) is added to them, depending on whether
the residue signal is in range or overrange.
Modern digitally corrected subranging ADCs generally obtain the additional quantization
levels by using an internal ADC with higher resolution for the N2 SADC. For instance, if
one additional bit is added to the N2 SADC, its range is doubled—then the residue
waveform can go outside either end of the range by ½ LSB referenced to the N1 SADC.
Adding two extra bits to N2 allows the residue waveform to go outside either end of the
range by 1½ LSBs referenced to the N1 SADC. The residue waveform is offset using
Horna's technique such that only a simple adder is required to perform the correction
logic. The details of how all this works are not immediately obvious, and can best be
explained by going through an actual example of a 6-bit ADC with a 3-bit MSB SADC
and a 4-bit LSB SADC providing one bit of error correction. The block diagram of the
example ADC is shown in Figure 3.70.
After passing through an input sample-and-hold, the signal is digitized by the 3-bit
SADC, reconstructed by a 3-bit SDAC, subtracted from the held analog signal and then
amplified and applied to the second 4-bit SADC. The gain of the amplifier, G, is chosen
so that the residue waveform occupies ½ the input range of the 4-bit SADC. The 3 LSBs
of the 6-bit output data word go directly from the second SADC to the output register.
The MSB of the 4-bit SADC controls whether or not the adder adds 001 to the 3 MSBs.
The carry output of the adder is used in conjunction with some simple overrange logic to
prevent the output bits from returning to the all-zeros state when the input signal goes
outside the positive range of the ADC.
3.65
ANALOG-DIGITAL CONVERSION
OFFSET
ANALOG
INPUT
RESIDUE
SIGNAL
–
N1
3-BIT
SADC
SAMPLE
AND HOLD
N1
3-BIT
SDAC
+
Σ
N2
4-BIT
SADC
G
OFFSET
SAMPLING
CLOCK
MSB
CONTROL
ADDER ( + 001 )
CARRY
OVERRANGE LOGIC AND OUTPUT REGISTER
DATA OUTPUT
SEE: T. C. Verster, "A Method to Increase the Accuracy of Fast SerialParallel Analog-to-Digital Converters," IEEE Transactions on Electronic
Computers,EC-13, 1964, pp. 471-473
Figure 3.70: 6-Bit Subranging Error Corrected ADC, N1 = 3, N2 = 4
The residue waveform for a full-scale ramp input will now be examined in more detail to
explain how the correction logic works. Figure 3.71 shows the ideal residue waveform
assuming perfect linearity in the first ADC and perfect alignment between the two stages.
Notice that the residue waveform occupies exactly ½ the range of the N2 SADC. The 4bit digital output of the N2 SADC are shown on the left-hand side of the figure. The
regions defined by the 3-bit uncorrected N1 SADC are shown on the bottom of the figure.
The regions defined by the 3-bit corrected N1 ADC are shown are shown at the top of the
figure.
ADD 001 TO MSB ADC
N2 –FS
OUTPUT
CODE
000
001
CORRECTED MSBs
RESIDUE
0
+FS
010
011
100
101
110
111
1 111
•
•
•
1 100
1 011
•
•
•
1 000
0 111
N2
RANGE
•
•
•
0 100
0 011
•
•
•
0 001
0 000
000
001
010
011
100
101
110
111
LSBs
UNCORRECTED MSBs
Figure 3.71: 6-Bit Error Corrected Subranging ADC N1 = 3, N2 = 4,
Ideal MSB SADC
3.66
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
Following the residue waveform from left-to-right—as the input first enters the overall
ADC range at –FS, the N2 SADC begins to count up, starting at 0000. When the N2
SADC reaches the 1000 code, 001 is added to the N1 SADC output causing it to change
from 000 to 001. As the residue waveform continues to increase, the N2 SADC continues
to count up until it reaches the code 1100, at which point the N1 SADC switches to the
next level, the SDAC switches and causes the residue waveform to jump down to the
0100 output code. The adder is now disabled because the MSB of the N2 SADC is zero,
so the N1 SADC output remains 001. The residue waveform then continues to pass
through each of the remaining regions until +FS is reached.
This method has some clever features worth mentioning. First, the overall transfer
function is offset by ½ LSB referred to the MSB SADC (which is 1/16th FS referred to
the overall ADC analog input). This is easily corrected by injecting an offset into the
input sample-and-hold. It is well-known that the points at which the internal N1 SADC
and SDAC switch are the most likely to have additional noise and are the most likely to
create differential nonlinearity in the overall ADC transfer function. Offsetting them by
1/16th FS ensures that low level signals (less than ±1/16th FS) near zero volts analog
input do not exercise the critical switching points and gives low noise and excellent DNL
where they are most needed in communications applications. Finally, since the ideal
residue signal is centered within the range of the N2 SADC, the extra range provided by
the N2 SADC allows up to a ±1/16th FS error in the N1 SADC conversion while still
maintaining no missing codes.
Figure 3.72 shows a residue signal where there are errors in the N1 SADC. Notice that
there is no effect on the overall ADC linearity provided the residue signal remains within
the range of the N2 SADC. As long as this condition is met, the error correction method
described corrects for the following errors: sample-and-hold droop error, sample-andhold settling time error, N1 SADC gain error, N1 SADC offset error, N1 SDAC offset
error, N1 SADC linearity error, residue amplifier offset error. In spite of its ability to
correct all these errors, it should be emphasized that this method does not correct for gain
and linearity errors associated with the N1 SDAC or gain errors in the residue amplifier.
The errors in these parameters must be kept less than 1 LSB referred to the N-bits of the
overall subranging ADC. Another way to look at this requirement is to realize that the
amplitude of the vertical transitions of the residue waveform, corresponding to the N1
SADC and SDAC changing levels, must remain within ±½ LSB referenced to the N2
SADC input in order for the correction to prevent missing codes.
Figure 3.73 shows two methods that can be used to design a pipeline stage in a
subranging ADC. Figure 3.73A shows two pipelined stages which use an interstage T/H
in order to provide interstage gain and give each stage the maximum possible amount of
time to process the signal at its input. In Figure 3.73B a multiplying DAC is used to
provide the appropriate amount of interstage gain as well as the subtraction function.
3.67
ANALOG-DIGITAL CONVERSION
CORRECTED MSBs
ADD 001 TO MSB ADC
N2
–FS
OUTPUT
000
CODE
1 111
RESIDUE
0
001
010
011
+FS
100
101
110
111
•
•
•
1 100
1 011
•
•
•
1 000
0 111
N2
RANGE
•
•
•
0 100
0 011
•
•
•
0 001
0 000
000
001
010 011
100
101
110
111
LSBs
UNCORRECTED MSBs
Figure 3.72: 6-Bit Error Corrected Subranging ADC 1 = 3, N2 = 4,
Nonlinear MSB SADC
(A)
+
T/H
–
SADC
N1 BITS
SDAC
N1 BITS
–
+
Σ
+
T/H,
G
–
+
SADC
N2 BITS
SDAC
N2 BITS
–
Σ
TO ERROR CORRECTING LOGIC
(B)
+
T/H
–
SADC
N1 BITS
S MDAC
N1 BITS
+
T/H
–
SADC
N2 BITS
S MDAC
N2 BITS
TO ERROR CORRECTING LOGIC
Figure 3.73: Generalized Pipeline Stages in a
Subranging ADC with Error Correction
The term "pipelined" architecture refers to the ability of one stage to process data from
the previous stage during any given clock cycle. At the end of each phase of a particular
clock cycle, the output of a given stage is passed on to the next stage using the T/H
functions and new data is shifted into the stage. Of course this means that the digital
outputs of all but the last stage in the "pipeline" must be stored in the appropriate number
3.68
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
of shift registers so that the digital data arriving at the correction logic corresponds to the
same sample.
Figure 3.74 shows a timing diagram of a typical pipelined subranging ADC. Notice that
the phases of the clocks to the T/H amplifiers are alternated from stage to stage such that
when a particular T/H in the ADC enters the hold mode it holds the sample from the
preceding T/H, and the preceding T/H returns to the track mode. The held analog signal
is passed along from stage to stage until it reaches the final stage in the pipelined ADC—
in this case, a flash converter. When operating at high sampling rates, it is critical that the
differential sampling clock be kept at a 50% duty cycle for optimum performance. Duty
cycles other than 50% affect all the T/H amplifiers in the chain—some will have longer
than optimum track times and shorter than optimum hold times; while others suffer
exactly the reverse condition. Several newer pipelined ADCs including the 12-bit, 65MSPS AD9235 and the 12-bit, 170-/210-MSPS AD9430 have on-chip clock conditioning
circuits to control the internal duty cycle while allowing some variation in the external
clock duty cycle.
CLOCK
CLOCK
VIN
T
H
INPUT
T/H
T
H
STAGE 1
T/H
T
H
STAGE 2
T/H
T
H
STAGE 3
T/H
H
FLASH
CLOCK
INPUT T/H
T
H
T
H
T
H
T
H
T
STAGE 1
H
T
H
T
H
T
H
T
H
STAGE 2
T
H
T
H
T
H
T
H
T
STAGE 3
H
T
H
T
H
T
H
T
H
FLASH
T
H
T
H
T
H
T
H
T
DATA OUT
DATA OUT
DATA OUT
DATA OUT
PIPELINE DELAY
Figure 3.74: Clock Issues in Pipelined ADCs
The effects of the "pipeline" delay (sometimes called latency) in the output data are
shown in Figure 3.75 for the AD9235 12-bit 65-MSPS ADC where there is a 7-clock
cycle pipeline delay.
Note that the pipeline delay is a function of the number of stages and the particular
architecture of the ADC under consideration—the data sheet should always be consulted
for the exact details of the relationship between the sampling clock and the output data
timing. In many applications the pipeline delay will not be a problem, but if the ADC is
inside a feedback loop the pipeline delay may cause instability. The pipeline delay can
also be troublesome in multiplexed applications or when operating the ADC in a "single3.69
ANALOG-DIGITAL CONVERSION
shot" mode. Other ADC architectures—such as successive approximation—may be better
suited to these types of applications.
PIPELINE DELAY (LATENCY) = 7 CLOCK CYCLES
Figure 3.75: Typical Pipelined ADC Timing for AD9235 12-Bit, 65-MSPS ADC
It is often erroneously assumed that all subranging ADCs are pipelined, and that all
pipelined ADCs are subranging. While it is true that most modern subranging ADCs are
pipelined in order to achieve the maximum possible sampling rate, they don't necessarily
have to be pipelined if designed for use at much lower speeds. For instance, the leading
edge of the sampling clock could initiate the conversion process, and any additional clock
pulses required to continue the conversion could be generated internal to the ADC using
an on-chip timing circuit. At the end of the conversion process, an end-of-conversion or
data-ready signal could be generated as an external indication that the data corresponding
to that particular sampling edge is valid.
Conversely, there are some ADCs which use other architectures than subranging and are
pipelined. For instance, many flash converters use an extra set of output latches (in
addition to the latch associated with the parallel comparators) which introduces pipeline
delay in the output data. Another example of a non-subranging architecture which
generally has quite a bit of pipeline delay is sigma-delta which will be covered in more
detail in the next section of this chapter. Note, however, that it is possible to modify the
timing of a normal sigma-delta ADC, reduce the output data rate, and make a "no
latency" sigma-delta ADC.
The pipelined error correcting ADC has become very popular in modern ADCs requiring
wide dynamic range and low levels of distortion. There are many possible ways to design
a pipelined ADC, and we will now look at just a few of the tradeoffs. Figure 3.76A shows
a pipelined ADC designed with identical stages of k-bits each. This architecture uses the
same core hardware in each stage, offers a few other advantages, but does necessarily
optimize the ADC for best possible performance. Figure 3.76B shows the simplest form
of this architecture where k = 1.
3.70
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
In order to optimize performance at the 12-bit level, for example, 1-bit per stage pipeline
is more commonly used with a multibit front-end and back-end ADC as shown in Figure
3.77.
+
VIN
k-BIT
ADC
T/H
+
Σ
2k
k-BIT
ADC
T/H
Σ
–
(A)
k-BITS
PER STAGE
T/H
–
k-BIT
DAC
k
2k
k-BIT
ADC
k-BIT
DAC
k
k
MSBs
LSBs
+
VIN
1-BIT
ADC
T/H
+
Σ
2
1-BIT
ADC
T/H
Σ
–
(B)
1-BIT
PER STAGE
2
T/H
–
1-BIT
DAC
1-BIT
ADC
1-BIT
DAC
MSB
LSB
Figure 3.76: Basic Pipelined ADC with Identical Stages
+
VIN
k-BIT
ADC
T/H
Σ
2k
–
k-BIT
DAC
k
MSBs
+
T/H
1-BIT
ADC
Σ
+
2
T/H
1-BIT
ADC
–
1-BIT
DAC
Σ
2
T/H
m-BIT
ADC
–
1-BIT
DAC
LSBs
1-BIT PIPELINED STAGES
Figure 3.77: Multibit and 1-Bit Pipelined Core Combined
3.71
ANALOG-DIGITAL CONVERSION
Error correction is used in practically all pipelined ADCs, including the simple 1-bit
stage. Figure 3.78 shows how an ADC constructed of uncorrected cascaded 1-bit stages
will ultimately result in missing codes unless each stage is nearly ideal.
+
VIN
1-BIT
ADC-1
T/H
Σ
+
2
T/H
1-BIT
ADC-2
Σ
–
–
1-BIT
DAC
0
1
IDEAL ADC-1
2
1-BIT
DAC
RESIDUE
WAVEFORMS
0
1
MISSING CODES
OFFSET ERROR IN ADC-1
Figure 3.78: 1-Bit per Stage Pipelined ADC
Error correction can be added to the simple 1-bit stage by adding a single extra
comparator—resulting in what is commonly referred to as a "1.5-bit" stage as shown in
Figure 3.79. Details of this architecture can be found in Reference 34. The two
comparators have three possible output codes: 00, 01, and 10. Note that three parallel
comparators form a complete 2-bit stage—which would be required for the final stage in
a pipelined 1.5-bit ADC, as one additional output level is required to generate the 11
code.
It would appear at first glance that simply adding a single comparator would not allow
the error correction scheme to work as previously illustrated for the multi-bit cases. The
explanation is as follows. The thresholds of the first 1.5-bit stage are designated M1, M2,
and the thresholds of the second 1.5-bit stage are designated L1, L2. The residue
waveform driving the second 1.5-bit stage is shown for an ideal first stage. The
"corrected" codes for the first stage are obtained by simply adding the output code of the
second stage to that of the first stage. Note that the correct output code is obtained as long
as the residue signal falls within the dotted box—a third comparator is not required
(except for the last stage which must be 2-bits or greater). The final bit decision for the
first stage is made based on the corrected bits as follows: a 00, 01, or 10 code indicates a
"0", and a 11 code indicates a "1".
3.72
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
+
VIN
1.5-BIT
ADC-1
M1,M2
T/H
THRESHOLDS = M1, M2
Σ
2
T/H
–
1.5-BIT
DAC
00, 01, 10
NOTE: 1.5-BIT ADC =
TWO COMPARATORS
+
1.5-BIT
ADC-2
L1, L2
RESIDUE
Σ
1.5-BIT
DAC
00, 01, 10
THRESHOLDS = L1, L2
ADC-1 CORRECTED BITS
1
0
00
01
10
RESIDUE
11
10
ADC-2
OUTPUT
CODES
2
–
L2
01
M1
01
00
00
ADC-2
INPUT
RANGE
M2
01
10
L1
ADC-1 UNCORRECTED BITS
Figure 3.79: 1.5-Bit per Stage Pipeline Showing Error Corrected Range
Figure 3.80 shows the effects of errors made in the first stage converter. Errors of up to
±½ LSB in the first stage can be corrected by this method. Figure 3.81 shows the effects
of error correction for a different set of first stage errors which still fall within the ±½
LSB correction range .
ADC-1 CORRECTED BITS
1
0
00
ADC-2
OUTPUT
CODES
01
10
11
10
L2
01
01
00
00
01
M1
10
ADC-2
INPUT
RANGE
L1
M2
ADC-1 UNCORRECTED BITS
Figure 3.80: Residue Waveform for 1.5-Bit ADC-2 Stage Input
with Nonlinear ADC-1 Stage, Case 1
3.73
ANALOG-DIGITAL CONVERSION
ADC-1 CORRECTED BITS
1
0
00
ADC-2
OUTPUT
CODES
01
10
11
10
L2
01
01
00
00
01
M1
ADC-2
INPUT
RANGE
L1
10
M2
ADC-1 UNCORRECTED BITS
Figure 3.81: Residue Waveform for 1.5-Bit ADC-2 Stage Input
with Nonlinear ADC-1 Stage, Case 2
Another less popular type of error corrected subranging architecture is the recirculating
subranging ADC. This is shown in Figure 3.82 and was proposed in a 1966 paper by
Kinniment, et.al. (Reference 31). The concept is similar to the error corrected subranging
architecture previously discussed, but in this architecture, the residue signal is
recirculated through a single ADC and DAC stage using switches and a programmable
gain amplifier (PGA). Figure 3.82 shows the additional buffer registers required to store
the pipelined data resulting in each conversion such that the data into the correction logic
(adder) corresponds to the same sample. To put things into historical perspective, Figure
3.83 from Kinniment's paper shows a pipelined error corrected subranging architecture
identical to those popular in many of today's ADCs.
ANALOG
INPUT
–
SAMPLE
AND HOLD
+
Σ
PGA
SAMPLING
CLOCK
3-BIT
FLASH
ADC
3-BIT
DAC
REG.
