We are using the AD73360L as AFE for a industrial product. We have found that if the unit has been left unpowered for +10 minutes and powered up again that there is a offset on one or two of the channels. We investigated this more by putting the unit in a oven to heat it up, where we found that the offset stayed the same. But when cooling the unit and then powering it up, there is a offset on our values. This offset does disappear after the unit is left running for 30-40minutes.
The unit was calibrated both offset and amplitude. We calibrate the offsets by letting the unit run for a minute or two to get the temperature to settle. Then without injecting any current we calibrate the each channels offset by taking the sum of 640 samples and dividing it by 640 to get a average offset value. The ARM doing the calibration also takes a 1000 samples to calculate the average offset value and they correspond.
It is just very interesting to us that the offset is a bit bigger when the unit is left switched off and then switched on (especially only on one or two channels). Then after being switched on for a few minutes the offset drifts back to zero.
We are using the AD73360 in full differential mode and sampling at 32 kHz. We did our layout as close as possible to examples and reference designs we could find on the AD76330. Is there anything we are missing? We first noticed this without cooling the unit in a cooler. But to make sure it is temperature effected we took it to extremes but never more than was specified in the datasheet.
We have changed the anti-aliasing filter and even took it off but got the same offset drift. Once it settles (after 30-40minutes) it stays stable. But then we switch the unit off and leave it off for +10 minutes and switch it back on and the offset is there again.
Any help or ideas would be appreciated.
Quintin van Wyngaardt
Sorry about the delay, since this was originally posted in a different section we weren't directly notified. I will help you out with this problem.
First off, is this still an issue?
If it is, then my next question is, are you using CT's as the current sensors in the application?
From the description of the issue, it sounds like it could be a capacitor that is the problem. In particular C38, 39, 42, 43, 46, 47, 50, 51, 54, 55, 58, and 59, all the capacitors in series with the input signal path. One good test for that would be to short VINPx and VINNx, to see if the offset still exists. Or by shorting out the caps listed, this should be ok if Current Transformers are used as the current sensor.