Inconsistent and erroneous analog differential measurements

I am having some difficulty getting consistent results using the Mayfly’s auxiliary 16-bit ADC. I have multiple stations set up that use the same code and wiring, but for some reason the measurements being recorded are not behaving as I would expect. For context, I am using Apogee’s shortwave pyranometers (both upward and downward looking) on the Mayfly’s auxiliary ADC. The measurement is taken off of a differential voltage reading between two wires connected to the thermopile of the sensor, so I’m taking two differential measurements, one for incoming shortwave radiation and one for outgoing. One of the problems I’ll have is that some stations, with some sensors, will read values outside of what the sensor is capable of reporting, even though I have taken a multimeter and measured the differential voltage on the terminal screws where the wire is plugged in and that reading is normal and does not match what is recorded. Another problem I’ll have is that sometimes a station will record data that is behaving as expected (with diurnal cycling that drops down to essentially zero at night and peaks midday), but then it will randomly stop working and not return to normal even though I have not accessed or altered the station in any way.

My guess is that I am not using the ADCs correctly, but I’ve looked through my coding multiple times over and I can’t seem to determine why I’m getting such conflicting results, especially since I do have one station that is operating perfectly with the same coding and wiring. It should be noted that I do have a couple extra Adafruit ADCs attached to the Mayfly that I have given different addresses. Again, I have one station that is running perfectly well with this setup, as far as I understand, that is not causing me any issues and has the same wiring and coding as the other stations. Any help would be greatly appreciated as I am trying to have these stations ready for winter which is quickly approaching.