CONTROL
Adapted from:
D.J. Kinniment, D. Aspinall, and D.B.G. Edwards,
"High-Speed Analogue-Digital Converter,"
IEE Proceedings, Vol. 113, pp. 2061-2069, Dec. 1966
REG.
REG.
REG.
REG.
REG.
ADDER
OUTPUT REGISTER
DATA OUTPUT
Figure 3.82: Kinniment, et. al., 1966 Pipelined 7-bit, 9-MSPS
Recirculating ADC Architecture
3.74
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
DELAY
ANALOG
INPUT
SAMPLE
AND
HOLD
3-BIT
DELAY
–
3-BIT +
FLASH
DAC
ADC
Σ
3-BIT
3-BIT +
FLASH
DAC
ADC
–
Σ
3-BIT
FLASH
ADC
REG.
SAMPLING
CLOCK
CONTROL
REG.
REG.
REG,
REG.
REG.
ADDER
OUTPUT REGISTER
DATA OUTPUT
Figure 3.83: Kinniment, et. al., 1966 Proposed Pipelined 7-bit ADC Architecture
The discussion of error corrected pipelined ADCs concludes with a few examples of
modern integrated circuit implementations of the popular architecture. These examples
show the flexibility of the technique in optimizing ADC performance at different
resolutions, sampling rates, power dissipation, etc.
The demand for wide dynamic range high speed ADCs suitable for communications
applications led to the development of a breakthrough product in 1995, the AD9042 12bit, 41-MSPS ADC (see Reference 35). A block diagram of the converter is shown in
Figure 3.84.
DELAY
DELAY
ADCs ARE GRAY CODE MagAMP™
Figure 3.84: AD9042 12-Bit 41-MSPS ADC, 1995
3.75
ANALOG-DIGITAL CONVERSION
The AD9042 uses an error corrected subranging architecture composed of a 6-bit MSB
ADC/DAC followed by a 7-bit LSB ADC and uses one bit of error correction. The
AD9042 yields 80-dB SFDR performance over the Nyquist bandwidth at a sampling rate
of 41 MSPS. Fabricated on a high speed complementary bipolar process, the device
dissipates 600 mW and operates on a single +5-V supply.
In order to meet the need for lower cost, lower power devices, Analog Devices initiated a
family of CMOS high performance ADCs such as the AD9225 12-bit, 25-MSPS ADC
released in 1998. The AD9225 dissipates 280 mW, has 85-dB SFDR, and operates on a
single +5 V supply. A simplified diagram of the AD9225 is shown in Figure 3.85.
Figure 3.85: AD9225 12-Bit 25-MSPS CMOS ADC, 1998
The AD9235 12-bit 65-MSPS CMOS ADC released in 2001 shows the progression of
CMOS high performance converters. The AD9235 operates on a single +3 V supply,
dissipates 300 mW (at 65 MSPS), and has a 90-dB SFDR over the Nyquist bandwidth.
The ADC uses 8 stages of 1.5-bit converters (2 comparators) previously described earlier
in this section of the book. A simplified diagram of the AD9235 is shown in Figure 3.86.
The 12-bit 210-MSPS AD9430 released in 2002 is shown in Figure 3.87 and is fabricated
on a BiCMOS process, has 80-dB SFDR up to 70-MHz inputs, operates on a single +3 V
supply and dissipates 1.3 W at 210 MSPS. Output data is provided on 2 ports at 105
MSPS each in the CMOS mode or on a single port at 210 MSPS in the LVDS (low
voltage digital signal) mode.
3.76
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
Figure 3.86: AD9235 12-Bit, 65-MSPS CMOC ADC, 2001
Figure 3.87: AD9430 12-Bit, 170-/210-MSPS ADC, 2002
The final example (certainly not to imply the conclusion of a complete listing!) is the 14bit 105-MSPS AD6645 ADC released in 2003 and fabricated on a high speed
complementary bipolar process (XFCB), has 90-dB SFDR, operates on a single +5 V
supply and dissipates 1.5 W. The 80-MSPS version of the AD6645 was released in 2002.
A simplified diagram of the AD6645 is shown in Figure 3.88.
3.77
ANALOG-DIGITAL CONVERSION
DELAY
DELAY
DELAY
DELAY
ADC1, ADC2, AND ADC3 ARE GRAY CODE MagAMP™
Figure 3.88: AD6645 14-Bit, 105-MSPS ADC,
2003 (80-MSPS Version Released in 2002)
Achieving the level of performance shown in the above examples is by no means easy. A
variety of circuit and process design tradeoffs must be made. Circuit design requires fastsettling op amps, track-and-hold amplifiers, linear DACs with low glitch, careful
attention to chip layout, etc. Bit shuffling using a thermometer DAC architecture is often
used in the internal DAC structures as well as the addition of dither signals to increase the
SFDR performance. In addition to the design of the actual IC, the system designer must
use extremely good layout, grounding, and decoupling techniques in order to achieve
specified performance. Many of these important hardware design techniques are covered
in detail in Chapters 6 and 9 of this book.
Serial Bit-Per-Stage Binary and Gray Coded (Folding) ADCs
Various architectures exist for performing A/D conversion using one stage per bit. Figure
3.89 shows the overall concept. In fact, a multistage subranging ADC with one bit per
stage and no error correction is one form as previously discussed. In this approach, the
input signal must be held constant during the entire conversion cycle. There are N stages,
each of which have a bit output and a residue output. The residue output of one stage is
the input to the next. The last bit is detected with a single comparator as shown.
One of the first references to these architectures appeared in an article by B. D. Smith in
1956 (Reference 36). Smith indicates, however, that previous work had been done at
M.I.T. by R. P. Sallen in a 1949 thesis. In the article, Smith describes both the binary and
the Gray (or folding) transfer functions required to implement the A/D conversion.
3.78
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
VREF
ANALOG
INPUT
STAGE
1
SHA
R1
BIT 1
MSB
STAGE
2
R2
STAGE
N-1
BIT 2
+
-
BIT N-1
BIT N
LSB
DECODE LOGIC AND OUTPUT REGISTERS
N
B. D. Smith, "An Unusual Electronic Analog-Digital Conversion Method,"
IRE Transactions on Instrumentation, June 1956, pp. 155-160.
Figure 3.89: Generalized Bit-Per-Stage ADC Architecture
The basic stage for performing a single binary bit conversion is shown in Figure 3.90. It
consists of a gain-of-two amplifier, a comparator, and a 1-bit DAC (changeover switch).
Assume that this is the first stage of the ADC. The MSB is simply the polarity of the
input, and that is detected with the comparator which also controls the 1-bit DAC. The 1bit DAC output is summed with the output of the gain-of-two amplifier. The resulting
residue output is then applied to the next stage. In order to better understand how the
circuit works, the diagram shows the residue output for the case of a linear ramp input
voltage which traverses the entire ADC range, –VR to +VR. Notice that the polarity of the
residue output determines the binary bit output of the next stage.
+VR
RESIDUE
INPUT
Σ
G=2
INPUT
t
0
+
-VR
-
+VR
+VR
SWITCH POSITION
SHOWN FOR
NEGATIVE INPUT
BIT OUTPUT
(BINARY CODE)
-VR
RESIDUE
0
t
-VR
Figure 3.90: Single-Stage Transfer Function for Binary ADC
3.79
ANALOG-DIGITAL CONVERSION
A simplified 3-bit serial-binary ADC is shown in Figure 3.91, and the residue outputs are
shown in Figure 3.92. Again, the case is shown for a linear ramp input voltage whose
range is between –VR and +VR. Each residue output signal has discontinuities which
correspond to the point where the comparator changes state and causes the DAC to
switch. The fundamental problem with this architecture is the discontinuity in the residue
output waveforms. Adequate settling time must be allowed for these transients to
propagate through all the stages and settle at the final comparator input. As presented
here, the prospects of making this architecture operate at high speed are dismal. However
using the 1.5-bit-per stage pipelined architecture previously discussed in this section
makes it much more attractive at high speeds.
±VR
ANALOG
INPUT
R1
R2
STAGE
2
STAGE
1
SHA
BIT 1
+
-
BIT 2
BIT 3
OUTPUT REGISTER
3
Figure 3.91: 3-bit Serial ADC with Binary Output
+VR
INPUT
t
0
–VR
R1
+VR
+VR
0
t
–VR
–VR
+VR
R2
+VR
0
t
–VR
–VR
BINARY
CODE
000
001
010
011
100
101
110
111
Figure 3.92: Input and Residue Waveforms of 3-Bit Binary Ripple ADC
3.80
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
Although the binary method is discussed in his paper, B. D. Smith also describes a much
preferred bit-per-stage architecture based on absolute value amplifiers (magnitude
amplifiers, or simply MagAMPs™). This scheme has often been referred to as serialGray (since the output coding is in Gray code), or folding converter because of the shape
of the transfer function. Performing the conversion using a transfer function that produces
an initial Gray code output has the advantage of minimizing discontinuities in the residue
output waveforms and offers the potential of operating at much higher speeds than the
binary approach.
The basic folding stage is shown functionally in Figure 3.93 along with its transfer
function. The input to the stage is assumed to be a linear ramp voltage whose range is
between –VR and +VR. The comparator detects the polarity of the input signal and
provides the Gray bit output for the stage. It also determines whether the overall stage
gain is +2 or –2. The reference voltage VR is summed with the switch output to generate
the residue signal which is applied to the next stage. The polarity of the residue signal
determines the Gray bit for the next stage. The transfer function for the folding stage is
also shown in Figure 3.93.
VR
INPUT
+VR
G = +2
INPUT
Σ
RESIDUE
0
t
G = -2
SWITCH POSITION
SHOWN FOR
NEGATIVE INPUT
-VR
+VR
+
RESIDUE
0
t
BIT OUTPUT
(GRAY CODE)
-VR
Figure 3.93: Folding Stage Functional Equivalent Circuit
A 3-bit MagAMP folding ADC is shown in Figure 3.94, and the corresponding residue
waveforms in Figure 3.95. As in the case of the binary ripple ADC, the polarity of the
residue output signal of a stage determines the value of the Gray bit for the next stage.
The polarity of the input to the first stage determines the Gray MSB; the polarity of R1
output determines the Gray bit-2; and the polarity of R2 output determines the Gray bit-3.
Notice that unlike the binary ripple ADC, there is no abrupt transition in any of the
folding stage residue output waveforms. This makes operation at high speeds quite
feasible.
3.81
ANALOG-DIGITAL CONVERSION
±VR
ANALOG
INPUT
MAGAMP
2
MAGAMP
1
SHA
BIT 1
+
-
BIT 2
BIT 3
GRAY CODE REGISTER
3
GRAY-TO-BINARY CONVERTER
3
OUTPUT REGISTER
3
Figure 3.94: 3-bit Folding ADC Block Diagram
+VR
INPUT
t
0
–VR
R1
+VR
0
t
–VR
–VR
+VR
R2
+VR
0
t
–VR
GRAY
CODE
–VR
–VR
000
001
011
010
110
111
101
100
Figure 3.95: Input and Residue Waveforms for 3-Bit Folding ADC
The key to operating this architecture at high speeds is the folding stage. N. E. Chasek of
Bell Telephone Labs describes a circuit for generating the folding transfer function using
nested diode bridges in a patent filed in 1960 (Reference 37). This circuit made use of
solid-state devices, but required different reference voltages for each stage (see Figure
3.96). Chasek's circuit also suffered from loss of headroom and gain when several stages
were cascaded to form higher resolution converters as shown in Figure 3.97. What is
really needed to make the folding ADC work at high resolutions is nearly ideal voltage or
current rectification.
3.82
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
GRAY CODE OUTPUT
BIT 2
BIT 1
BIT 3
BIAS
BIAS
+
+
BALANCED
INPUT
R1
R2
+
+
BIAS
BIAS
Adapted from: N. E. Chasek, "Pulse Code Modulation Encoder,"
U.S. Patent 3,035,258, Filed November 14, 1960, Issued May 15, 1962
Figure 3.96: 3-Bit Folding ADC Based on N. E. Chasek's Design
+VR
INPUT
t
0
–VR
R1
+VR/2
0
t
–VR/2
–VR/2
+VR/4
R2
0
–VR/4
t
–VR/4
–VR
GRAY
CODE
000
001
011
010
110
111
101
100
Figure 3.97: Single-Ended Waveforms in Chasek's Folding ADC
3.83
ANALOG-DIGITAL CONVERSION
F. D. Waldhaur of Bell Telephone Labs remedied the problems of Chasek's nested diode
bridge circuits in a classic patent filed in 1962 (Reference 38). Figure 3.98 shows
Waldhaur's elegant implementation of the folding transfer function using solid state op
amps with diodes in the feedback loop. The gain-of-two op amps allow the same
reference voltages to be used for each stage and maintain the same signal level at each
residue output with nearly ideal rectification.
R
2R
IOFFSET
I
R
–
+
i
2R
BALANCED
INPUT
i
i
2i
R
2i – I
2i
BIT
OUTPUT
R
2R
BALANCED
OUTPUT
2i
i
3i
–
R
3i
2i
2i – I
+
2R
R
IOFFSET
I
Extracted from: F. D. Waldhauer, "Analog-to-digital Converter,"
U.S. Patent 3,187,325, Filed July 2, 1962, Issued June 1, 1965
Figure 3.98: F. D. Waldhaur's Classic Folding Stage using Rectifier Amplifiers
J. O. Edson and H. H. Henning describe the operation and performance of this type of
ADC in greater detail in a 1965 Bell System Technical Journal article (Reference 39). An
operational 9-bit, 6-MSPS ADC of this type was used in experimental studies on 224Mbit/second PCM terminals. These terminals were supposed to handle data as well as
voice signals. The voiceband objective was to digitize an entire 600-channel, 2.4-MHz
FDM band, therefore requiring a minimum sampling rate of approximately 6 MSPS.
It is interesting to note that the experimental terminal was also supposed to handle video
as well, which required a higher sampling rate of approximately 12-MSPS. For this
requirement, the latest (and final) generation Bell Labs' electron beam coder was needed
to meet the ADC requirement, as the solid-state coder based on Waldhaur's patent did not
have the necessary accuracy at the higher sampling rates.
The first commercial ADC using Waldhaur's Gray code architecture was the 8-bit, 10MSPS HS-810 from Computer Labs, Inc., in 1966 (see Chapter 1 of this book). The
instrument used all discrete transistor circuits (no ICs) and was designed to be mounted in
a 19" rack.
The folding Gray code architecture was used in a few instrument and modular ADCs in
the early 1970s, but was largely replaced by the error-corrected subranging architecture.
With improvements in IC processes, there was renewed interest in the folding
3.84
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
architecture in the late 1970s and throughout the 1980s—with quite a number of designs
reported in the various journals over the period (References 40-44).
Analog Devices developed the first high speed fully complementary bipolar (CB) process
in the mid-1980s, and in 1994 Frank Murden and Carl Moreland filed patents on a
significantly improved current-steering architecture for a Gray code MagAMP-based
ADC (References 45-49). The technique was first implemented for building block cores
in the AD9042 12-bit, 41-MSPS ADC released in 1995, and refinements of the technique
and a higher speed CB process, XFCB, (References 50 and 51) pushed the core
technology to 14-bits with the release of the AD6644 14-bit 65-MSPS ADC in 1999, the
AD6645 14-bit 80-MSPS ADC in 2001, and a 105-MSPS version of the AD6645 in
2003. Although these ADCs use the error-corrected subranging architecture, the internal
building block core ADCs utilize the MagAMP architecture.
Modern IC circuit designs implement the transfer function using current-steering openloop gain techniques which can be made to operate much faster. Fully differential stages
(including the SHA) also provide speed, lower distortion, and yield 8-bit accurate folding
stages with no requirement for thin film resistor laser trimming.
An example of a fully differential gain-of-two MagAMP folding stage is shown in Figure
3.99 (see References 45, 46, 48). The differential input signal is applied to the
degenerated-emitter differential pair Q1,Q2 and the comparator. The differential input
voltage is converted into a differential current which flows in the collectors of Q1, Q2. If
+IN is greater than –IN, cascode-connected transistors Q3, Q6 are on, and Q4, Q6 are off.
The differential signal currents therefore flow through the collectors of Q3, Q6 into levelshifting transistors Q7, Q8 and into the output load resistors, developing the differential
output voltage between +OUT and –OUT. The overall differential voltage gain of the
circuit is two.
2I - IOFF
2I + IOFF
Q7
VBIAS
Q3
Q8
VBIAS
Q4
Q5
Q6
+OUT
-OUT
i+
GRAY
GRAY
-
+
+IN
i-
-IN
Q2
Q1
R
I
R
R
I
Figure 3.99: A Modern Current-Steering MagAMP™ Stage
3.85
ANALOG-DIGITAL CONVERSION
If +IN is less than –IN (negative differential input voltage), the comparator changes state
and turns Q4, Q5 on and Q3, Q6 off. The differential signal currents flow from Q5 to Q7
and from Q4 to Q8, thereby maintaining the same relative polarity at the differential
output as for a positive differential input voltage. The required offset voltage is developed
by adding a current IOFF to the emitter current of Q7 and subtracting it from the emitter
current of Q8.
The differential residue output voltage of the stage drives the next stage input, and the
comparator output represents the Gray code output for the stage.
