AD708
Rev. C | Page 11 of 16
OPERATION WITH A GAIN OF −100
To show the outstanding dc precision of the AD708 in a real
application,
Table 3 shows an error budget calculation for a gain
of −100. This configuration is shown in
Figure 28.
Table 3.
Error Sources
Maximum Error Contribution
A
V
= 100 (S Grade)
(Full Scale: V
OUT
= 10 V, V
IN
= 100 mV)
V
OS
30 μV/100 mV = 300 ppm
I
OS
(100 kΩ)(1 nA)/10 V = 10 ppm
Gain (2 kΩ Load) 10 V/(5 × 106)/100 mV = 20 ppm
Noise 0.35 mV/100 mV = 4 ppm
V
OS
Drift (0.3 mV/°C)/100 mV = 3 ppm/°C
Total Unadjusted
Error @ 25°C = 334 ppm > 11 bits
−55°C to +125°C = 634 ppm > 10 bits
With Offset
Calibrated Out @ 25°C = 34 ppm > 14 bits
−55°C to +125°C = 334 ppm > 11 bits
1/2
AD708
V
OUT
V
IN
+
–
100k
1k
1k
2
3
7
4
6
0.1µF
0.1µF
+V
S
–V
S
5789-028
Figure 28. Gain of −100 Configuration
This error budget assumes no error in the resistor ratio and no
error from power supply variation (the 120 dB minimum PSRR
of the AD708S makes this a good assumption). The external
resistors can cause gain error from mismatch and drift over
temperature.
HIGH PRECISION PROGRAMMABLE GAIN
AMPLIFIER
The three op amp programmable gain amplifier shown in
Figure 29 takes advantage of the outstanding matching
characteristics of the AD708 to achieve high dc precision.
05789-029
S1
S2
A0
A1
S3
S4
S8
S7
+V
S
–V
S
S6
S5
AD7502
OUT
1–4
OUT
5–8
1/2
AD708
V
INA
1/2
AD708
V
INB
10k
26.1
100
10k
10k
26.1
1k
10k
10k
26.1
10k
10k
AD707
10k
10k
9.9k R
A
9.9k
R
B
Figure 29. Precision PGA
The gains of the circuit are controlled by the select lines, A0 and
A1, of the
AD7502 multiplexer, and are 1, 10, 100, and 1000 in
this design.
The input stage attains very high dc precision due to the 30 V
maximum offset voltage match of the AD708S and the 1 nA
maximum input bias current match. The accuracy is main-
tained over temperature because of the ultralow drift
performance of the AD708.
To achieve 0.1% gain accuracy, along with high common-mode
rejection, the circuit should be trimmed.
To maximize common-mode rejection
1. Set the select lines for gain = 1 and ground V
INB
.
2. Apply a precision dc voltage to V
INA
and trim R
A
until
V
O
= −V
INA
to the required precision.
3. Connect V
INB
to V
INA
and apply an input voltage equal to
the full-scale common mode expected.
4. Trim R
B
until VB
O
= 0 V.
To minimize gain errors
1. Select gain = 10 with the control lines and apply a
differential input voltage.
2. Adjust the 100 potentiometer to V
O
= 10 V
IN
(adjust V
IN
magnitude as necessary).
3. Repeat Step 1 and Step 2 for gain = 100 and gain = 1000,
adjusting the 1 k and 10 k potentiometers, respectively.
The design shown in
Figure 29 should allow for 0.1% gain
accuracy and 0.1 V/V common-mode rejection when ±1%
resistors and ±5% potentiometers are used.