AD622
The AD622 provides greater accuracy at lower cost. The higher
cost of the homebrew circuit is dominated in this case by the
matched resistor network. One could also realize a homebrew
design using cheaper discrete resistors that are either trimmed
or hand selected to give high common-mode rejection. This
level of common-mode rejection, however, degrades significantly
Table 4. Make vs. Buy Error Budget
Total Error in ppm
Relative to 1 V FS
AD622
Homebrew
400
2.5
25
427.5
3300
210
0.12
3510.12
10
0.6
10.6
3948
1600
15
50
1665
3000
1080
9.3
4089.3
20
0.778
20.778
5775
over temperature due to the drift mismatch of the discrete
resistors.
Note that for the homebrew circuit, the LT1013 specification for
noise has been multiplied by √2. This is because a two op amp
type instrumentation amplifier has two op amps at its inputs,
both contributing to the overall noise.
Error Source
ABSOLUTE ACCURACY at T
A
= 25°C
Total RTI Offset Voltage, μV
Input Offset Current, nA
CMR, dB
DRIFT TO 85°C
Gain Drift, ppm/°C
Total RTI Offset Voltage, μV/°C
Input Offset Current, pA/°C
RESOLUTION
Gain Nonlinearity, ppm of Full Scale
Typ 0.1 Hz to 10 Hz Voltage Noise, μV p-p
AD622 Circuit Calculation
250 μV + 1500 μV/10
2.5 nA × 1 kΩ
86 dB→50 ppm × 0.5 V
Homebrew Circuit Calculation
800 μV × 2
15 nA × 1 kΩ
(0.1% Match × 0.5 V)/10 V
Total Absolute Error
(50 ppm)/°C × 60°C
9 μV/°C × 2 × 60°C
155 pA/°C × 1 kΩ × 60°C
Total Drift Error
20 ppm
0.55 μV p-p × √2
Total Resolution Error
Grand Total Error
(50 ppm + 5 ppm) × 60°C
(2 μV/°C + 15 μV/°C /10) × 60°C
2 pA/°C × 1 kΩ × 60°C
10 ppm
0.6 μV p-p
Rev. D | Page 10 of 16
AD [ ANALOG DEVICES ]