The MagAMP architecture offers lower power and can be extended to sampling rates
previously dominated by flash converters. For example, the AD9054A 8-bit, 200-MSPS
ADC is shown in Figure 3.100. The first five bits (Gray code) are derived from five
differential MagAMP stages. The differential residue output of the fifth MagAMP stage
drives a 3-bit flash converter, rather than a single comparator.
The Gray-code output of the five MagAMPs and the binary-code output of the 3-bit flash
are latched, all converted into binary, and latched again in the output data register.
Because of the high data rate, a demultiplexed output option is provided.
ANALOG
INPUT
SHA
DIFFERENTIAL
OUTPUTS ON
BITS 1 - 5
MAGAMP
1
MAGAMP
2
MAGAMP
3
MAGAMP
4
MAGAMP
5
BIT
1
GRAY
BIT
2
GRAY
BIT
3
GRAY
BIT
4
GRAY
BIT
5
GRAY
3-BIT
FLASH
ADC
3
BINARY
REGISTER
5
GRAY-TO-BINARY CONVERTER
5
OUTPUT REGISTERS
8
Figure 3.100: AD9054A 8-bit, 200-MSPS ADC Functional Diagram
3.86
3
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
A summary of the timeline for the popular high speed ADC architectures discussed in
this section is shown in Figure 3.101 to put things into a historical perspective.
Reeve’s counting ADC
1939
Successive approximation
1946
Flash (electron tube coders)
1948
Bit-per-stage (binary and folding-Gray)
1956
Subranging
1956
Subranging with error correction
1964
Pipeline with error correction
1966
Note: Dates are first publications or patent filings
Figure 3.101: High Speed ADC Architecture Timeline
Counting and Integrating ADC Architectures
Although counting-based ADCs are not well suited for high speed applications, they are
ideal for high resolution low frequency applications, especially when combined with
integrating techniques. The evolution of counting-based ADCs follows the approximate
timeline shown in Figure 3.102. Notice that the development of the various
counting/integrating ADC architectures runs roughly in parallel with the development of
the higher speed architectures (refer back to Figure 3.101). By 1957 (with the exception
of triple and quad slope) all the fundamental architectures had been proposed in one form
or another.
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 3.102: Counting and Integrating ADC Architecture Timeline
3.87
ANALOG-DIGITAL CONVERSION
A. H. Reeves' 5-Bit Counting ADC
As previously discussed, the first ADC suitable for PCM applications was the one
documented by A. H. Reeves in his comprehensive 1939 PCM patent (Reference 24). A
simplified diagram of the ADC is repeated here in Figure 3.103. The early ADCs for
PCM typically had 5-7 bits of resolution and sampling rates of 6-10 kSPS. Interestingly
enough, Reeves' ADC was based on a counting technique, probably because of his
general interests in counters—the Eccles-Jordan bistable multivibrator had been invented
only a few years earlier. However other architectures such as successive approximation,
flash, bit-per-stage, subranging, and pipeline were much more widely used in later PCM
applications.
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 3.103: A. H. Reeves' 5-bit Counting ADC
The counting ADC technique (see Figure 3.103) 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 which 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 was used, and a 100:1 divider generated 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 full-scale signal. The technique can obviously be extended
to higher resolutions.
3.88
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
Charge Run-Down ADC
The charge run-down ADC architecture (see Reference 52) shown in Figure 3.104 first
samples the analog input and stores the voltage on a fixed capacitor. The capacitor is then
discharged with a constant current source, and the time required for complete discharge is
measured using a counter. Notice that in this approach, the overall accuracy is dependent
on the quality and magnitude of the capacitor, the magnitude of the current source, as
well as the accuracy of the timebase.
ANALOG
INPUT
+
–
C
I
COUNTER
CK
R
DIGITAL
OUTPUT
TIMER
AND
CONTROL
OSC.
Figure 3.104: Charge Run-Down ADC
Ramp Run-Up ADC
In the ramp run-up architecture shown in Figure 3.105 (see Reference 53), a ramp
generator is started at the beginning of the conversion cycle. The counter then measures
the time required for the ramp voltage to equal the analog input voltage. The counter
output is therefore proportional to the value of the analog input. In an alternate version
(shown dotted in Figure 3.105), the ramp voltage generator is replaced by a DAC which
is driven by the counter output. The advantage of using the ramp is that the ADC is
always monotonic, whereas overall monotonicity is determined by the DAC when it is
used as a substitute.
The accuracy of the ramp run-up ADC depends on the accuracy of the ramp generator (or
the DAC) as well as the oscillator.
3.89
ANALOG-DIGITAL CONVERSION
ANALOG
INPUT
+
–
* ALTERNATE
DAC
COUNTER
CK
R
* RAMP
GENERATOR
DIGITAL
OUTPUT
TIMER
AND
CONTROL
OSC.
Figure 3.105: Ramp Run-Up ADC
Tracking ADC
The tracking ADC architecture shown in Figure 3.106 (see References 54 and 55)
continually compares the input signal with a reconstructed representation of the input
signal. The up/down counter is controlled by the comparator output. If the analog input
exceeds the DAC output, the counter counts up until they are equal. If the DAC output
exceeds the analog input, the counter counts down until they are equal. It is evident that if
the analog input changes slowly, the counter will follow, and the digital output will
remain close to its correct value. If the analog input suddenly undergoes a large step
change, it will be many hundreds or thousands of clock cycles before the output is again
valid. The tracking ADC therefore responds quickly to slowly changing signals, but
slowly to a quickly changing one.
ANALOG
INPUT
+
U/D
–
UP/DOWN
COUNTER
OSC.
CK
DAC
Figure 3.106: Tracking ADC
3.90
DIGITAL
OUTPUT
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
The simple analysis above ignores the behavior of the ADC when the analog input and
DAC output are nearly equal. This will depend on the exact nature of the comparator and
counter. If the comparator is a simple one, the DAC output will cycle by 1 LSB from just
above the analog input to just below it, and the digital output will, of course, do the same
—there will be 1 LSB of flicker. Note that the output in such a case steps every clock
cycle, irrespective of the exact value of analog input, and hence always has unity
Mark/Space ratio. In other words, there is no possibility of taking a mean value of the
digital output and increasing resolution by oversampling.
A more satisfactory, but more complex arrangement would be to use a window
comparator with a window 1-2 LSB wide. When the DAC output is high or low the
system behaves as in the previous description, but if the DAC output is within the
window, the counter stops. This arrangement eliminates the flicker, provided that the
DAC DNL never allows the DAC output to step across the window for 1 LSB change in
code.
Tracking ADCs are not very common. Their slow step response makes them unsuitable
for many applications, but they do have one asset: their output is continuously available.
Most ADCs perform conversions: i.e., on receipt of a "start convert" command (which
may be internally generated), they perform a conversion and, after a delay, a result
becomes available. Providing that the analog input changes slowly, the output of a
tracking ADC is always available. This is valuable in synchro-to-digital and resolver to
digital converters (SDCs and RDCs), and this is the application where tracking ADCs are
most often used. Another valuable characteristic of tracking ADCs is that a fast transient
on the analog input causes the output to change only one count. This is very useful in
noisy environments. Notice the similarity between a tracking ADC and a successive
approximation ADC. Replacing the up/down counter with SAR logic yields the
architecture for a successive approximation ADC.
Voltage-to-Frequency Converters (VFCs)
A voltage-to-frequency converter (VFC) is an oscillator whose frequency is linearly
proportional to a control voltage. The VFC/counter ADC is monotonic and free of
missing codes, integrates noise, and can consume very little power. It is also very useful
for telemetry applications, since the VFC, which is small, cheap and low-powered can be
mounted on the experimental subject (patient, wild animal, artillery shell, etc.) and
communicate with the counter by a telemetry link as shown in Figure 3.107.
There are two common VFC architectures: the current- steering multivibrator VFC and
the charge-balance VFC (Reference 56). The charge-balanced VFC may be made in
asynchronous or synchronous (clocked) forms. There are many more VFO (variable
frequency oscillator) architectures, including the ubiquitous 555 timer, but the key feature
of VFCs is linearity—few VFOs are very linear.
The current-steering multivibrator VFC is actually a current-to-frequency converter
rather than a VFC, but, as shown in Figure 3.108, practical circuits invariably contain a
voltage-to-current converter at the input. The principle of operation is evident: the current
discharges the capacitor until a threshold is reached, and when the capacitor terminals are
reversed, the half-cycle repeats itself. The waveform across the capacitor is a linear
3.91
ANALOG-DIGITAL CONVERSION
triangular wave, but the waveform on either terminal with respect to ground is the more
complex waveform shown.
ANALOG
INPUT
VFC
FREQUENCY
COUNTER
OUTPUT
CONNECTION NEED NOT BE DIRECT
CIRCUIT IS IDEAL FOR TELEMETRY
Figure 3.107: Voltage-to-Frequency Converter (VFC) and Frequency Counter
Make a Low-Cost, Versatile, High-Resolution ADC
Figure 3.108: A Current-Steering VFC
Practical VFCs of this type have linearities around 14-bits, and comparable stability,
although they may be used in ADCs with higher resolutions without missing codes. The
performance limits are set by comparator threshold noise, threshold temperature
coefficient, and the stability and dielectric absorption (DA) of the capacitor, which is
generally a discrete component. The comparator/voltage reference structure shown in the
diagram is more of a representation of the function performed than the actual circuit used,
which is much more integrated with the switching, and correspondingly harder to
analyze.
This type of VFC is simple, inexpensive, and low-powered, and most run from a wide
range of supply voltages. They are ideally suited for low cost medium accuracy ADC and
data telemetry applications.
3.92
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
The charge balance VFC shown in Figure 3.109 is more complex, more demanding in its
supply voltage and current requirements, and more accurate. It is capable of 16-18 bit
linearity.
ANALOG
INPUT
DIGITAL
OUTPUT
–
PRECISION
ONE-SHOT
+
FREQUENCY
COUNTER
VREF
FREQUENCY
OUTPUT
I
PRECISION
CURRENT
SOURCE
Figure 3.109: Charge Balance Voltage-to-Frequency Converter (VFC)
The integrator capacitor charges from the signal as shown in Figure 3.109. When it
passes the comparator threshold, a fixed charge is removed from the capacitor, but the
input current continues to flow during the discharge, so no input charge is lost. The fixed
charge is defined by the precision current source and the pulse width of the precision
monostable. The output pulse rate is thus accurately proportional to the rate at which the
integrator charges from the input.
At low frequencies, the limits on the performance of this VFC are set by the stability of
the current source and the monostable timing (which depends on the monostable
capacitor, among other things). The absolute value and temperature stability of the
integration capacitor do not affect the accuracy, although its leakage and dielectric
absorption (DA) do. At high frequencies, second-order effects, such as switching
transients in the integrator and the precision of the monostable when it is retriggered very
soon after the end of a pulse, take their toll on accuracy and linearity.
The changeover switch in the current source addresses the integrator transient problem.
By using a changeover switch instead of the on/off switch more common on older VFC
designs: (a) there are no on/off transients in the precision current source and (b) the
output stage of the integrator sees a constant load—most of the time the current from the
source flows directly in the output stage; during charge balance, it still flows in the output
stage, but through the integration capacitor.
The stability and transient behavior of the precision monostable present more problems,
but the issue may be avoided by replacing the monostable with a clocked bistable
multivibrator. This arrangement is known as a synchronous VFC or SVFC and is shown
in Figure 3.110.
3.93
ANALOG-DIGITAL CONVERSION
ANALOG
INPUT
CLOCK INPUT
DIGITAL
OUTPUT
–
+
D
D
FLIP-FLOP
FREQUENCY
COUNTER
VREF
FREQUENCY
OUTPUT
I
PRECISION
CURRENT
SOURCE
Figure 3.110: Synchronous VFC (SVFC)
The difference from the previous circuit is quite small, but the charge balance pulse
length is now defined by two successive edges of the external clock. If this clock has low
jitter, the charge will be very accurately defined. The output pulse will also be
synchronous with the clock. SVFCs of this type are capable of up to 18-bit linearity and
excellent temperature stability.
This synchronous behavior is convenient in many applications, since synchronous data
transfer is often easier to handle than asynchronous. It does mean, however, that the
output of an SVFC is not a pure tone (plus harmonics, of course) like a conventional
VFC, but contains components harmonically related to the clock frequency. The display
of an SVFC output on an oscilloscope is especially misleading and is a common cause of
confusion—a change of input to a VFC produces a smooth change in the output
frequency, but a change to an SVFC produces a change in probability density of output
pulses N and N+1 clock cycles after the previous output pulse, which is often
misinterpreted as severe jitter and a sign of a faulty device (see Figure 3.111).
Another problem with SVFCs is nonlinearity at output frequencies related to the clock
frequency. If we study the transfer characteristic of an SVFC, we find nonlinearities close
to sub-harmonics of the clock frequency FC as shown in Figure 3.112. They can be found
at FC/3, FC/4, and FC/6. This is due to stray capacitance on the chip (and in the circuit
layout!) and the coupling of the clock signal into the SVFC comparator which causes the
device to behave as an injection-locked phase-locked loop (PLL). This problem is
intrinsic to SVFCs, but is not often serious: if the circuit card is well laid out, and clock
amplitude and dv/dts kept as low as practical, the effect is a discontinuity in the transfer
characteristic of less than 8 LSBs (at 18-bit resolution) at FC/3 and FC/4, and less at other
sub-harmonics. This is frequently tolerable, since the frequencies where it occurs are
known. Of course, if the circuit layout or decoupling is poor, the effect may be much
larger, but this is the fault of poor design and not the SVFC itself.
3.94
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
VFC
T
T VARIES SMOOTHLY WITH INPUT
T1
SVFC
T2
SVFC CLOCK
THE RELATIVE PROBABILITY OF DELAY
T1 OR T2 VARIES WITH INPUT, BUT
BOTH DELAYS DO OCCUR
Figure 3.111: VFC and SVFC Waveforms
Nonlinearities caused by
injection locking due to clock
feedthrough
Should only be 6-8 LSBs at
18-bits with proper layout and
decoupling
Nonlinearities occur at
subharmonics of the clock,
where FS = Fclock/2
Figure 3.112: SVFC Nonlinearity
It is evident that the SVFC is quantized, while the basic VFC is not. It does NOT follow
from this that the counter/VFC ADC has higher resolution (neglecting nonlinearities)
than the counter/SVFC ADC, because the clock in the counter also sets a limit to the
resolution.
When a VFC has a large input, it runs quickly and (counting for a short time) gives good
resolution, but it is hard to get good resolution in a reasonable sample time with a
slow-running VFC. In such a case, it may be more practical to measure the period of the
VFC output (this does not work for an SVFC), but of course the resolution of this system
3.95
ANALOG-DIGITAL CONVERSION
deteriorates as the input (and the frequency) increases. However, if the counter/timer
arrangement is made "smart," it is possible to measure the approximate VFC frequency
and the exact period of not one, but N cycles (where the value of N is determined by the
approximate frequency), and maintain high resolution over a wide range of inputs. The
AD1170 modular ADC released in 1986 is an example of this architecture.
VFCs have more applications than as a component in ADCs. Since their output is a pulse
stream, it may easily be sent over a wide range of transmission media (PSN, radio,
optical, IR, ultrasonic, etc.). It need not be received by a counter, but by another VFC
configured as a frequency-to-voltage converter (FVC). This gives an analog output, and a
VFC-FVC combination is a very useful way of sending a precision analog signal across
an isolation barrier. There are a number of issues to be considered in building FVCs from
VFCs, and these are considered in References 57-60.
Dual Slope/Multi-Slope ADCs
The dual-slope ADC architecture was truly a breakthrough in ADCs for high resolution
applications such as digital voltmeters, etc. (see References 61-64). A simplified diagram
is shown in Figure 3.113, and the integrator output waveforms are shown in Figure 3.114.
ANALOG
INPUT
VIN
C
R
–
–VREF
+
TIMER AND
CONTROL
DIGITAL
OUTPUT
OSC.
COUNTER
CK
R
Figure 3.113: Dual slope ADC
The input signal is applied to an integrator; at the same time a counter is started, counting
clock pulses. After a pre-determined amount of time (T), a reference voltage having
opposite polarity is applied to the integrator. At that instant, the accumulated charge on
the integrating capacitor is proportional to the average value of the input over the interval
T. The integral of the reference is an opposite-going ramp having a slope of VREF/RC. At
the same time, the counter is again counting from zero. When the integrator output
reaches zero, the count is stopped, and the analog circuitry is reset. Since the charge
gained is proportional to VIN · T, and the equal amount of charge lost is proportional to
3.96
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
VREF · tx, then the number of counts relative to the full scale count is proportional to tx/T,
or VIN/VREF. If the output of the counter is a binary number, it will therefore be a binary
representation of the input voltage.
SLOPE =
VIN
SLOPE =
RC
VREF
RC
(CONSTANT SLOPE)
t
0
T
tx
VIN
RC
T
=
tx
=
VREF
RC
VIN
VREF
tx
T
HIGH NORMAL MODE REJECTION AT MULTIPLES OF
1
T
Figure 3.114: Dual Slope ADC Integrator Output Waveforms
Dual-slope integration has many advantages. Conversion accuracy is independent of both
the capacitance and the clock frequency, because they affect both the up-slope and the
down-slope by the same ratio.
The fixed input signal integration period results in rejection of noise frequencies on the
analog input that have periods that are equal to or a sub-multiple of the integration time
T. Proper choice of T can therefore result in excellent rejection of 50-Hz and 60-Hz line
ripple as shown in Figure 3.115.
