AD977/AD977A
–19–REV. D
OFFSET AND GAIN ADJUSTMENT
The AD977/AD977A is factory trimmed to minimize gain,
offset and linearity errors. In some applications, where the ana-
log input signal is required to meet the full dynamic range of the
ADC, the gain and offset errors need to be externally trimmed
to zero. Figures 12 and 13 show the required trim circuitry to
correct for these offset and gain errors.
Where adjustment is required, offset error must be corrected
before gain error. To achieve this in the bipolar input configura-
tion, trim the offset potentiometer with the input voltage set to
1/2 LSB below ground. Then adjust the potentiometer until the
major carry transition is located between 1111 1111 1111 1111
and 0000 0000 0000 0000. To adjust the gain error, an analog
signal should be input at either the first code transition (ADC
negative full scale) or the last code transition (ADC positive full
scale). Thus, to adjust for full-scale error, an input voltage of
FS/2 – 3/2 LSBs can be applied to V
IN
, and the gain potentiom-
eter should be adjusted until the output code flickers between
the last positive code transition 0111 1111 1111 1111 and 0111
1111 1111 1110. Should the first code transition need adjust-
ing, the trim procedure should consist of applying an analog
input signal of –FS/2 + 1/2 LSB to the V
IN
input and adjust-
ing the trim until the output code flickers between 1000 0000
0000 0000 and 1000 0000 0000 0001.
AC PERFORMANCE
The AD977/AD977A is fully specified and tested for dynamic
performance specifications. The ac parameters are required for
signal processing applications such as speech recognition and
spectrum analysis. These applications require information on
the ADC’s effect on the spectral content of the input signal.
Hence, the parameters for which the AD977/AD977A is specified
include S/(N+D), THD and Spurious Free Dynamic Range.
These terms are discussed in greater detail in the following
sections.
As a general rule, it is recommended that the results from sev-
eral conversions be averaged to reduce the effects of noise and
thus improve parameters such as S/(N+D) and THD. The ac
performance of the AD977/AD977A can be optimized by operat-
ing the ADC at its maximum sampling rate of 100 kHz/200 kHz
and digitally filtering the resulting bit stream to the desired signal
bandwidth. By distributing noise over a wider frequency range
the noise density in the frequency band of interest can be
reduced. For example, if the required input bandwidth is 50 kHz,
FREQUENCY kHz
0
10
0 10010 20 30 40
40
70
130
20
30
60
50
AMPLITUDE dB
90
120
80
110
100
50 60 70 80 90 955 1525354555657585
5280 POINT FFT
F
SAMPLE
= 200kHz
F
IN
= 20kHz, 0dB
SNRD = 86dB
THD = 101dB
Figure 17. FFT Plot
the AD977/AD977A could be oversampled by a factor of 2/4.
This would yield a 3/6 dB improvement in the effective SNR
performance.
DC PERFORMANCE
The factory calibration scheme used for the AD977/AD977A
compensates for bit weight errors that may exist in the capacitor
array. The mismatch in capacitor values is adjusted (using the
calibration coefficients) during a conversion resulting in excel-
lent dc linearity performance. Figures 18, 19, 20, 21, 22 and 23,
respectively, show typical INL, typical DNL, typical positive and
negative INL and DNL distribution plots for the AD977/AD977A
at 25°C.
A histogram test is a statistical method for deriving an A/D
converter’s differential nonlinearity. A ramp input is sampled by
the ADC and a large number of conversions are taken at each
voltage level, averaged then stored. The effect of averaging is to
reduce the transition noise by 1/n. If 64 samples are averaged at
each point, the effect of transition noise is reduced by a factor of
8, i.e., a transition noise of 0.8 LSBs rms is reduced to
0.1 LSBs rms. Theoretically the codes, during a test of DNL,
would all be the same size and therefore have an equal number
of occurrences. A code with an average number of occurrences
would have a DNL of “0.” A code that is different from the
average would have a DNL that was either greater or less than
zero LSB. A DNL of –1 LSB indicates that there is a missing
code present at the 16-bit level and that the ADC exhibits 15-
bit performance.
OUTPUT CODE K
0665 101520253035
2.0
2.0
0
0.5
1.0
1.5
1.0
0.5
1.5
40 45 50 55 60
100%
LSB
Figure 18. INL Plot
OUTPUT CODE K
0665 101520253035
2.0
2.0
0
0.5
1.0
1.5
1.0
0.5
1.5
40 45 50 55 60
100%
LSB
Figure 19. DNL Plot
AD977/AD977A
–20–
REV. D
POSITIVE INL LSB
0
50
10
25
20
15
40
35
45
NUMBER OF UNITS
5
0
30
0.2
0.3
0.4
0.5
0.6
0.8
0.9
1
1.1
1.2
2.9
1.4
1.5
1.6
1.7
1.8
2
2.1
2.2
2.3
2.4
2.6
2.7
2.8
Figure 20. Typical Positive INL Distribution (999 Units)
NEGATIVE INL LSB
2.5
2.4
2.3
2.2
2.1
2.0
1.9
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
60
30
20
50
NUMBER OF UNITS
10
0
40
0.1
Figure 21. Typical Negative INL Distribution (999 Units)
POSITIVE DNL LSB
60
20
NUMBER OF UNITS
0
40
80
100
120
0.02
0.10
0.19
0.27
0.36
0.44
0.52
0.61
0.69
0.77
0.86
0.94
1.02
1.11
1.19
1.27
1.36
1.44
1.53
1.61
1.69
1.78
1.86
1.94
2.03
Figure 22. Typical Positive DNL Distribution (999 Units)
60
30
20
50
NUMBER OF UNITS
10
0
40
1.17
1.12
1.02
0.92
NEGATIVE DNL LSB
1.07
0.97
70
80
90
1.22
0.87
0.82
0.78
0.73
0.68
0.63
0.58
0.53
0.48
0.43
0.38
0.33
0.28
0.23
0.18
0.14
0.09
0.04
Figure 23. Typical Negative DNL Distribution (999 Units)
SNR+D (dB) FOR AD977A
90
80
70
60
50
40
30
20
10
SINAD (dB) FOR V
IN
= 0dB
1 10 100 1000
INPUT SIGNAL FREQUENCY kHz
SNR+D (dB) FOR AD977
Figure 24. S/(N+D) vs. Input Frequency
TEMPERATURE C
110
100
80
75 15050
SFDR, S/N+D dB
25 0 25 50 75 100 125
90
105
95
85
80
90
110
100
85
95
105
SFDR
THD
SNRD
THD dB
Figure 25. AC Parameters vs. Temperature
AD977/AD977A
–21–REV. D
DC CODE UNCERTAINTY
Ideally, a fixed dc input should result in the same output code
for repetitive conversions; however, as a consequence of unavoid-
able circuit noise within the wideband circuits of the ADC, a
range of output codes may occur for a given input voltage.
