Attached is a block diagram of the circuit. I didn't think it was necessary to include the specifics of the amplifier. The voltage meter clearly states that the measured signal cannot share a ground with the meter's power supply. The negatives of the power supplies are not connected directly--although I can measure a voltage between them, and I assume that this relationship is the cause of the problem.

Yup, one of those eBay/China type designs. You'll need a separate supply or a $6 isolation converter.

 

is the opamp properly decoupled
Without the detail schema we cannot be sure.

Instead of asking provides all the details. this way we can help u sooner and better

 

Yup, one of those eBay/China type designs. You'll need a separate supply or a $6 isolation converter.

The meter is a LASCAR EMV 1125 from Newark. It was pretty expensive. Last one I used was from MPJA was much cheaper and worked fine with separate power supplies.

Maybe I should go back to the cheap one?

Can you recommend an isolator and/or a meter that will work?

 

The meter is a LASCAR EMV 1125 from Newark. It was pretty expensive. Last one I used was from MPJA was much cheaper and worked fine with separate power supplies.

Maybe I should go back to the cheap one?

Can you recommend an isolator and/or a meter that will work?

The isolator would depend on the voltage and current requirements of the meter, mine were the 5V red LED style and I used a NKE0505SC isolator.

Due to all the confusion there are quite a few inexpensive ones on eBay now that will work with a common supply.