Errors caused by bias currents and the offset voltages of the integrating amplifier and the
comparator as well as gain errors can be cancelled by using additional charge/discharge
cycles to measure "zero" and "full-scale" and using the results to digitally correct the
initial measurement, as in the quad-slope architecture discussed in Reference 65.
The triple-slope architecture (see References 66-68) retains the advantages of the dualslope, but greatly increases the conversion speed at the cost of added complexity. The
increase in conversion speed is achieved by accomplishing the reference integration
(ramp-down) at two distinct rates: a high-speed rate, and a "vernier" lower speed rate.
The counter is likewise divided into two sections, one for the MSBs and one for the
LSBs. In a properly designed triple-slope converter, a significant increase in speed can be
achieved while retaining the inherent linearity, differential linearity, and stability
characteristics associated with dual-slope ADCs.
3.97
ANALOG-DIGITAL CONVERSION
Figure 3.115: Frequency Response of Integrating ADC
Optical Converters
Among the most popular position measuring sensors, optical encoders find use in
relatively low reliability and low resolution applications. An incremental optical encoder
(left-hand diagram in Figure 3.116) is a disc divided into sectors that are alternately
transparent and opaque. A light source is positioned on one side of the disc, and a light
sensor on the other side. As the disc rotates, the output from the detector switches
alternately on and off, depending on whether the sector appearing between the light
source and the detector is transparent or opaque.
Thus, the encoder produces a stream of square wave pulses which, when counted,
indicate the angular position of the shaft. Available encoder resolutions (the number of
opaque and transparent sectors per disc) range from 100 to 65,000, with absolute
accuracies approaching 30 arc-seconds (1/43,200 per rotation). Most incremental
encoders feature a second light source and sensor at an angle to the main source and
sensor, to indicate the direction of rotation. Many encoders also have a third light source
and detector to sense a once-per-revolution marker. Without some form of revolution
marker, absolute angles are difficult to determine. A potentially serious disadvantage is
that incremental encoders require external counters to determine absolute angles within a
given rotation. If the power is momentarily shut off, or if the encoder misses a pulse due
to noise or a dirty disc, the resulting angular information will be in error.
3.98
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
INCREMENTAL
θ
θ
ABSOLUTE
LIGHT
SOURCES
LIGHT
SOURCES
DISC
SHAFT
DISC
SENSORS
CONDITIONING
ELECTRONICS
SHAFT
5 BITS
SENSORS
5 BITS
CONDITIONING
ELECTRONICS
Figure 3.116: Incremental and Absolute Optical Encoders
The absolute optical encoder (right-hand diagram in Figure 3.116) overcomes these
disadvantages but is more expensive. An absolute optical encoder's disc is divided up into
N sectors (N = 5 for example shown), and each sector is further divided radially along its
length into opaque and transparent sections, forming a unique N-bit digital word with a
maximum count of 2N – 1. The digital word formed radially by each sector increments in
value from one sector to the next, usually employing Gray code. Binary coding could be
used, but can produce large errors if a single bit is incorrectly interpreted by the sensors.
Gray code overcomes this defect: the maximum error produced by an error in any single
bit of the Gray code is only 1 LSB after the Gray code is converted into binary code. A
set of N light sensors responds to the N-bit digital word which corresponds to the disc's
absolute angular position. Industrial optical encoders achieve up to 16-bit resolution, with
absolute accuracies that approach the resolution (20 arc seconds). Both absolute and
incremental optical encoders, however, may suffer damage in harsh industrial
environments.
Resolver-to-Digital Converters (RDCs) and Synchros
Machine-tool and robotics manufacturers have increasingly turned to resolvers and
synchros to provide accurate angular and rotational information. These devices excel in
demanding factory applications requiring small size, long-term reliability, absolute
position measurement, high accuracy, and low-noise operation.
A diagram of a typical synchro and resolver is shown in Figure 3.117. Both sycnchros
and resolvers employ single-winding rotors that revolve inside fixed stators. In the case
of a simple synchro, the stator has three windings oriented 120º apart and electrically
3.99
ANALOG-DIGITAL CONVERSION
connected in a Y-connection. Resolvers differ from synchros in that their stators have
only two windings oriented at 90º.
S1
S2
STATOR
SYNCHRO
ROTOR
ROTOR
R1
θ
S1 TO S3 = V sin ωt sin θ
S3 TO S2 = V sin ωt sin (θ + 120°)
S2 TO S1 = V sin ωt sin (θ + 240°)
V sin ωt
R2
S3
ROTOR
S4
R1
STATOR
V sin ωt
S1 TO S3 = V sin ωt sin θ
S4 TO S2 = V sin ωt sin (θ + 90°)
= V sin ωt cos θ
S2
STATOR
R2
S3
RESOLVER
S1
Figure 3.117: Synchros and Resolvers
Because synchros have three stator coils in a 120º orientation, they are more difficult than
resolvers to manufacture and are therefore more costly. Today, synchros find decreasing
use, except in certain military and avionic retrofit applications.
Modern resolvers, in contrast, are available in a brushless form that employ a transformer
to couple the rotor signals from the stator to the rotor. The primary winding of this
transformer resides on the stator, and the secondary on the rotor. Other resolvers use
more traditional brushes or slip rings to couple the signal into the rotor winding.
Brushless resolvers are more rugged than synchros because there are no brushes to break
or dislodge, and the life of a brushless resolver is limited only by its bearings. Most
resolvers are specified to work over 2-V to 40-V rms and at frequencies from 400 Hz to
10 kHz. Angular accuracies range from 5 arc-minutes to 0.5 arc-minutes. (There are 60
arc-minutes in one degree, and 60 arc-seconds in one arc-minute. Hence, one arc-minute
is equal to 0.0167 degrees).
In operation, synchros and resolvers resemble rotating transformers. The rotor winding is
excited by an ac reference voltage, at frequencies up to a few kHz. The magnitude of the
voltage induced in any stator winding is proportional to the sine of the angle, θ, between
the rotor coil axis and the stator coil axis. In the case of a synchro, the voltage induced
across any pair of stator terminals will be the vector sum of the voltages across the two
connected coils.
3.100
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
For example, if the rotor of a synchro is excited with a reference voltage, Vsinωt, across
its terminals R1 and R2, then the stator's terminal will see voltages in the form:
S1 to S3 = V sinωt sinθ
S3 to S2 = V sinωt sin (θ + 120º)
S2 to S1 = V sinωt sin (θ + 240º),
Eq. 3.2
Eq. 3.3
Eq. 3.4
where θ is the shaft angle.
In the case of a resolver, with a rotor ac reference voltage of Vsinωt, the stator's terminal
voltages will be:
S1 to S3 = V sinωt sin θ
S4 to S2 = V sinωt sin(θ + 90º) = V sinωt cosθ.
Eq. 3.5
Eq. 3.6
It should be noted that the 3-wire synchro output can be easily converted into the
resolver-equivalent format using a Scott-T transformer. Therefore, the following signal
processing example describes only the resolver configuration.
A typical resolver-to-digital converter (RDC) is shown functionally in Figure 3.118. The
two outputs of the resolver are applied to cosine and sine multipliers. These multipliers
incorporate sine and cosine lookup tables and function as multiplying digital-to-analog
converters. Begin by assuming that the current state of the up/down counter is a digital
number representing a trial angle, ϕ. The converter seeks to adjust the digital angle, ϕ,
continuously to become equal to, and to track θ, the analog angle being measured. The
resolver's stator output voltages are written as:
V1 = V sinωt sinθ
V2 = V sinωt cosθ
Eq. 3.7
Eq. 3.8
where θ is the angle of the resolver's rotor. The digital angle ϕ is applied to the cosine
multiplier, and its cosine is multiplied by V1 to produce the term:
V sinωt sinθ cosϕ.
Eq. 3.9
The digital angle ϕ is also applied to the sine multiplier and multiplied by V2 to product
the term:
V sinωt cosθ sinϕ.
Eq. 3.10
These two signals are subtracted from each other by the error amplifier to yield an ac
error signal of the form:
V sinωt [sinθ cosϕ – cosθ sinϕ].
Eq. 3.11
Using a simple trigonometric identity, this reduces to:
V sinωt [sin (θ –ϕ)].
Eq. 3.12
3.101
ANALOG-DIGITAL CONVERSION
The detector synchronously demodulates this ac error signal, using the resolver's rotor
voltage as a reference. This results in a dc error signal proportional to
sin(θ–ϕ).
The dc error signal feeds an integrator, the output of which drives a voltage-controlledoscillator (VCO). The VCO, in turn, causes the up/down counter to count in the proper
direction to cause:
Sin (θ – ϕ) → 0.
Eq. 3.13
When this is achieved,
θ – ϕ → 0,
Eq. 3.14
ϕ=θ
Eq. 3.15
and therefore
to within one count. Hence, the counter's digital output, ϕ, represents the angle θ. The
latches enable this data to be transferred externally without interrupting the loop's
tracking.
V sin ωt
V sin ωt sin θ
STATOR
INPUTS
V sin ωt cos θ
ROTOR REFERENCE
COSINE
MULTIPLIER
V sin ωt sin θ cos ϕ
_
ϕ
SINE
MULTIPLIER
+
V sin ωt cos θ sin ϕ
ϕ
UP / DOWN
COUNTER
V sin ωt [sin (θ – ϕ )]
DETECTOR
ERROR
K sin (θ – ϕ )
INTEGRATOR
VCO
ϕ = DIGITAL ANGLE
LATCHES
ϕ
WHEN ERROR = 0,
ϕ = θ ± 1 LSB
Figure 3.118: Resolver-to-Digital Converter (RDC)
3.102
VELOCITY
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
This circuit is equivalent to a so-called type-2 servo loop, because it has, in effect, two
integrators. One is the counter, which accumulates pulses; the other is the integrator at the
output of the detector. In a type-2 servo loop with a constant rotational velocity input, the
output digital word continuously follows, or tracks the input, without needing externally
derived convert commands, and with no steady state phase lag between the digital output
word and actual shaft angle. An error signal appears only during periods of acceleration
or deceleration.
As an added bonus, the tracking RDC provides an analog dc output voltage directly
proportional to the shaft's rotational velocity. This is a useful feature if velocity is to be
measured or used as a stabilization term in a servo system, and it makes additional
tachometers unnecessary.
Since the operation of an RDC depends only on the ratio between input signal
amplitudes, attenuation in the lines connecting them to resolvers doesn't substantially
affect performance. For similar reasons, these converters are not greatly susceptible to
waveform distortion. In fact, they can operate with as much as 10% harmonic distortion
on the input signals; some applications actually use square-wave references with little
additional error.
Tracking ADCs are therefore ideally suited to RDCs. While other ADC architectures,
such as successive approximation, could be used, the tracking converter is the most
accurate and efficient for this application.
Because the tracking converter doubly integrates its error signal, the device offers a high
degree of noise immunity (12-dB-per-octave rolloff). The net area under any given noise
spike produces an error. However, typical inductively coupled noise spikes have equal
positive and negative going waveforms. When integrated, this results in a zero net error
signal. The resulting noise immunity, combined with the converter's insensitivity to
voltage drops, lets the user locate the converter at a considerable distance from the
resolver. Noise rejection is further enhanced by the detector's rejection of any signal not
at the reference frequency, such as wideband noise.
The AD2S90 is one of a number of integrated RDCs offered by Analog Devices. The
general architecture is similar to that of Figure 3.118. Further details on synchro and
resolver-to-digital converters can be found in References 69 and 70.
3.103
ANALOG-DIGITAL CONVERSION
REFERENCES:
3.2 ADC ARCHITECTURES
1.
Reza Moghimi, "Amplifiers as Comparators," Ask the Applications Engineer 31, Analog Dialogue,
Vol. 37-04, Analog Devices, April 2003, http://www.analog.com.
2.
James N. Giles, "High Speed Transistor Difference Amplifier," U.S. Patent 3,843,934, filed January
31 1973, issued October 22, 1974. (describes one of the first high-speed ECL comparators, the
AM685).
3.
Christopher W. Mangelsdorf, A 400-MHz Input Flash Converter with Error Correction, IEEE
Journal of Solid-State Circuits, Vol. 25, No. 1, February 1990, pp. 184-191. (a discussion of the
AD770, an 8-bit 200 MSPS flash ADC. The paper describes the comparator metastable state problem
and how to optimize the ADC design to minimize its effects).
4.
Charles E. Woodward, A Monolithic Voltage-Comparator Array for A/D Converters, IEEE Journal of
Solid State Circuits, Vol. SC-10, No. 6, December 1975, pp. 392-399. (an early paper on a 3-bit flash
converter optimized to minimize metastable state errors).
5.
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 5-bit
electro-mechanical ADC described is probably the first documented flash converter. The patent was
largely ignored and forgotten until many years after the various Reeves' patents were issued in 19391942).
6.
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).
7.
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).
8.
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 at the time).
9.
R. Staffin and R. D. Lohman, "Signal Amplitude Quantizer," U.S. Patent 2,869,079, filed December
19, 1956, issued January 13, 1959. (describes flash and subranging conversion using tubes and
transistors).
10. Goto, et. al., "Esaki Diode High-Speed Logical Circuits," IRE Transactions on Electronic
Computers, Vol. EC-9, March 1960, pp. 25-29. (describes how to use tunnel diodes as logic
elements).
11. T. Kiyomo, K. Ikeda, and H. Ichiki, "Analog-to-Digital Converter Using an Esaki Diode Stack," IRE
Transactions on Electronic Computers, Vol. EC-11, December 1962, pp. 791-792. (description of a
low resolution 3-bit flash ADC using a stack of tunnel diodes).
12. H. R. Schindler, "Using the Latest Semiconductor Circuits in a UHF Digital Converter," Electronics,
August 1963, pp. 37-40. (describes a 6-bit 50-MSPS subranging ADC using three 2-bit tunnel diode
flash converters).
13. J. B. Earnshaw, "Design for a Tunnel Diode-Transistor Store with Nondestructive Read-out of
Information," IEEE Transactions on Electronic Computers, EC-13, 1964 , pp. 710-722. (use of
tunnel diodes as memory elements).
3.104
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
14. Willard K. Bucklen, "A Monolithic Video A/D Converter," Digital Video, Vol. 2, Society of Motion
Picture and Television Engineers, March 1979, pp. 34-42. (describes the revolutionary TDC1007J 8bit 20MSPS video flash converter. Originally introduced at the February 3, 1979 SMPTE Winter
Conference in San Francisco, Bucklen accepted an Emmy award for this product in 1988 and was
responsible for the initial marketing and applications support for the device).
15. J. Peterson, "A Monolithic video A/D Converter," IEEE Journal of Solid-State Circuits, Vol. SC-14,
No. 6, December 1979, pp. 932-937. (another detailed description of the TRW TDC1007J 8-bit, 20MSPS flash converter).
16. Yukio Akazawa et. al., A 400MSPS 8 Bit Flash A/D Converter, 1987 ISSCC Digest of Technical
Papers, pp. 98-99. (describes a monolithic flash converter using Gray decoding).
17. A. Matsuzawa et al., An 8b 600MHz Flash A/D Converter with Multi-stage Duplex-gray Coding,
Symposium VLSI Circuits, Digest of Technical Papers, May 1991, pp. 113-114. (describes a
monolithic flash converter using Gray decoding).
18. Chuck Lane, A 10-bit 60MSPS Flash ADC, Proceedings of the 1989 Bipolar Circuits and
Technology Meeting, IEEE Catalog No. 89CH2771-4, September 1989, pp. 44-47. (describes an
interpolating method for reducing the number of preamps required in a flash converter).
19. 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).
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. (the ground-breaking patent on PCM. Interestingly enough, the ADC and DAC proposed by
Reeves are counting types, and not successive approximation).
21. John C. Schelleng, "Code Modulation Communication System," U.S. Patent 2,453,461, filed June 19,
1946, issued November 9, 1948. (an interesting description of a rather cumbersome successive
approximation ADC based on vacuum tube technology. This converter was not very practical, but did
illustrate the concept. Also in the patent is a description of a corresponding binary DAC).
22. 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, 8KSPS successive
approximation ADC based on the subtraction of binary weighted charges from a capacitor to
implement the internal subtraction/DAC function. It required 5 internal reference voltages).
23. Harold R. Kaiser, et al, "High-Speed Electronic Analogue-to-Digital Converter System," U.S. Patent
2,784,396, filed April 2, 1953, issued March 5, 1957. (one of the first SAR ADCs to use an actual
binary-weighted DAC internally).
24. B. D. Smith, "Coding by Feedback Methods," Proceedings of the I. R. E., Vol. 41, August 1953, pp.
1053-1058. (Smith uses an internal DAC and also points out that a non-linear transfer function can be
achieved by using a DAC with non-uniform bit weights, a technique which is widely used in today's
voiceband ADCs with built-in companding).
25. 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
non-linear diode-based compressors and expanders for generating a non-linear ADC/DAC transfer
function).
26. 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, 20kSPS
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.)
3.105
ANALOG-DIGITAL CONVERSION
27. 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).
28. J. R. Gray and S. C. Kitsopoulos, "A Precision Sample-and-Hold Circuit with Subnanosecond
Switching," IEEE Transactions on Circuit Theory, CT11, September 1964, pp. 389-396. (one of the
first papers on the detailed analysis of a sample-and-hold circuit).
29. T. C. Verster, "A Method to Increase the Accuracy of Fast Serial-Parallel Analog-to-Digital
Converters," IEEE Transactions on Electronic Computers, EC-13, 1964, pp. 471-473. (one of the
first references to the use of error correction in a subranging ADC).
