Instrumentation amplifier temperature drift problem AD8422

[qat]

Hi,

I have a circuit to precicely measure temperature, but I've stumbeled upon a stange issue.

The idea is that the system can use the analog multiplexer to measure across a known resistance to get an idea of the momentary offset of the measurement compared to the theoretically ideal voltage (to compensate for things such as temperature drift), then immediately proceed to measure across the thermistor and compensate using the offset. I've assembeled my board, and to test things out I've soldered on a precision resistor in the thermistors place, and almost everything is working fine.

The problem I have is that my tempearure drift is a lot higher than expected, but not always. When measuring across the "thermistor" (replaced by a precision resisotr) the voltage is bang on, and the temperature drift across a few degrees is almost not measurable, however, when measuring across the "resistor" (also precision resistor) the tempeature drift resulting from the AD8422 is almost 1mV/K! To simulate tempearure changes I've got a thin straw to blow air through to precicely target only the AD8422.

The two measurements are almost identical! The only exception is that the common mode voltage of one measurement is higher (around 7.1V vs. 4.1V), but that would only result in a greater offset voltage at the output defined by the CMRR?

Does anyone have any ideas why this might be? Should I increase R6 to decrease the CM voltage?

See my circuit attached to this post.

[David Hess]

You are exceeding the input common mode range of the instrumentation amplifier.  Both inputs have to be 1.2 volts above the negative supply voltage, and 1.5 volts below the positive supply voltage.

When measuring the resistor, the positive input is only 0.4 volts below the +10 volt positive supply which is too high.  Th 2 kilohm resistor needs to be at least 7.5 kilohms.

[qat]

You are exceeding the input common mode range of the instrumentation amplifier.  Both inputs have to be 1.2 volts above the negative supply voltage, and 1.5 volts below the positive supply voltage.

When measuring the resistor, the positive input is only 0.4 volts below the +10 volt positive supply which is too high.  Th 2 kilohm resistor needs to be at least 7.5 kilohms.

This is correct, thanks for the response. I tried swapping the 2k for a 10k, and had to increase the overall supply voltage and now it works as expected! It's strange how the issue manifested itself in the temperature drift