Instrument Amplifier (AD623) incorrect (too low) and inconsistent gain

Question

I'm getting an incorrect and inconsisten gain from an AD623 instrument amplifier. The datasheet states that RG = 100 kΩ/(G − 1) where RG is the resistor between pins 1 and 8 and G is gain. My R3 is 268ohm which should provide a gain of 374.

R2 is a current shunt. R1 and R4 are power resistors.

Resistor values and voltage values are as measured.

• Using R1 I'm measuring 4.2mV across the shunt (R2) and reading 0.52V on the meter. Gain of 123.8. Expected 1.57V.
• Using R4 I'm measuring 6.35mV across the shunt (R2) and reading 1.175V on the meter. Gain of 185. Expected 2.37V.

I've considered:

• Are the voltages beyond the rails. No, rail voltage is 5V which I've measured at 5V.
• Do I have enough current into pins 2 and 3 - I don't think i'm doing anything that would limit the current.
• The datasheet shows a settling time of a few hundred microseconds. My results are stable and incorrect so I don't think this is the issue.
• Is my voltage too small. I can't find anything in the datasheet that would indicate this is a problem.
• Is my chip damaged. I have two and they're both showing the same result (they could both be similarly damaged).

Any suggestions as to what I'm doing wrong?

^{simulate this circuit – Schematic created using CircuitLab}

Answer 1

This answer was given before the op changed their mind about the circuit.

Your power supply is 5 volts and you are trying to measure a voltage that is raised up at 12 volts. Have you checked the input common mode voltage range for the chosen device?

Have you also confirmed that the ref pin works ok when tied to the most negative rail in the system. It happens to be ground in your circuit but some IAs don't like ref being attached to their negative power rail.

Answer 2

Since you are drawing multiple amps through your shunt, and you have the - input tied to power ground, I suspect that you have got an unwanted voltage drop on that input.

Instead of measuring your input voltage at the shunt, make a differential measurement at the op amp. Connect your meter ground to pin 4. Then measure to pin 3, and then to pin 2. Subtract the two to get your real input voltage.

In general, for this sort of measurement, you are better off isolating the inputs somewhat. Try inserting 10k resistors before BOTH inputs.

^{simulate this circuit – Schematic created using CircuitLab}

Also, make your shunt connections as a Kelvin configuration. Run separate traces or wires from the amp to the shunt independent of the power/ground connnection, and make the connections directly to the shunt.