30. G. G. Gorbatenko, High-Performance Parallel-Serial Analog to Digital Converter with Error
Correction, IEEE National Convention Record, New York, March 1966. (another early reference to
the use of error correction in a subranging ADC).
31. D. J. Kinniment, D. Aspinall, and D.B.G. Edwards, "High-Speed Analogue-Digital Converter," IEE
Proceedings, Vol. 113, pp. 2061-2069, Dec. 1966. (a 7-bit 9MSPS three-stage pipelined error
corrected converter is described based on recircuilating through a 3-bit stage three times. Tunnel
(Esaki) diodes are used for the individual comparators. The article also shows a proposed faster
pipelined 7-bit architecture using three individual 3-bit stages with error correction. The article also
describes a fast bootstrapped diode-bridge sample-and-hold circuit).
32. O. A. Horna, "A 150Mbps A/D and D/A Conversion System," Comsat Technical Review, Vol. 2, No.
1, pp. 52-57, 1972. (a detailed description and analysis of a subranging ADC with error correction).
33. J. L. Fraschilla, R. D. Caveney, and R. M. Harrison, "High Speed Analog-to-Digital Converter," U.S.
Patent 3,597,761, filed Nov. 14, 1969, issued Aug. 13, 1971. (Describes an 8-bit, 5-MSPS subranging
ADC with switched references to second comparator bank).
34. Stephen H. Lewis, Scott Fetterman, George F. Gross, Jr., R. Ramachandran, and T. R. Viswanathan,
"A 10-b 20-Msample/s Analog-Digital Converter, IEEE Journal of Solid-State Circuits, Vol. 27, No.
3, March 1992, pp. 351-358. (a detailed description and analysis of an error corrected subranging
ADC using 1.5-bit pipelined stages).
35. Roy Gosser and Frank Murden, "A 12-bit 50MSPS Two-Stage A/D Converter," 1995 ISSCC Digest
of Technical Papers, p. 278. (a description of the AD9042 error corrected subranging ADC using
MagAMP stages for the internal ADCs).
36. B. D. Smith, "An Unusual Electronic Analog-Digital Conversion Method," IRE Transactions on
Instrumentation, June 1956, pp. 155-160. (possibly the first published description of the binary-coded
and Gray-coded bit-per-stage ADC architectures. Smith mentions similar work partially covered in R.
P. Sallen's 1949 thesis at M.I.T.).
37. N. E. Chasek, "Pulse Code Modulation Encoder," U.S. Patent 3,035,258, filed November 14, 1960,
issued May 15, 1962. (an early patent showing a diode-based circuit for realizing the Gray code
folding transfer function).
38. F. D. Waldhauer, "Analog-to-Digital Converter," U.S. Patent 3,187,325, filed July 2, 1962, issued
June 1, 1965. (a classic patent using op amps with diode switches in the feedback loops to implement
the Gray code folding transfer function).
39. 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. (a further description of a 9-bit
ADC based on Waldhauer's folding stage).
3.106
DATA CONVERTER ARCHITECTURES
3.2 ADC ARCHITECTURES
40. Udo Fiedler and Dieter Seitzer, "A High-Speed 8 Bit A/D Converter Based on a Gray-Code Multiple
Folding Circuit," IEEE Journal of Solid-State Circuits, Vol. SC-14, No. 3, June 1979, pp. 547-551.
(an early monolithic folding ADC).
41. Rudy J. van de Plassche and Rob E. J. van de Grift, "A High-Speed 7 Bit A/D Converter," IEEE
Journal of Solid-State Circuits, Vol. SC-14, No. 6, December 1979, pp. 938-943. (a monolithic
folding ADC).
42. Rob. E. J. van de Grift and Rudy J. van de Plassche, "A Monolithic 8-bit Video A/D Converter, IEEE
Journal of Solid State Circuits, Vol. SC-19, No. 3, June 1984, pp. 374-378. (a monolithic folding
ADC).
43. Rob. E. J. van de Grift, Ivo W. J. M. Rutten and Martien van der Veen, "An 8-bit Video ADC
Incorporating Folding and Interpolation Techniques," IEEE Journal of Solid State Circuits, Vol. SC22, No. 6, December 1987, pp. 944-953. (another monolithic folding ADC).
44. Rudy van de Plassche, Integrated Analog-to-Digital and Digital-to-Analog Converters, Kluwer
Academic Publishers, 1994, pp. 148-187. (a good textbook on ADCs and DACs with a section on
folding ADCs indicated by the referenced page numbers).
45. Carl Moreland, "An 8-bit 150 MSPS Serial ADC," 1995 ISSCC Digest of Technical Papers, Vol. 38,
p. 272. (a description of an 8-bit ADC with 5 folding stages followed by a 3-bit flash converter).
46. Carl Moreland, An Analog-to-Digital Converter Using Serial-Ripple Architecture, Masters' Thesis,
Florida State University College of Engineering, Department of Electrical Engineering, 1995.
(Moreland's early work on folding ADCs).
47. Frank Murden, "Analog to Digital Converter Using Complementary Differential Emitter Pairs," U.S.
Patent 5,550,492, filed December 1, 1994, issued August 27, 1996. (a description of an ADC based on
the MagAMP folding stage
48. Carl W. Moreland, "Analog to Digital Converter Having a Magnitude Amplifier with an Improved
Differential Input Amplifier," U.S. Patent 5,554,943, filed December 1, 1994, issued September 10,
1996. (a description of an 8-bit ADC with 5 folding stages followed by a 3-bit flash converter).
49. Frank Murden and Carl W. Moreland, "N-bit Analog-to-Digital Converter with N-1 Magnitude
Amplifiers and N Comparators," U.S. Patent 5,684,419, filed December 1, 1994, issued November 4,
1997. (another patent on the MagAMP folding architecture applied to an ADC).
50. Carl Moreland, Frank Murden, Michael Elliott, Joe Young, Mike Hensley, and Russell Stop, "A 14-bit
100-Msample/s Subranging ADC, IEEE Journal of Solid State Circuits, Vol. 35, No. 12, December
2000, pp. 1791-1798. (describes the architecture used in the 14-bit AD6645 ADC).
51. Frank Murden and Michael R. Elliott, "Linearizing Structures and Methods for Adjustable-Gain
Folding Amplifiers," U.S. Patent 6,172,636B1, filed July 13, 1999, issued January 9, 2001. (describes
methods for trimming the folding amplifiers in an ADC).
52. Bernard M. Oliver and Claude E. Shannon, "Communication System Employing Pulse Code
Modulation," U.S. Patent 2,801,281, filed February 21, 1946, issued July 30, 1957. (charge run-down
ADC and Shannon-Rack DAC).
53. Arthur H. Dickinson, "Device to Manifest an Unknown Voltage as a Numerical Quantity," U.S. Patent
2,872,670, filed May 26, 1951, issued February 3, 1959. (ramp run-up ADC).
54. K. Howard Barney, "Binary Quantizer," U.S. Patent 2,715,678, filed May 26, 1950, issued August 16,
1955. (tracking ADC).
55. Bernard M. Gordon and Robert P. Talambiras, "Information Translating Apparatus and Method," U.S.
Patent 2,989,741, filed July 22, 1955, issued June 20, 1961. (tracking ADC).
3.107
ANALOG-DIGITAL CONVERSION
56. John L. Lindesmith, "Voltage-to-Digital Measuring Circuit," U.S. Patent 2,835,868, filed September
16, 1952, issued May 20, 1958. (voltage-to-frequency ADC).
57. Paul Klonowski, "Analog-to-Digital Conversion Using Voltage-to-Frequency Converters,"
Application Note AN-276, Analog Devices, Inc. (a good application note on VFCs).
58. James M. Bryant, "Voltage-to-Frequency Converters," Application Note AN-361, Analog Devices,
Inc. (a good overview of VFCs).
59. Walt Jung, "Operation and Applications of the AD654 IC V-F Converter," Application Note AN-278,
Analog Devices, Inc.
60. Steve Martin, "Using the AD650 Voltage-to-Frequency Converter as a Frequency-to-Voltage
Converter," Application Note AN-279, Analog Devices, Inc. (a description of a frequency-to-voltage
converter using the popular AD650 VFC).
61. Robin N. Anderson and Howard A. Dorey, "Digital Voltmeters," U.S. Patent 3,267,458, filed August
20, 1962, issued August 16, 1966. (charge balance dual slope voltmeter ADC).
62. Richard Olshausen, "Analog-to-Digital Converter," U.S. Patent 3,281,827, filed June 27, 1963, issued
October 25, 1966. (charge balance dual slope ADC).
63. Roswell W. Gilbert, "Analog-to-Digital Converter," U.S. Patent 3,051,939, filed May 8, 1957, issued
August 28, 1962. (dual-slope ADC).
64. Stephan K. Ammann, "Integrating Analog-to-Digital Converter," U.S. Patent 3,316,547, filed July 15,
1964, issued April 25, 1967. (dual-slope ADC).
65. 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. (quad-slope ADC).
66. Hans Bent Aasnaes, "Triple Integrating Ramp Analog-to-Digital Converter," U.S. Patent 3,577,140,
filed June 27, 1967, issued May 4, 1971. (triple-slope ADC).
67. Frederick Bondzeit, Lewis J. Neelands, "Multiple Slope Analog-to-Digital Converter," U.S. Patent
3,564,538, filed January 29, 1968, issued February 16, 1971. (triple-slope ADC).
68. Desmond Wheable, "Triple-Slope Analog-to-Digital Converters," U.S. Patent 3,678,506, filed
October 2, 1968, issued July 18, 1972. (triple-slope ADC).
69. Dan Sheingold, Analog-Digital Conversion Handbook, Prentice-Hall, 1986, ISBN-0-13-032848-0,
pp. 441-471. (this chapter contains an excellent tutorial on optical, synchro, and resolver-to-digital
conversion).
70. Dennis Fu, "Circuit Applications of the AD2S90 Resolver-to-Digital Converter," Application Note
AN-230, Analog Devices. (applications of the AD2S90 RTD).
3.108
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
SECTION 3.3: SIGMA-DELTA CONVERTERS
Walt Kester, James Bryant
Historical Perspective
The sigma-delta (Σ-∆) ADC architecture had its origins in the early development phases
of pulse code modulation (PCM) systems—specifically, those related to transmission
techniques called delta modulation and differential PCM. (An excellent discussion of
both the history and concepts of the sigma-delta ADC can be found by Max Hauser in
Reference 1). Delta modulation was first invented, like classical PCM, at the ITT
Laboratories in France by the director of the laboratories, E. M. Deloraine, S. Van
Mierlo, and B. Derjavitch in 1946 (References 2, 3). The principle was rediscovered,
several years later, at the Phillips Laboratories in Holland, whose engineers published the
first extensive studies both of the single-bit and multi-bit concepts in 1952 and 1953
(References 4, 5). In 1950, C. C. Cutler of Bell Telephone Labs in the U.S. filed a
seminal patent on differential PCM which covered the same essential concepts
(Reference 6).
The driving force behind delta modulation and differential PCM was to achieve higher
transmission efficiency by transmitting the changes (delta) in value between consecutive
samples rather than the actual samples themselves.
In delta modulation, the analog signal is quantized by a one-bit ADC (a comparator) as
shown in Figure 3.119A. The comparator output is converted back to an analog signal
with a 1-bit DAC, and subtracted from the input after passing through an integrator. The
shape of the analog signal is transmitted as follows: a "1" indicates that a positive
excursion has occurred since the last sample, and a "0" indicates that a negative excursion
has occurred since the last sample.
If the analog signal remains at a fixed dc level for a period of time, a pattern alternating
of "0s" and "1s" is obtained. It should be noted that differential PCM (see Figure 3.119B)
uses exactly the same concept except a multibit ADC is used rather than a comparator to
derived the transmitted information.
Since there is no limit to the number of pulses of the same sign that may occur, delta
modulation systems are capable of tracking signals of any amplitude. In theory, there is
no peak clipping. However, the theoretical limitation of delta modulation is that the
analog signal must not change too rapidly. The problem of slope clipping is shown in
Figure 3.120. Here, although each sampling instant indicates a positive excursion, the
analog signal is rising too quickly, and the quantizer is unable to keep pace.
3.109
ANALOG-DIGITAL CONVERSION
SAMPLING CLOCK
ANALOG
INPUT
+
∑
DIGITAL
OUTPUT
–
(A) DELTA MODULATION
∫
1-BIT
DAC
SAMPLING CLOCK
ANALOG
INPUT
(B) DIFFERENTIAL PCM
+
∑
N-BIT
FLASH
ADC
DIGITAL
OUTPUT
–
∫
N-BIT
DAC
Figure 3.119: Delta Modulation and Differential PCM
SLOPE OVERLOAD
Figure 3.120: Quantization Using Delta Modulation
Slope clipping can be reduced by increasing the quantum step size or increasing the
sampling rate. Differential PCM uses a multibit quantizer to effectively increase the
quantum step sizes at the increase of complexity. Tests have shown that in order to obtain
the same quality as classical PCM, delta modulation requires very high sampling rates,
typically 20× the highest frequency of interest, as opposed to Nyquist rate of 2×.
For these reasons, delta modulation and differential PCM have never achieved any
significant degree of popularity, however a slight modification of the delta modulator
3.110
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
leads to the basic sigma-delta architecture, one of the most popular high resolution ADC
architectures in use today.
In 1954 C. C. Cutler of Bell Labs filed a very significant patent which introduced the
principle of oversampling and noise shaping with the specific intent of achieving higher
resolution (Reference 7). His objective was not specifically to design a Nyquist ADC, but
to transmit the oversampled noise-shaped signal without reducing the data rate. Thus
Cutler's converter embodied all the concepts in a sigma-delta ADC with the exception of
digital filtering and decimation which would have been too complex and costly at the
time using vacuum tube technology.
The basic single and multibit first-order sigma-delta ADC architecture is shown in Figure
3.121A and 3.121B, respectively. Note that the integrator operates on the error signal,
whereas in a delta modulator, the integrator is in the feedback loop. The basic
oversampling sigma-delta modulator increases the overall signal-to-noise ratio at low
frequencies by shaping the quantization noise such that most of it occurs outside the
bandwidth of interest. The digital filter then removes the noise outside the bandwidth of
interest, and the decimator reduces the output data rate back to the Nyquist rate.
SAMPLING CLOCK
+
ANALOG
INPUT
∑
DIGITAL
OUTPUT
( K•fs )
∫
DIGITAL
FILTER
–
(A) SINGLE BIT
( fs )
1-BIT
DAC
SAMPLING CLOCK
(K•fs)
+
ANALOG
INPUT
(B) MULTIBIT
∑
∫
N-BIT
FLASH
ADC
DIGITAL
OUTPUT
DIGITAL
FILTER
–
N-BIT
DAC
( fs )
Figure 3.121: Single and Multibit Sigma-Delta ADCs
Occasional work continued on these concepts over the next several years, including an
important patent of C. B. Brahm filed in 1961 which gave details of the analog design of
the loop filter for a second-order multibit noise shaping ADC (Reference 8). Transistor
circuits began to replace vacuum tubes over the period, and this opened up many more
possibilities for implementation of the architecture.
In 1962, Inose, Yasuda, and Murakami elaborated on the single-bit oversampling noiseshaping architecture proposed by Cutler in 1954 (Reference 9). Their experimental
3.111
ANALOG-DIGITAL CONVERSION
circuits used solid state devices to implement first and second-order sigma-delta
modulators. The 1962 paper was followed by a second paper in 1963 which gave
excellent theoretical discussions on oversampling and noise-shaping (Reference 10).
These two papers were also the first to use the name delta-sigma to describe the
architecture. The name delta-sigma stuck until the 1970s when AT&T engineers began
using name sigma-delta. Since that time, both names have been used; however, sigmadelta may be the more correct of the two (see later discussion in this section by Dan
Sheingold on the terminology).
It is interesting to note that all the work described thus far was related to transmitting an
oversampled digitized signal directly rather than the implementation of a Nyquist ADC.
In 1969 D. J. Goodman at Bell Labs published a paper describing a true Nyquist sigmadelta ADC with a digital filter and a decimator following the modulator (Reference 11).
This was the first use of the sigma-delta architecture for the explicit purpose of producing
a Nyquist ADC. In 1974 J. C. Candy, also of Bell Labs, described a multibit
oversampling sigma-delta ADC with noise shaping, digital filtering, and decimation to
achieve a high resolution Nyquist ADC (Reference 12).
The IC sigma-delta ADC offers several advantages over the other architectures,
especially for high resolution, low frequency applications. First and foremost, the singlebit sigma-delta ADC is inherently monotonic and requires no laser trimming. The sigmadelta ADC also lends itself to low cost foundry CMOS processes because of the digitally
intensive nature of the architecture. Examples of early monolithic sigma-delta ADCs are
given in References 13-21. Since that time there have been a constant stream of process
and design improvements in the fundamental architecture proposed in the early works
cited above. A summary of the key developments relating to sigma-delta is shown in
Figure 3.122.
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 3.122: Sigma-Delta ADC Architecture Timeline
3.112
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
Sigma-Delta (Σ-∆) or Delta-Sigma (∆-Σ)? Editor's Notes from Analog
Dialogue Vol. 24-2, 1990, by Dan Sheingold
This is not the most earth-shaking of controversies, and many readers may wonder what
the fuss is all about—if they wonder at all. The issue is important to both editor and
readers because of the need for consistency; we'd like to use the same name for the same
thing whenever it appears. But which name? In the case of the modulation technique that
led to a new oversampling A/D conversion mechanism, we chose sigma-delta. Here's
why.
