We retested the failed ai channels. Most of them got fixed by applying the inverse calibration in the yaml file.
We still find some anomalous channels, mostly in the DB25 connector. Turns out, their limits were ill-defined in the EPICS database. Specifying the right limit fixed the issue.
We find one miswired channel (BNC4). We connected the BNC to the right channel on the Acromag unit which fixed the issue.
Overall all the ai channels were successfully bench-tested.
I checked the failed channels against the EPICS database definitions and the yaml file inputted to IFOTest. The channels where the readings are something other than +10/0 V, but the high/low values do change, I think can be attributed to one of two things:
I fixed the channel gains/offsets in the master yaml file (PSL.yaml). I also disabled smoothing in the EPICS defintions of the new PSL channels for the purpose of testing. We can uncomment those lines after installing the new chassis if noise is a problem. Please go ahead and re-test the channels again.
I marked in red the failed channels. Some of them are probably calibration issues, but the ones that show the same voltage for high and low are probably disconnected wires.