I have swapped out both background detector preamps for preamps from Dave's spares box, configured by Scott with gain 2MΩ and offset +0.1V.
This removes a negative 1.2 V offset from isource and pmt_led.
I identified an even number of cable labeling swaps between aux2 and aux3, at the preamp. Both aux2 and aux3 respond to LED light (but aux2 does not respond to LED light in aux3 or vice versa, which didn't make sense until I figured it out). This means that the detector under study in the tee'd preamp business was pmt_only, rather than pmt_lightguide, and that the narrow- and wide-band channels in the analysis were reversed. However data before and after the tee'd preamp study are mapped correctly.
I had four preamps (negative offset, killed by overvoltage, blows a fuse, and wideband) surveyed out of the hall and will get them moved to EEL tomorrow.