Ordinarily, when a new concept is named by its creators, the name sticks; it should not be
changed unless it is erroneous or flies in the face of precedent. The seminal paper on this
subject was published in 1962 (References 9, 10), and its authors chose the name "deltasigma modulation," since it was based on delta modulation but included an integration
(summation, hence Σ).
Delta-sigma was apparently unchallenged until the 1970s, when engineers at AT&T were
publishing papers using the term sigma-delta. Why? According to Hauser (Reference 1),
the precedent had been to name variants of delta modulation with adjectives preceding
the word "delta." Since the form of modulation in question is a variant of delta
modulation, the sigma, used as an adjective—so the argument went—should precede the
delta.
Many engineers who came upon the scene subsequently used whatever term caught their
fancy, often without knowing why. It was even possible to find both terms used
interchangeably in the same paper. As matters stand today, sigma-delta is in widespread
use, probably for the majority of citations. Would its adoption be an injustice to the
inventors of the technique?
We think not. Like others, we believe that the name delta-sigma is a departure from
precedent. Not just in the sense of grammar, but also in relation to the hierarchy of
operations. Consider a block diagram for embodying an analog root-mean-square
(finding the square root of the mean of a squared signal) computer. First the signal is
squared, then it is integrated, and finally it is rooted (see Figure 3.123).
If we were to name the overall function after the causal order of operations, it would have
to be called a "square mean root" function. But naming in order of the hierarchy of its
mathematical operations gives us the familiar—and undisputed—name, root meansquare. Consider now a block diagram for taking a difference (delta), and then
integrating it (sigma).
Its causal order would give delta-sigma, but in functional hierarchy it is sigma-delta,
since it computes the integral of a difference. We believe that the latter term is correct
and follows precedent; and we have adopted it as our standard.
Dan Sheingold, 1990.
3.113
ANALOG-DIGITAL CONVERSION
Root-Mean-Square:
SQUARE
MEAN
ROOT
Hierarchy of Mathematical Operations: ROOT MEAN (SQUARE)
3
2
1
Sigma-Delta:
Sigma
Delta
+
∑
∫
Hierarchy of Mathematical Operations: SIGMA (DELTA)
2
1
Figure 3.123: Sigma-Delta (Σ-∆) or Delta-Sigma (∆-Σ)?
Basics of Sigma-Delta ADCs
Sigma-Delta Analog-Digital Converters (Σ-∆ ADCs) have been known for over thirty
years, but only recently has the technology (high-density digital VLSI) existed to
manufacture them as inexpensive monolithic integrated circuits. They are now used in
many applications where a low-cost, low-bandwidth, low-power, high-resolution ADC is
required.
There have been innumerable descriptions of the architecture and theory of Σ-∆ ADCs,
but most commence with a maze of integrals and deteriorate from there. Some engineers
who do not understand the theory of operation of Σ-∆ ADCs are convinced, from study of
a typical published article, that it is too complex to comprehend easily.
There is nothing particularly difficult to understand about Σ-∆ ADCs, as long as you
avoid the detailed mathematics, and this section has been written in an attempt to clarify
the subject. A Σ-∆ ADC contains very simple analog electronics (a comparator, voltage
reference, a switch, and one or more integrators and analog summing circuits), and quite
complex digital computational circuitry. This circuitry consists of a digital signal
processor (DSP) which acts as a filter (generally, but not invariably, a low pass filter). It
is not necessary to know precisely how the filter works to appreciate what it does. To
understand how a Σ-∆ ADC works, familiarity with the concepts of over-sampling,
quantization noise shaping, digital filtering, and decimation is required (see Figure
3.124).
3.114
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
Low Cost, High Resolution (to 24-bits)
Excellent DNL
Low Power, but Limited Bandwidth (Voiceband, Audio)
Key Concepts are Simple, but Math is Complex
Oversampling
Quantization Noise Shaping
Digital Filtering
Decimation
Ideal for Sensor Signal Conditioning
High Resolution
Self, System, and Auto Calibration Modes
Wide Applications in Voiceband and Audio Signal
Processing
Figure 3.124: Sigma-Delta ADCs
Let us consider the technique of over-sampling with an analysis in the frequency domain.
Where a dc conversion has a quantization error of up to ½ LSB, a sampled data system
has quantization noise. A perfect classical N-bit sampling ADC has an rms quantization
noise of q/√12 uniformly distributed within the Nyquist band of dc to fs/2 (where q is the
value of an LSB and fs is the sampling rate) as shown in Figure 3.125A. Therefore, its
SNR with a full-scale sinewave input will be (6.02N + 1.76) dB. If the ADC is less than
perfect, and its noise is greater than its theoretical minimum quantization noise, then its
effective resolution will be less than N-bits. Its actual resolution (often known as its
Effective Number of Bits or ENOB) will be defined by
ENOB =
SNR − 1.76dB
.
6.02dB
Eq. 3.16
If we choose a much higher sampling rate, Kfs (see Figure 3.125B), the rms quantization
noise remains q/√12, but the noise is now distributed over a wider bandwidth dc to Kfs/2.
If we then apply a digital low pass filter (LPF) to the output, we remove much of the
quantization noise, but do not affect the wanted signal—so the ENOB is improved. We
have accomplished a high resolution A/D conversion with a low resolution ADC. The
factor K is generally referred to as the oversampling ratio. It should be noted at this point
that oversampling has an added benefit in that it relaxes the requirements on the analog
antialiasing filter.
3.115
ANALOG-DIGITAL CONVERSION
A
fs
QUANTIZATION
NOISE = q / 12
q = 1 LSB
Nyquist
Operation
ADC
B
Oversampling
+ Digital Filter
Kfs
+ Decimation
ADC
C
Kfs
Σ∆
MOD
fs
fs
2
DIGITAL FILTER
DIGITAL
DEC
FILTER
Oversampling
+ Noise Shaping
+ Digital Filter
+ Decimation
fs
REMOVED NOISE
fs
2
Kfs
2
fs
DIGITAL
DEC
FILTER
Kfs
REMOVED NOISE
fs
2
Kfs
2
Kfs
Figure 3.125: Oversampling, Digital Filtering, Noise Shaping, and Decimation
Since the bandwidth is reduced by the digital output filter, the output data rate may be
lower than the original sampling rate (Kfs) and still satisfy the Nyquist criterion. This
may be achieved by passing every Mth result to the output and discarding the remainder.
The process is known as "decimation" by a factor of M. Despite the origins of the term
(decem is Latin for ten), M can have any integer value, provided that the output data rate
is more than twice the signal bandwidth. Decimation does not cause any loss of
information (see Figure 3.125B).
If we simply use oversampling to improve resolution, we must oversample by a factor of
22N to obtain an N-bit increase in resolution. The Σ-∆ converter does not need such a high
oversampling ratio because it not only limits the signal passband, but also shapes the
quantization noise so that most of it falls outside this passband as shown in Figure
3.125C.
If we take a 1-bit ADC (generally known as a comparator), drive it with the output of an
integrator, and feed the integrator with an input signal summed with the output of a 1-bit
DAC fed from the ADC output, we have a first-order Σ-∆ modulator as shown in Figure
3.126. Add a digital low pass filter (LPF) and decimator at the digital output, and we
have a Σ-∆ ADC—the Σ-∆ modulator shapes the quantization noise so that it lies above
the passband of the digital output filter, and the ENOB is therefore much larger than
would otherwise be expected from the over-sampling ratio.
3.116
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
CLOCK
Kfs
INTEGRATOR
VIN
+
∫
∑
fs
A
+
_
_
DIGITAL
FILTER
AND
DECIMATOR
LATCHED
COMPARATOR
(1-BIT ADC)
B
N-BITS
fs
+VREF
1-BIT
DAC
1-BIT DATA
STREAM
1-BIT,
Kfs
–VREF
SIGMA-DELTA MODULATOR
Figure 3.126: First-Order Sigma-Delta ADC
Intuitively, a Σ-∆ ADC operates as follows. Assume a dc input at VIN. The integrator is
constantly ramping up or down at node A. The output of the comparator is fed back
through a 1-bit DAC to the summing input at node B. The negative feedback loop from
the comparator output through the 1-bit DAC back to the summing point will force the
average dc voltage at node B to be equal to VIN. This implies that the average DAC
output voltage must equal the input voltage VIN. The average DAC output voltage is
controlled by the ones-density in the 1-bit data stream from the comparator output. As the
input signal increases towards +VREF, the number of "ones" in the serial bit stream
increases, and the number of "zeros" decreases. Similarly, as the signal goes negative
towards –VREF, the number of "ones" in the serial bit stream decreases, and the number of
"zeros" increases. From a very simplistic standpoint, this analysis shows that the average
value of the input voltage is contained in the serial bit stream out of the comparator. The
digital filter and decimator process the serial bit stream and produce the final output data.
For any given input value in a single sampling interval, the data from the 1-bit ADC is
virtually meaningless. Only when a large number of samples are averaged, will a
meaningful value result. The sigma-delta modulator is very difficult to analyze in the
time domain because of this apparent randomness of the single-bit data output. If the
input signal is near positive full-scale, it is clear that there will be more 1s than 0s in the
bit stream. Likewise, for signals near negative full-scale, there will be more 0s than 1s in
the bit stream. For signals near midscale, there will be approximately an equal number of
1s and 0s. Figure 3.127 shows the output of the integrator for two input conditions. The
first is for an input of zero (midscale). To decode the output, pass the output samples
through a simple digital lowpass filter that averages every four samples. The output of
the filter is 2/4. This value represents bipolar zero. If more samples are averaged, more
dynamic range is achieved. For example, averaging 4 samples gives 2 bits of resolution,
while averaging 8 samples yields 4/8, or 3 bits of resolution. In the bottom waveform of
3.117
ANALOG-DIGITAL CONVERSION
Figure 3.127, the average obtained for 4 samples is 3/4, and the average for 8 samples is
6/8.
Figure 3.127: Sigma-Delta Modulator Waveforms
The sigma-delta ADC can also be viewed as a synchronous voltage-to-frequency
converter followed by a counter. If the number of 1s in the output data stream is counted
over a sufficient number of samples, the counter output will represent the digital value of
the input. Obviously, this method of averaging will only work for dc or very slowly
changing input signals. In addition, 2N clock cycles must be counted in order to achieve
N-bit effective resolution, thereby severely limiting the effective sampling rate.
Further time-domain analysis is not productive, and the concept of noise shaping is best
explained in the frequency domain by considering the simple Σ-∆ modulator model in
Figure 3.128.
The integrator in the modulator is represented as an analog lowpass filter with a transfer
function equal to H(f) = 1/f. This transfer function has an amplitude response which is
inversely proportional to the input frequency. The 1-bit quantizer generates quantization
noise, Q, which is injected into the output summing block. If we let the input signal be X,
and the output Y, the signal coming out of the input summer must be X – Y. This is
multiplied by the filter transfer function, 1/f, and the result goes to one input of the output
summer. By inspection, we can then write the expression for the output voltage Y as:
1
Y = (X − Y) + Q .
Eq. 3.17
f
This expression can easily be rearranged and solved for Y in terms of X, f, and Q:
Y=
3.118
X
Q⋅f
+
.
f +1 f +1
Eq. 3.18
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
X
ANALOG FILTER
H(f) = 1
f
∑
+
_
Q=
QUANTIZATION
NOISE
1 (X–Y)
f
X–Y
∑
Y
Y
Y=
1 (X–Y)
+ Q
f
REARRANGING, SOLVING FOR Y:
Y=
X
+
f+1
SIGNAL TERM
Qf
f+1
NOISE TERM
Figure 3.128: Simplified Frequency Domain Linearized Model
of a Sigma-Delta Modulator
Note that as the frequency f approaches zero, the output voltage Y approaches X with no
noise component. At higher frequencies, the amplitude of the signal component
approaches zero, and the noise component approaches Q. At high frequency, the output
consists primarily of quantization noise. In essence, the analog filter has a lowpass effect
on the signal, and a highpass effect on the quantization noise. Thus the analog filter
performs the noise shaping function in the Σ-∆ modulator model.
For a given input frequency, higher order analog filters offer more attenuation. The same
is true of Σ-∆ modulators, provided certain precautions are taken.
By using more than one integration and summing stage in the Σ-∆ modulator, we can
achieve higher orders of quantization noise shaping and even better ENOB for a given
over-sampling ratio as is shown in Figure 3.129 for both a first and second-order Σ-∆
modulator.
The block diagram for the second-order Σ-∆ modulator is shown in Figure 3.130. Third,
and higher, order Σ-∆ ADCs were once thought to be potentially unstable at some values
of input—recent analyses using finite rather than infinite gains in the comparator have
shown that this is not necessarily so, but even if instability does start to occur, it is not
important, since the DSP in the digital filter and decimator can be made to recognize
incipient instability and react to prevent it.
3.119
ANALOG-DIGITAL CONVERSION
2ND ORDER
DIGITAL
FILTER
1ST ORDER
Kfs
2
fs
2
Figure 3.129: Sigma-Delta Modulators Shape Quantization Noise
INTEGRATOR
VIN
+
∑
_
∫
CLOCK
Kfs
INTEGRATOR
+
∑
∫
+
_
_
1-BIT
DAC
1-BIT
DATA
STREAM
DIGITAL FILTER
AND
DECIMATOR
N-BITS
fs
Figure 3.130: Second-Order Sigma-Delta ADC
Figure 3.131 shows the relationship between the order of the Σ-∆ modulator and the
amount of over-sampling necessary to achieve a particular SNR. For instance, if the
oversampling ratio is 64, an ideal second-order system is capable of providing an SNR of
about 80 dB. This implies approximately 13 effective number of bits (ENOB). Although
the filtering done by the digital filter and decimator can be done to any degree of
precision desirable, it would be pointless to carry more than 13 binary bits to the outside
world. Additional bits would carry no useful signal information, and would be buried in
the quantization noise unless post-filtering techniques were employed. Additional
resolution can be obtained by increasing the oversampling ratio and/or by using a higherorder modulator.
3.120
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
120
THIRD-ORDER LOOP*
21dB / OCTAVE
100
SECOND-ORDER LOOP
15dB / OCTAVE
80
SNR
(dB)
60
FIRST-ORDER LOOP
9dB / OCTAVE
40
* > 2nd ORDER LOOPS DO NOT
OBEY LINEAR MODEL
20
0
4
8
16
32
64
128
256
OVERSAMPLING RATIO, K
Figure 3.131: SNR Versus Oversampling Ratio for
First, Second, and Third-Order Loops
Idle Tone Considerations
In our discussion of sigma-delta ADCs up to this point, we have made the assumption
that the quantization noise produced by the sigma-delta modulator is random and
uncorrelated with the input signal. Unfortunately, this is not entirely the case, especially
for the first-order modulator. Consider the case where we are averaging 16 samples of
the modulator output in a 4 bit sigma-delta ADC.
Figure 3.132 shows the bit pattern for two input signal conditions: an input signal having
the value 8/16, and an input signal having the value 9/16. In the case of the 9/16 signal,
the modulator output bit pattern has an extra "1" every 16th output. This will produce
energy at fs/16, which translates into an unwanted tone. If the oversampling ratio is less
than 16, this tone will fall into the passband. In audio, the idle tones can be heard just
above the noise floor as the input changes from negative to positive fullscale.
Figure 3.133 shows the correlated idling pattern behavior for a first order sigma-delta
modulator, and Figure 3.134 shows the relatively uncorrelated pattern for a second-order
modulator. For this reason, virtually all sigma-delta ADCs contain at least a secondorder modulator loop, and some use up to fifth-order loops.
3.121
ANALOG-DIGITAL CONVERSION
Figure 3.132: Repetitive Bit Pattern in Sigma-Delta Modulator Output
Figure 3.133: Idling Patterns for First-Order Sigma-Delta Modulator
(Integrator Output)
Figure 3.134: Idling Patterns for Second-Order Sigma-Delta Modulator
(Integrator Output)
3.122
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
Higher Order Loop Considerations
In order to achieve wide dynamic range, sigma-delta modulator loops greater than
second-order are necessary, but present real design challenges. First of all, the simple
linear models previously discussed are no longer fully accurate. Loops of order greater
than two are generally not guaranteed to be stable under all input conditions. The
instability arises because the comparator is a non-linear element whose effective "gain"
varies inversely with the input level. This mechanism for instability causes the following
behavior: if the loop is operating normally, and a large signal is applied to the input that
overloads the loop, the average gain of the comparator is reduced. The reduction in
comparator gain in the linear model causes loop instability. This causes instability even
when the signal that caused it is removed. In actual practice, such a circuit would
normally oscillate on power-up due to initial conditions caused by turn-on transients. The
AD1879 dual audio ADC released in 1994 by Analog Devices used a 5th order loop.
Extensive nonlinear stabilization techniques were required in this and similar higherorder loop designs (References 22-26).
Multi-Bit Sigma-Delta Converters
So far we have considered only sigma-delta converters which contain a single-bit ADC
(comparator) and a single-bit DAC (switch). The block diagram of Figure 3.135 shows a
multi-bit sigma-delta ADC which uses an n-bit flash ADC and an n-bit DAC. Obviously,
this architecture will give a higher dynamic range for a given oversampling ratio and
order of loop filter. Stabilization is easier, since second-order loops can generally be
used. Idling patterns tend to be more random thereby minimizing tonal effects.
