We were just trying a TSAMs change after the RH moves, see if the optical gain still gets worse as we turn it up (it still reduces optical gain).
Around this time we saw twice a big glitch at 280Hz, Which looks like the same one I saw during my last trip and appears to be coherent with ASC motion (66035).
There's a big shift in alignment in the Y position and angle sensed by the OMC angular dithers. This seems very similar to what we've seen twice at LLO (LLO alog 63995). Looking at the OSEMs, OM2 has a large wander in pitch, and to some extent in yaw. It reverses exactly when the glitches happen, and the sudden shift is seen on the dithers. OM3 can be seen following the POS/ANG signal. See the attached plots. Like LLO, the OMC already has strong lines around 270 Hz in pitch and yaw as seen by the QPDs, which are already there before the glitches in DARM. The misalignment just couples them in, with a lot of nonstationarity and glitchiness. See page 3 of the attached plots. We can compare the expected feedback from the alignment to what the OSEMs actually do. It seems in both the LLO and LHO cases like OM2 is responsible and is not following the control signal.
After checking MEDM, it seems that neither OM2 nor OM1 are being controlled by the angular loop (at least in this lock). So OM2 had a very large drift, hit some limit, and suddenly reversed direction. Then the beam is misaligned for a few minutes until the angular loop can get the beam realigned into the OMC. To prevent this would require keeping OM2 from suddenly reversing direction (or wandering to its limit in the first place). Note for detchar - LHO is using the QPDs for alignment, not dithers, while LLO is using dither.
Ansel, Robert, and I had done some previous investigation of noise near 280 Hz. (See alog 68261 for combs-related follow up). We had noticed glitching near 280 Hz in HVeto runs that was happening at the same time as glitching at ~35.4 Hz in DARM and ASC channels. This glitching was present in LHO data since the Oct 2022 commissioning break, although the rate of this glitching varied with time.
- HVeto example in Feb 2023: https://ldas-jobs.ligo-wa.caltech.edu/~detchar/summary/day/20230208/detchar/hveto/
- HVeto example in Dec 2022: https://ldas-jobs.ligo-wa.caltech.edu/~detchar/summary/day/20221209/detchar/hveto/
- Example omega scan: https://ldas-jobs.ligo-wa.caltech.edu/~detchar/hveto/day/20230208/1359849618-1359936018/scans/1359907908.523437/
Looking closer at the three lines near ~280 Hz, we noticed that there were sidebands surrounding all lines with the same pattern (see attached images). These sidebands only appeared after switching into the low noise ASC Guardian state. However, the sidebands were no longer present a week after Feb 11. This was after CHARD YAW sensing was switched from RF9 to RF45, but we aren’t sure if this was actually related to the sidebands. The lines themselves were still present immediately after Feb 11. We didn't identify any HVeto runs after this date that have picked up the same type of glitching.
Following the connection between 280 Hz and 35.4 Hz, Ansel, Robert, and I investigated potential PEM sources of this noise at 35.4 Hz. We noticed that a line was present in many of the LVEA accelerometers, including PSL, HAM1, and HAM6 accelerometers. However, the strongest correlation between DARM and the HAM6 accelerometers. We investigated if the source of the motion was near HAM6 or the electronics bay, but did not identify a clear source. We did notice some time variation of the peak position, ~a few hundredths of a Hz on weekly/monthly time scales, with the peak width <1/100th Hz on hourly timescales.
