We have a design for 2 isolated low-frequency voltage measurement channels, supplied by ADUM5401, using ADR03AKS reference, and the AD7924 ADC. We use only channels 0 and 1.
Everything seems to work well, but we have had a few cases where the converter would return results about 31% high on both channels at the same time. I have 1 sample that acted up in the system. On the bench I saw the problem, probed around and the problem disappeared. The same board has the problem again when reinstalled. Back to the bench, the problem is gone. Another sample in the system is working fine.
I'm thinking something like ESD or common mode issues, but the problem is there even when the our power converter hadn't run yet. We have 6V1 (yes a bit high) ESD zeners and 100k resistance in the divider, plus 100nF ceramic, in front of the converter for protection.
Before I replace the chips and test, I want to duplicate it reliably. The fact that something happened when I was probing suggests compoent failure, but I have heated and cooled the circuitry, and the error % doesn't correspond to a failure by some passive component. The data sheet says the converter can take 100mA without latchup. The on-off failure mode would be even stranger for the bipolar reference chip.
Has anyone else seen this behaviour? Any clues would be appreciated.
PCS Power Converter Solutions