CLOCK
Kfs
fs
INTEGRATOR
VIN
+
∑
∫
FLASH
ADC
n-BITS
_
n-BIT
DAC
DIGITAL
FILTER
AND
DECIMATOR
N-BITS
fs
n-BITS,
n-BIT DATA
STREAM
Kfs
Figure 3.135: Multi-Bit Sigma-Delta ADC
3.123
ANALOG-DIGITAL CONVERSION
The real disadvantage of this technique is that the linearity depends on the DAC linearity,
and thin film laser trimming is required to approach 16-bit performance levels. This
makes the multi-bit architecture extremely impractical to implement on mixed-signal ICs
using traditional binary DAC techniques.
However, fully decoded thermometer DACs coupled with proprietary data scrambling
techniques as used in a number of Analog Devices' audio ADCs and DACs, including the
24-bit stereo AD1871 (see References 27 and 28) can achieve high SNR and low
distortion using the multibit architecture. The multibit data scrambling technique both
minimizes idle tones and ensures better differential linearity. A simplified block diagram
of the AD1871 ADC is shown in Figure 3.136.
Figure 3.136: AD1871 24-Bit 96-kSPS Stereo Audio Multi-Bit Sigma-Delta ADC
The AD1871's analog Σ-∆ modulator section comprises a second order multibit
implementation using Analog Device's proprietary technology for best performance. As
shown in Figure 3.137, the two analog integrator blocks are followed by a flash ADC
section that generates the multibit samples.
The output of the flash ADC, which is thermometer encoded, is decoded to binary for
output to the filter sections and is scrambled for feedback to the two integrator stages.
The modulator is optimized for operation at a sampling rate of 6.144 MHz (which is
128×fs at 48-kHz sampling and 64×fs at 96-kHz sampling). The A-weighted dynamic
range of the AD1871 is typically 105 dB. Key specifications for the AD1871 are
summarized in Figure 3.138.
3.124
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
Figure 3.137: Details of the AD1871 Second-Order Modulator
and Data Scrambler
Single +5 V Power Supply
Differential Dual-Channel Analog Inputs
16-/20-/24-Bit Word Lengths Supported
105 dB (typ) A-Weighted Dynamic Range
103 dB (typ) THD+N (–20 dBFS input)
0.01 dB Decimator Filter Passband Ripple
Second-Order, 128-/64-Times Oversampling Multibit
Modulator with Data Scrambling
Less than 350 mW (typ)
Power-Down Mode
On-Chip Voltage Reference
Flexible Serial Output Interface
28-Lead SSOP Package
Figure 3.138: AD1871 24-Bit, 96-kSPS Stereo
Sigma-Delta ADC Key Specifications
Digital Filter Implications
The digital filter is an integral part of all sigma-delta ADCs—there is no way to remove
it. The settling time of this filter affects certain applications especially when using sigmadelta ADCs in multiplexed applications. The output of a multiplexer can present a step
function input to an ADC if there are different input voltages on adjacent channels. In
fact, the multiplexer output can represent a full-scale step voltage to the sigma-delta ADC
when channels are switched. Adequate filter settling time must be allowed, therefore, in
such applications. This does not mean that sigma-delta ADCs shouldn't be used in
multiplexed applications, just that the settling time of the digital filter must be
considered. Some newer sigma-delta ADCs such are actually optimized for use in
multiplexed applications.
For example, the group delay through the AD1871 digital filter is 910 µs (sampling at 48
kSPS) and 460 µs (sampling at 96 kSPS)—this represents the time it takes for a step
function input to propagate through one-half the number of taps in the digital filter. The
3.125
ANALOG-DIGITAL CONVERSION
total settling time is therefore approximately twice the group delay time. The input
oversampling frequency is 6.144 MSPS for both conditions. The frequency response of
the digital filter in the AD1871 ADC is shown in Figure 1.139.
Figure 1.139: AD1871 24-Bit, 96-kSPS Stereo Sigma-Delta ADC
Digital Filter Characteristics
In other applications, such as low frequency, high resolution 24-bit measurement sigmadelta ADCs (such as the AD77xx-series), other types of digital filters may be used. For
instance, the SINC3 response is popular because it has zeros at multiples of the
throughput rate. For instance a 10-Hz throughput rate produces zeros at 50 Hz and 60 Hz
which aid in ac power line rejection.
Multistage Noise Shaping (MASH) Sigma-Delta Converters
As has been discussed, nonlinear stabilization techniques can be difficult for 3rd order
loops or higher. In many cases, the multibit architecture is preferable. An alternative
approach to either of these, called multistage noise shaping (MASH), utilizes cascaded
stable first-order loops (see References 29 and 30). Figure 3.140 shows a block diagram
of a three-stage MASH ADC. The output of the first integrator is subtracted from the
first DAC output to yield the first stage quantization noise, Q1. Q1 is then quantized by
the second stage. The output of the second integrator is subtracted from the second DAC
output to yield the second stage quantization noise which is in turn quantized by the third
stage.
The output of the first stage is summed with a single digital differentiation of the second
stage output and a double differentiation of the third stage output to yield the final output.
The result is that the quantization noise Q1 is suppressed by the second stage, and the
quantization noise Q2 is suppressed by the third stage yielding the same suppression as a
third-order loop. Since this result is obtained using three first-order loops, stable
operation is assured.
3.126
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
Figure 3.140: Multi-Stage Noise Shaping Sigma-Delta ADC (MASH)
High Resolution Measurement Sigma-Delta ADCs
In order to better understand the capability of sigma-delta measurement ADCs and the
power of the technique, a modern example, the AD7730, will be examined in detail. The
AD7730 is a member of the AD77XX family and is shown in Figure 3.141. This ADC
was specifically designed to interface directly to bridge outputs in weigh scale
applications. The device accepts low-level signals directly from a bridge and outputs a
serial digital word. There are two buffered differential inputs which are multiplexed,
buffered, and drive a PGA. The PGA can be programmed for four differential unipolar
analog input ranges: 0 V to +10 mV, 0 V to +20 mV, 0 V to +40 mV, and 0 V to +80 mV
and four differential bipolar input ranges: ±10 mV, ±20 mV, ±40 mV, and ±80 mV.
The maximum peak-to-peak, or noise-free resolution achievable is 1 in 230,000 counts,
or approximately 18-bits. It should be noted that the noise-free resolution is a function of
input voltage range, filter cutoff, and output word rate. Noise is greater using the smaller
input ranges where the PGA gain must be increased. Higher output word rates and
associated higher filter cutoff frequencies will also increase the noise.
The analog inputs are buffered on-chip allowing relatively high source impedances. Both
analog channels are differential, with a common mode voltage range that comes within
1.2 V of AGND and 0.95 V of AVDD. The reference input is also differential, and the
common mode range is from AGND to AVDD.
The 6-bit DAC is controlled by on-chip registers and can remove TARE (pan weight)
values of up to ±80 mV from the analog input signal range. The resolution of the TARE
function is 1.25 mV with a +2.5 V reference and 2.5 mV with a +5 V reference.
3.127
ANALOG-DIGITAL CONVERSION
AVDD
DVDD
REFIN(–)
REFIN(+)
REFERENCE DETECT
100nA
AIN1(+)
AIN1(–)
MUX
+
STANDBY
SIGMA-DELTA ADC
BUFFER
_
AIN2(+)/D1
AIN2(–)/D0
AD7730
+
∑
SIGMADELTA
MODULATOR
PGA
+/–
100nA
SERIAL INTERFACE
AND CONTROL LOGIC
6-BIT
DAC
PROGRAMMABLE
DIGITAL
FILTER
CLOCK
GENERATION
REGISTER BANK
VBIAS
ACX
MCLK IN
MCLK OUT
SCLK
CS
CALIBRATION
MICROCONTROLLER
ACX
SYNC
DIN
DOUT
AC
EXCITATION
CLOCK
AGND
DGND
POL
RDY
RESET
Figure 3.141: AD7730 Sigma-Delta Single-Supply Bridge ADC
The output of the PGA is applied to the Σ-∆ modulator and programmable digital filter.
The serial interface can be configured for three-wire operation and is compatible with
microcontrollers and digital signal processors. The AD7730 contains self-calibration and
system-calibration options and has an offset drift of less than 5nV/ºC and a gain drift of
less than 2 ppm/ºC. This low offset drift is obtained using a chop mode which operates
similarly to a chopper-stabilized amplifier.
The oversampling frequency of the AD7730 is 4.9152 MHz, and the output data rate can
be set from 50 Hz to 1200 Hz. The clock source can be provided via an external clock or
by connecting a crystal oscillator across the MCLK IN and MCLK OUT pins.
The AD7730 can accept input signals from a dc-excited bridge. It can also handle input
signals from an ac-excited bridge by using the ac excitation clock signals (ACX and
ACX ). These are non-overlapping clock signals used to synchronize the external
switches which drive the bridge. The ACX clocks are demodulated on the AD7730 input.
The AD7730 contains two 100-nA constant current generators, one source current from
AVDD to AIN(+) and one sink current from AIN(–) to AGND. The currents are switched
to the selected analog input pair under the control of a bit in the Mode Register. These
currents can be used in checking that a sensor is still operational before attempting to take
measurements on that channel. If the currents are turned on and a full-scale reading is
obtained, then the sensor has gone open circuit. If the measurement is 0V, the sensor has
gone short circuit. In normal operation, the burnout currents are turned off by setting the
proper bit in the Mode Register to 0.
The AD7730 contains an internal programmable digital filter. The filter consists of two
sections: a first stage filter, and a second stage filter. The first stage is a sinc3 lowpass
3.128
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
filter. The cutoff frequency and output rate of this first stage filter is programmable. The
second stage filter has three modes of operation. In its normal mode, it is a 22-tap FIR
filter that processes the output of the first stage filter. When a step change is detected on
the analog input, the second stage filter enters a second mode (FASTStep™) where it
performs a variable number of averages for some time after the step change, and then the
second stage filter switches back to the FIR filter mode. The third option for the second
stage filter (SKIP mode) is that it is completely bypassed so the only filtering provided on
the AD7730 is the first stage. Both the FASTStep mode and SKIP mode can be enabled
or disabled via bits in the control register.
Figure 3.142 shows the full frequency response of the AD7730 when the second stage
filter is set for normal FIR operation. This response is with the chop mode enabled and an
output word rate of 200 Hz and a clock frequency of 4.9152 MHz. The response is shown
from dc to 100 Hz. The rejection at 50 Hz ± 1 Hz and 60 Hz ± 1 Hz is better than 88 dB.
0
–10
–20
SINC3 + 22-TAP FIR FILTER,
CHOP MODE ENABLED
–30
GAIN –40
(dB)
–50
–60
–70
–80
–90
–110
–120
–130
0
10
20
30
40
50
60
70
80
90
100
FREQUNCY (Hz)
Figure 3.142: AD7730 Digital Filter Response
Figure 3.143 shows the step response of the AD7730 with and without the FASTStep
mode enabled. The vertical axis shows the code value and indicates the settling of the
output to the input step change. The horizontal axis shows the number of output words
required for that settling to occur. The positive input step change occurs at the 5th output.
In the normal mode (FASTStep disabled), the output has not reached its final value until
the 23rd output word. In FASTStep mode with chopping enabled, the output has settled
to the final value by the 7th output word. Between the 7th and the 23rd output, the
FASTStep mode produces a settled result, but with additional noise compared to the
specified noise level for normal operating conditions. It starts at a noise level comparable
to the SKIP mode, and as the averaging increases ends up at the specified noise level. The
complete settling time required for the part to return to the specified noise level is the
same for FASTStep mode and normal mode.
3.129
ANALOG-DIGITAL CONVERSION
20,000,000
FASTStep ENABLED
15,000,000
CODE
10,000,000
FASTStep DISABLED
5,000,000
0
0
5
10
15
20
25
NUMBER OF OUTPUT SAMPLES
Figure 3.143: AD7730 Digital Filter Settling Time Showing FASTStep™ Mode
The FASTStep mode gives a much earlier indication of where the output channel is going
and its new value. This feature is very useful in weigh scale applications to give a much
earlier indication of the weight, or in an application scanning multiple channels where the
user does not have to wait the full settling time to see if a channel has changed.
Note, however, that the FASTStep mode is not particularly suitable for multiplexed
applications because of the excess noise associated with the settling time. For
multiplexed applications, the full 23-cycle output word interval should be allowed for
settling to a new channel. This points out the fundamental issue of using Σ-∆ ADCs in
multiplexed applications. There is no reason why they won't work, provided the internal
digital filter is allowed to settle fully after switching channels.
The calibration modes of the AD7730 are given in Figure 3.144. A calibration cycle may
be initiated at any time by writing to the appropriate bits of the Mode Register.
Calibration removes offset and gain errors from the device.
The AD7730 gives the user access to the on-chip calibration registers allowing an
external microprocessor to read the device's calibration coefficients and also to write its
own calibration coefficients to the part from prestored values in external E2PROM. This
gives the microprocessor much greater control over the AD7730's calibration procedure.
It also means that the user can verify that the device has performed its calibration
correctly by comparing the coefficients after calibration with prestored values in
E2PROM. Since the calibration coefficients are derived by performing a conversion on
the input voltage provided, the accuracy of the calibration can only be as good as the
noise level the part provides in the normal mode. To optimize calibration accuracy, it is
recommended to calibrate the part at its lowest output rate where the noise level is lowest.
The coefficients generated at any output rate will be valid for all selected output update
3.130
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
rates. This scheme of calibrating at the lowest output data rate does mean that the
duration of the calibration interval is longer.
Internal Zero-ScaleCalibration
22 Output Cycles (CHP = 0)
24 Output Cycles (CHP = 1)
Internal Full-Scale Calibration
44 Output Cycles (CHP = 0)
48 Output Cycles (CHP = 1)
Calibration Programmed via the Mode Register
Calibration Coefficients Stored in Calibration
Registers
External Microprocessor Can Read or Write to
Calibration Coefficient Registers
Figure 3.144: AD7730 Calibration Options
The AD7730 requires an external voltage reference, however, the power supply may be
used as the reference in the ratiometric bridge application shown in Figure 3.145. In this
configuration, the bridge output voltage is directly proportional to the bridge drive
voltage which is also used to establish the reference voltages to the AD7730. Variations
in the supply voltage will not affect the accuracy. The SENSE outputs of the bridge are
used for the AD7730 reference voltages in order to eliminate errors caused by voltage
drops in the lead resistances.
+5V
+FORCE
RLEAD
6-LEAD
BRIDGE
+SENSE
DVDD
AVDD
+ VREF
VO
– SENSE
– FORCE
+5V/+3V
AD7730
ADC
+ AIN
– AIN
24 BITS
– VREF
RLEAD
AGND
DGND
Figure 3.145: AD7730 Bridge Application (Simplified Schematic)
3.131
ANALOG-DIGITAL CONVERSION
Bandpass Sigma-Delta Converters
The Σ-∆ ADCs that we have described so far contain integrators, which are low pass
filters, whose passband extends from dc. Thus, their quantization noise is pushed up in
frequency. At present, most commercially available Σ-∆ ADCs are of this type (although
some which are intended for use in audio or telecommunications applications contain
bandpass rather than lowpass digital filters to eliminate any system dc offsets). But there
is no particular reason why the filters of the Σ-∆ modulator should be LPFs, except that
traditionally ADCs have been thought of as being baseband devices, and that integrators
are somewhat easier to construct than bandpass filters. If we replace the integrators in a
Σ-∆ ADC with bandpass filters (BPFs) as shown in Figure 3.146, the quantization noise
is moved up and down in frequency to leave a virtually noise-free region in the pass-band
(see References 31, 32, and 33). If the digital filter is then programmed to have its
pass-band in this region, we have a Σ-∆ ADC with a bandpass, rather than a lowpass
characteristic. Such devices would appear to be useful in direct IF-to-digital conversion,
digital radios, ultrasound, and other undersampling applications. However, the modulator
and the digital BPF must be designed for the specific set of frequencies required by the
system application, thereby somewhat limiting the flexibility of this approach.
CLOCK
+
Σ
-
ANALOG
BPF
+
Σ
-
fs
Kfs
ANALOG
BPF
DIGITAL
BPF AND
DECIMATOR
1-BIT
DAC
fc
DIGITAL BPF
RESPONSE
fs > 2 BW
BW
SHAPED
QUANTIZATION
NOISE
f
Figure 3.146: Replacing Integrators with Resonators
Gives a Bandpass Sigma-Delta ADC
In an undersampling application of a bandpass Σ-∆ ADC, the minimum sampling
frequency must be at least twice the signal bandwidth, BW. The signal is centered around
a carrier frequency, fc. A typical digital radio application using a 455-kHz center
frequency and a signal bandwidth of 10 kHz is described in Reference 32. An
oversampling frequency Kfs = 2 MSPS and an output rate fs = 20 kSPS yielded a
dynamic range of 70 dB within the signal bandwidth.
3.132
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
Another example of a bandpass is the AD9870 IF Digitizing Subsystem having a nominal
oversampling frequency of 18 MSPS, a center frequency of 2.25 MHz, and a bandwidth
of 10 kHz - 150 kHz (see details in Reference 33).
Sigma-Delta DACs
Sigma-delta DACs operate very similarly to sigma-delta ADCs, however in a sigma-delta
DAC, the noise shaping function is accomplished with a digital modulator rather than an
analog one.
