Related: https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=63422
1. Strange offsets in M0.
In the first attachment, you're looking at the Euler-based OSEM DAMP values, Voltmon (analog readback of low-passed coil output voltage) and MASTER_OUT (last stage of coil output before DAC) for ETMX M0 stage. At around t=-4d on this plot is the lock loss of the last lock before Monday evening, and at around -1d they manually put big offsets to V and T.
Anyway, if you look at MASTER_OUT for LF, RT and SD (enclosed by a green box on the plot), they were very quiet after the lock loss until t~ -2.75d (~May/30 22:40 PDT), became really noisy at around -2.5d (~ May/31 06:40 PDT), but the mean of the digital outputs were always about zero until they manually put V and T offset.
However, if you look at VOLTMON for LF, RT and SD (yellow box), they started showing some offset at t~-2.75d and then a big offset of ~-140 counts at t~-2.5d.
These offsets are likely real (i.e. not the readback problem) because you can see that both R and T jumped at the same time without any reason (red boxes).
I believe that the offsets in VOLTMONs are gone now (see the next paragraph), if there indeed were analog offsets.
Looking at good coils (F1, F2 and F3), at DC, VMON/MASTER_OUT~5.05E-2. Right now, all of the three weird coils have almost exactly 20000 (or -20000) counts in MASTER_OUT and about 1010 (or -1010) counts in VMON, so VMON/MASTER_OUT=5.05E-2, these are almost exactly what they're supposed to be. If the offset of ~-140 count is still there in the analog world we could easily spot that because the ratio should be more like (1010-140)/20000=4.35E-2 or (1010+140)/20000=5.75e-2. It seems like the offsets are gone.
It could also be that, instead of simple analog offsets, there could have been analog saturation somewhere, which was clipping the waveform asymmetrically around zero when things were noisy, which of course makes DC offsets. It's hard to imagine that mere 1000 counts in MASTER_OUT causes saturation regardless of the frequency, but it could have been analog oscillation. But then I don't know why there were similar offsets in three coils, and the question remains why the problem went away when they put offsets on LF, RT and SD.
Either way, since offsets seem to be gone, at some point you could remove V and T offsets to see if the problem comes back.
2. Strange offsets in R0. They're still causing problems.
I got curious and checked R0 channels (2nd attachment, top 2 rows), and offsets showed up in R0 LF and RT but not in SD. However, SD as well as F1 got somewhat noisier than before. And these are still unfixed.
If you look at the 3rd attachment, to the left are M0 EULER based DAMP input signals and to the right are R0. References are before the offset showed up and current traces are now. You want to compare red with pink, blue with light blue and green with darker green. For R0, all current traces are noisier than the references.
You could put offsets on V or R and see if the problem goes away. If it does, disable the offset and see if the problem comes back.