Thus, when a dc signal is applied to the AD977/AD977A input
and 10,000 conversions are recorded, the result will be a distri-
bution of codes as shown in Figure 26. This histogram shows a
bell shaped curve consistent with the Gaussian nature of thermal
noise. The histogram is approximately seven codes wide. The
standard deviation of this Gaussian distribution results in a code
transition noise of 1 LSB rms.
4000
3500
0
3
2000
1500
1000
500
3000
2500
2 10 1 2 3 4
Figure 26. Histogram of 10,000 Conversions of a DC Input
USE OF THE TAG INPUT
The AD977/AD977A provides a TAG input pin for cascading
multiple converters together. This feature is useful for reducing
component count in systems where an isolation barrier must be
crossed and is also useful for systems with a limited capacity for
interfacing to a large number of converters.
The tag feature only works in the external clock mode and
requires that the DATA output of a “upstream” device be con-
nected to the TAG input of an “downstream” device.
An example of the concatenation of two devices is shown in
Figure 27 and their resultant output is shown in Figure 28.
In Figure 27, the paralleled R/C ensures that each AD977/
AD977A will simultaneously sample their inputs. In Figure 28,
a “null” bit is shown between each 16-bit word associated with
each ADC in the serial data output stream. This is the result of
a minimum value for “External Data Clock to Data Valid Delay
(t
18
) that is greater than the “TAG Valid Setup Time” (t
23
). In
other words, when you concatenate two or more AD977/AD977As
the MSB on the downstream device will not be present on the
TAG input of the upstream device in time to meet the setup
time requirement of the TAG input.
If the serial data stream is going to a parallel port of a micro-
processor that is also providing the serial data clock, then the
microprocessor’s firmware can be written to “throw away” the
null bit. If the serial data stream is going to a serial port then
external “glue” logic will have to be added to make the interface
work. If the serial port has a “sync” input then this can be used
to throw away the null bit if the sync input is toggled each time
the null bit appears.
If the application does not require simultaneous sampling, the
null bit can be completely avoided by delaying the R/C signal
of each upstream device by one clock cycle with respect to its
immediate downstream device. This bit time delay can be accom-
plished through a D-type flip-flop that delays the R/C signal at
its D-input by one cycle of the serial data clock that is at its
clock input.
DATA OUT
DCLK IN
R/
IN
IN
TAG DATA
DCLK
AD977/AD977A
#2
(UPSTREAM)
AD977/AD977A
#1
(DOWNSTREAM)
TAG
DATA
DCLK
CS
R/C
CS
R/C
Figure 27. Two AD977/AD977A’s Utilizing Tag
It is not recommended that the TAG feature be used with the
read during convert mode because this will require data to be
clocked out during the second half of the conversion process. It
is recommended that the read after convert mode be used in an
application that wants to take advantage of the TAG feature. To
improve the data throughput a combination of the two data read
methods can be used and is described as follows.
If two or more AD977/AD977As are to have their data output
concatenated together in a single data stream, and if data
throughput is to be maximized, a system could be designed such
that the upstream device data is read during the first half of its
conversion process and the remainder of the downstream devices
read during the time between conversions. Assume three AD977As
are to have their data concatenated. Assume the further most
downstream device is referred to as device #1 and the further
most upstream device as #3. Each device is driven from a com-
mon DATACLK and R/C control signal, the CS input of each
device is tied to ground. The three BUSY outputs should be
OR’d together to form a composite BUSY. After the conversion
is complete, as indicated by the composite BUSY going high, an
external, normally low, 15.15 MHz DATACLK can be toggled
34 times to first read the data first from device #3 and then
from device #2. When the composite BUSY goes low to indicate
the beginning of the conversion process the external DATA-
CLK can be toggled 17 times to read the data from device #1
during the first half of the conversion process. Using this tech-
nique it would be possible to read in the data from the three
devices in approximately 6.4 µs for a throughput of approxi-
mately 156 kHz The receiving device would have to deal with
the null bit between data from device #2 and #3. The receiving
device would also have to be capable of starting and stopping
the external DATACLK at the appropriate times.
The TAG input, when unused, should always be tied either high
or low and not be allowed to float.

AD977ABRSZRL

Mfr. #:
Manufacturer:
Analog Devices Inc.
Description:
Analog to Digital Converters - ADC 16-Bit 200 kSPS Serial I/O
Lifecycle:
New from this manufacturer.
Delivery:
DHL FedEx Ups TNT EMS
Payment:
T/T Paypal Visa MoneyGram Western Union