A Σ-∆ DAC, unlike the Σ-∆ ADC, is mostly digital (see Figure 3.147A). It consists of an
"interpolation filter" (a digital circuit which accepts data at a low rate, inserts zeros at a
high rate, and then applies a digital filter algorithm and outputs data at a high rate), a Σ-∆
modulator (which effectively acts as a low pass filter to the signal but as a high pass filter
to the quantization noise, and converts the resulting data to a high speed bit stream), and a
1-bit DAC whose output switches between equal positive and negative reference
voltages. The output is filtered in an external analog LPF. Because of the high
oversampling frequency, the complexity of the LPF is much less than the case of
traditional Nyquist operation.
SINGLE BIT
N-BITS @ fs
N-BITS @ K fs
DIGITAL
INTERPOLATION
FILTER
1- BIT @ KfS
DIGITAL
Σ∆
MODULATOR
ANALOG SIGNAL:
2 LEVELS
1-BIT
DAC
ANALOG
OUTPUT
FILTER
ANALOG
OUTPUT
MULTIBIT
N-BITS @ fs
N-BITS @ K fs
DIGITAL
INTERPOLATION
FILTER
M- BITS @ KfS
DIGITAL
MULTIBIT
Σ∆
ANALOG SIGNAL:
2M LEVELS
M-BIT
DAC
MODULATOR
ANALOG
OUTPUT
FILTER
ANALOG
OUTPUT
Figure 3.147: Sigma-Delta DACs
It is possible to use more than one bit in the Σ-∆ DAC, and this leads to the multibit
architecture shown in Figure 3.147B. The concept is similar to that of interpolating DACs
previously discussed in Chapter 2, with the addition of the digital sigma-delta modulator.
In the past, multibit DACs have been difficult to design because of the accuracy
requirement on the n-bit internal DAC (this DAC, although only n-bits, must have the
linearity of the final number of bits, N). The AD185x-series of audio DACs, however use
a proprietary data scrambling technique (called data directed scrambling) which
3.133
ANALOG-DIGITAL CONVERSION
overcomes this problem and produces excellent performance with respect to all audio
specifications (see References 27 and 28). For instance, the AD1853 dual 24-bit, 192kSPS DAC has greater than 104-dB THD + N at a 48-kSPS sampling rate.
One of the newest members of this family is the AD1955 multibit sigma-delta audio
DAC shown in Figure 3.148. The AD1955 also uses data directed scrambling, supports a
multitude of DVD audio formats and has an extremely flexible serial port. THD + N is
typically 110 dB.
Figure 3.148: AD1955 Multibit Sigma-Delta Audio DAC
Summary
Sigma-delta ADCs and DACs have proliferated into many modern applications including
measurement, voiceband, audio, etc. The technique takes full advantage of low cost
CMOS processes and therefore makes integration with highly digital functions such as
DSPs practical. Resolutions up to 24-bits are currently available, and the requirements on
analog antialiasing/anti-imaging filters are greatly relaxed due to oversampling. Modern
techniques such as the multibit data scrambled architecture minimize problems with idle
tones which plagued early sigma-delta products.
Many sigma-delta converters offer a high level of user programmability with respect to
output data rate, digital filter characteristics, and self-calibration modes. Multi-channel
sigma-delta ADCs are now available for data acquisition systems, and most users are
well-educated with respect to the settling time requirements of the internal digital filter in
these applications. Figure 3.149 summarizes some final thoughts about sigma-delta
converters.
3.134
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
Inherently Excellent Linearity
High Resolution Possible (24-Bits)
Oversampling Relaxes Analog Antialiasing Filter Requirements
Ideal for CMOS Processes, no Trimming
No SHA Required
Added Functionality: On-Chip PGAs, Analog Filters,
Autocalibration
On-Chip Programmable Digital Filters (AD7725: Lowpass,
Highpass, Bandpass, Bandstop)
Upper Sampling Rate Currently Limits Applications to
Measurement, Voiceband, and Audio, except for Bandpass SigmaDelta ADCs
Analog Multiplexer Switching Speed Limited by Internal Filter
Settling Time.
Figure 3.149: Sigma-Delta Summary
3.135
ANALOG-DIGITAL CONVERSION
REFERENCES:
3.3 SIGMA-DELTA CONVERTERS
1.
Max W. Hauser, "Principles of Oversampling A/D Conversion," Journal Audio Engineering Society,
Vol. 39, No. 1/2, January/February 1991, pp. 3-26. (one of the best tutorials and practical discussions
of the sigma-delta ADC architecture and its history).
2.
E. M. Deloraine, S. Van Mierlo, and B. Derjavitch, "Methode et systéme de transmission par
impulsions," French Patent 932,140, issued August, 1946. Also British Patent 627,262, issued 1949.
3.
E. M. Deloraine, S. Van Mierlo, and B. Derjavitch, "Communication System Utilizing Constant
Amplitude Pulses of Opposite Polarities," U.S. Patent 2,629,857, filed October 8, 1947, issued
February 24, 1953.
4.
F. de Jager, "Delta Modulation: A Method of PCM Transmission Using the One Unit Code," Phillips
Research Reports, Vol. 7, 1952, pp. 542-546. (additional work done on delta modulation during the
same time period).
5.
H. Van de Weg, "Quantizing Noise of a Single Integration Delta Modulation System with an N-Digit
Code," Phillips Research Reports, Vol. 8, 1953, pp. 367-385. (additional work done on delta
modulation during the same time period).
6.
C. C. Cutler, "Differential Quantization of Communication Signals," U.S. Patent 2,605,361, filed June
29, 1950, issued July 29, 1952. (recognized as the first patent on differential PCM or delta modulation,
although actually first invented in the Paris labs of the International Telephone and Telegraph
Corporation by E. M. Deloraine, S. Mierlo, and B. Derjavitch a few years earlier)
7.
C. C. Cutler, "Transmission Systems Employing Quantization," U.S. Patent 2,927,962, filed April 26,
1954, issued March 8, 1960. (a ground-breaking patent describing oversampling and noise shaping
using first and second-order loops to increase effective resolution. The goal was transmission of
oversampled noise shaped PCM data without decimation, not a Nyquist-type ADC).
8.
C. B. Brahm, "Feedback Integrating System," U.S. Patent 3,192,371, filed September 14, 1961,
issued June 29, 1965. (describes a second-order multibit oversampling noise shaping ADC).
9.
H. Inose, Y. Yasuda, and J. Murakami, "A Telemetering System by Code Modulation: ∆-Σ
Modulation," IRE Transactions on Space Electronics Telemetry, Vol. SET-8, September 1962, pp.
204-209. Reprinted in N. S. Jayant, Waveform Quantization and Coding, IEEE Press and John
Wiley, 1976, ISBN 0-471-01970-4. (an elaboration on the 1-bit form of Cutler's noise-shaping
oversampling concept. This work coined the description of the architecture as 'delta-sigma
modulation').
10. H. Inose and Y. Yasuda, "A Unity Bit Coding Method by Negative Feedback," IEEE Proceedings,
Vol. 51, November 1963, pp. 1524-1535. (further discussions on their 1-bit 'delta-sigma' concept).
11. D. J. Goodman, "The Application of Delta Modulation of Analog-to-PCM Encoding," Bell System
Technical Journal, Vol. 48, February 1969, pp. 321-343. Reprinted in N. S. Jayant, Waveform
Quantization and Coding, IEEE Press and John Wiley, 1976, ISBN 0-471-01970-4. (the first
description of using oversampling and noise shaping techniques followed by digital filtering and
decimation to produce a true Nyquist-rate ADC).
12. J. C. Candy, "A Use of Limit Cycle Oscillations to Obtain Robust Analog-to-Digital Converters,"
IEEE Transactions on Communications, Vol. COM-22, December 1974, pp. 298-305. (describes a
multibit oversampling noise shaping ADC with output digital filtering and decimation to interpolate
between the quantization levels).
13. R. J. van de Plassche, "A Sigma-Delta Modulator as an A/D Converter," IEEE Transactions on
Circuits and Systems, Vol. CAS-25, July 1978, pp. 510-514.
3.136
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
14. B. A. Wooley and J. L. Henry, "An Integrated Per-Channel PCM Encoder Based on Interpolation,"
IEEE Journal of Solid State Circuits, Vol. SC-14, February 1979, pp. 14-20. (one of the first allintegrated CMOS sigma-delta ADCs).
15. B. A. Wooley et al, "An Integrated Interpolative PCM Decoder," IEEE Journal of Solid State
Circuits, Vol. SC-14, February 1979, pp. 20-25.
16. J. C. Candy, B. A. Wooley, and O. J. Benjamin, "A Voiceband Codec with Digital Filtering," IEEE
Transactions on Communications, Vol. COM-29, June 1981, pp. 815-830.
17. J. C. Candy and Gabor C. Temes, Oversampling Delta-Sigma Data Converters, IEEE Press, ISBN
0-87942-258-8, 1992.
18. R. Koch, B. Heise, F. Eckbauer, E. Engelhardt, J. Fisher, and F. Parzefall, "A 12-bit Sigma-Delta
Analog-to-Digital Converter with a 15 MHz Clock Rate," IEEE Journal of Solid-State Circuits, Vol.
SC-21, No. 6, December 1986.
19. D. R. Welland, B. P. Del Signore and E. J. Swanson, "A Stereo 16-Bit Delta-Sigma A/D Converter for
Digital Audio," J. Audio Engineering Society, Vol. 37, No. 6, June 1989, pp. 476-485.
20. B. Boser and Bruce Wooley, "The Design of Sigma-Delta Modulation Analog-to-Digital Converters,"
IEEE Journal of Solid-State Circuits, Vol. 23, No. 6, December 1988, pp. 1298-1308.
21. J. Dattorro, A. Charpentier, D. Andreas, "The Implementation of a One-Stage Multirate 64:1 FIR
Decimator for use in One-Bit Sigma-Delta A/D Applications," AES 7th International Conference,
May 1989.
22. W.L. Lee and C.G. Sodini, "A Topology for Higher-Order Interpolative Coders," ISCAS PROC.
1987.
23. P.F. Ferguson, Jr., A. Ganesan and R. W. Adams, "One-Bit Higher Order Sigma-Delta A/D
Converters," ISCAS PROC. 1990, Vol. 2, pp. 890-893.
24. Wai Laing Lee, A Novel Higher Order Interpolative Modulator Topology for High Resolution
Oversampling A/D Converters, MIT Masters Thesis, June 1987.
25. R. W. Adams, "Design and Implementation of an Audio 18-Bit Analog-to-Digital Converter Using
Oversampling Techniques," J. Audio Engineering Society, Vol. 34, March 1986, pp. 153-166.
26. P. Ferguson, Jr., A. Ganesan, R. Adams, et. al., "An 18-Bit 20-kHz Dual Sigma-Delta A/D Converter,"
ISSCC Digest of Technical Papers, February 1991.
27. Robert Adams, Khiem Nguyen, and Karl Sweetland, "A 113 dB SNR Oversampling DAC with
Segmented Noise-Shaped Scrambling, " ISSCC Digest of Technical Papers, vol. 41, 1998, pp. 62,
63, 413. (describes a segmented audio DAC with data scrambling).
28. Robert W. Adams and Tom W. Kwan, "Data-directed Scrambler for Multi-bit Noise-shaping D/A
Converters," U.S. Patent 5,404,142, filed August 5, 1993, issued April 4, 1995. (describes a
segmented audio DAC with data scrambling).
29. Y. Matsuya, et. al., "A 16-Bit Oversampling A/D Conversion Technology Using Triple-Integration
Noise Shaping," IEEE Journal of Solid-State Circuits, Vol. SC-22, No. 6, December 1987, pp. 921929.
30. Y. Matsuya, et. al., "A 17-Bit Oversampling D/A Conversion Technology Using Multistage Noise
Shaping," IEEE Journal of Solid-State Circuits, Vol. 24, No. 4, August 1989, pp. 969-975.
3.137
ANALOG-DIGITAL CONVERSION
31. Paul H. Gailus, William J. Turney, and Francis R. Yester, Jr., "Method and Arrangement for a Sigma
Delta Converter for Bandpass Signals," U.S. Patent 4,857,928, filed January 28, 1988, issued August
15, 1989.
32. S.A. Jantzi, M. Snelgrove, and P.F. Ferguson Jr., "A 4th-Order Bandpass Sigma-Delta Modulator,"
IEEE Journal of Solid State Circuits, Vol. 38, No. 3, March 1993, pp. 282-291.
33. Paul Hendriks, Richard Schreier, Joe DiPilato, "High Performance Narrowband Receiver Design
Simplified by IF Digitizing Subsystem in LQFP," Analog Dialogue, Vol. 35-3, June-July 2001.
available at http://www.analog.com (describes an IF subsystem with a bandpass sigma-delta ADC
having a nominal oversampling frequency of 18MSPS, a center frequency of 2.25MHz, and a
bandwidth of 10kHz - 150kHz).
3.138
DATA CONVERTER ARCHITECTURES
3.3 SIGMA-DELTA CONVERTERS
General References on Data Conversion and Related Topics
1.
Alfred K. Susskind, Notes on Analog-Digital Conversion Techniques, John Wiley, 1957.
2.
David F. Hoeschele, Jr., Analog-to-Digital/Digital-to-Analog Conversion Techniques, John Wiley
and Sons, 1968.
3.
K. W. Cattermole, Principles of Pulse Code Modulation, American Elsevier Publishing Company,
Inc., 1969, New York NY, ISBN 444-19747-8.
4.
Hermann Schmid, Electronic Analog/Digital Conversions, Van Nostrand Reinhold Co., 1970.
5.
Dan Sheingold, Analog-Digital Conversion Handbook, First Edition, Analog Devices, 1972.
6.
Donald B. Bruck, Data Conversion Handbook, Hybrid Systems Corporation, 1974.
7.
Eugene R. Hnatek, A User's Handbook of D/A and A/D Converters, John Wiley, New York, 1976,
ISBN 0-471-40109-9.
8.
Nuggehally S. Jayant, Waveform Quantizing and Coding, John Wiley-IEEE Press, 1976, ISBN 087942-074-X.
9.
Dan Sheingold, Analog-Digital Conversion Notes, Analog Devices, 1977.
10. C. F. Kurth, editor, IEEE Transactions on Circuits and Systems Special Issue on Analog/Digital
Conversion, CAS-25, No. 7, July 1978.
11. Daniel J. Dooley, Data Conversion Integrated Circuits, John Wiley-IEEE Press, 1980, ISBN 0-47108155-8.
12. Bernard M. Gordon, The Analogic Data-Conversion Systems Digest, Fourth Edition, Analogic
Corporation, 1981.
13. Eugene L. Zuch, Data Acquisition and Conversion Handbook, Datel-Intersil, 1982.
14. Frank F. E. Owen, PCM and Digital Transmission Systems, McGraw-Hill, 1982, ISBN 0-07047954-2.
15. Dan Sheingold, Analog-Digital Conversion Handbook, Analog Devices/Prentice-Hall, 1986, ISBN
0-13-032848-0.
16. Matthew Mahoney, DSP-Based Testing of Analog and Mixed-Signal Circuits, IEEE Computer
Society Press, 1987, ISBN 0-8186-0785-8.
17. Michael J. Demler, High-Speed Analog-to-Digital Conversion, Academic Press, Inc., 1991, ISBN 012-209048-9.
18. J. C. Candy and Gabor C. Temes, Oversampling Delta-Sigma Data Converters, IEEE Press, ISBN
0-87942-258-8, 1992.
19. David F. Hoeschele, Jr., Analog-to-Digital and Digital-to-Analog Conversion Techniques, Second
Edition, John Wiley and Sons, 1994, ISBN-0-471-57147-4.
20. Rudy van de Plassche, Integrated Analog-to-Digital and Digital-to-Analog Converters, Kluwer
Academic Publishers, 1994, ISBN 0-7923-9436-4.
21. David A. Johns and Ken Martin, Analog Integrated Circuit Design, John Wiley, 1997, ISBN 0-47114448-7.
3.139
ANALOG-DIGITAL CONVERSION
22. Mikael Gustavsson, J. Jacob Wikner, and Nianxiong Nick Tan, CMOS Data Converters for
Communications, Kluwer Academic Publishers, 2000, ISBN 0-7923-7780-X.
23. R. Jacob Baker, CMOS Circuit Design Volumes I and II, John Wiley-IEEE Computer Society, 2002,
ISBN 0-4712-7256-6.
24. Rudy van de Plassche, CMOS Integrated Analog-to-Digital and Digital-to-Analog Converters,
Second Edition, Kluwer Academic Publishers, 2003, ISBN 1-4020-7500-6.
Analog Devices' Seminar Series:
25. Walt Kester, Practical Analog Design Techniques, Analog Devices, 1995, ISBN 0-916550-16-8,
available for download at http://www.analog.com.
26. Walt Kester, High Speed Design Techniques, Analog Devices, 1996, ISBN 0-916550-17-6, available
for download at http://www.analog.com.
27. Walt Kester, Practical Design Techniques for Power and Thermal Management, Analog Devices,
1998, ISBN 0-916550-19-2, available for download at http://www.analog.com.
28. Walt Kester, Practical Design Techniques for Sensor Signal Conditioning, Analog Devices, 1999,
ISBN 0-916550-20-6, available for download at http://www.analog.com.
29. Walt Kester, Mixed-Signal and DSP Design Techniques, Analog Devices, 2000, ISBN 0-916550-222, available for download at http://www.analog.com.
30. Walt Kester, Mixed-Signal and DSP Design Techniques, Analog Devices and Newnes (An Imprint
of Elsevier Science), ISBN 0-75067-611-6, 2003.
31. Walter G. Jung, Op Amp Applications, Analog Devices, 2002, ISBN 0-916550-26-5.
3.140
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