which is the cause of a negative offset the output of an AD620 with zero input and zero reference, the circuit is show in the file attached. The problem is which the output of AD620 is aprox -0.5 V, with 0V of input and the reference at 0 V, which cause, that the output of the filter is zero while that the output of AD620 is negative.
how is fix the problem?
Your circuit looks pretty reasonable to me. The 100k pull up resistor seems a little small, you might want to bump it up to 1 Mohm. If this gave you some error, it would be in the positive direction, so I don't think this is the issue.
To debug, I would recommend replacing the thermocouple with a short as close as possible to the AD620. This will help you figure out whether the issue is from the thermocouple or from the AD620 circuit.
If you replace with a short and the problem goes away, I can think of three possible explanations:
1) If there is a temperature difference on the two ends of the thermocouple, it will generate a voltage. I don't know what gain you are using, but if it is high enough, it might be enough to generate -500 mV at the output of the AD620.
2) If you have very long thermocouple leads, it might be RFI rectification. It is normally good practice to put an RFI filter (as shown in the AD620 datasheet) both to prevent RFI rectification, as well as provide some input protection for the AD620.
3) The tip of your thermocouple is electrically conductive, and you are measuring the potential of whatever the tip is touching.
Note that the AD8221 is our next generation version of the AD620 and will give you a little bit better performance. If you are using a J or K type thermocouple, we also have some amplifiers specifically designed for thermocouple measurement, that have their own cold junction compensation: the AD8494-AD8497. You may also find this analog dialogue article helpful.