J. Kissel, O. Patane We happened to be catching up with the aLOG at the moment (minutes) the IFO began exhibiting "the" 1 Hz oscillation that has an unknown source and yet has troubling the lock acquisition sequence since the vent (and had gotten worse yesterday after maintenance). We found Elenna mentioned increasing the ASC CSOFT P gain from 20.0 to 25.0 helped last night (LHO:85643), so we repeated that gain increase in CSOFT P with a 5 sec ramp (the ISC_LOCK guardian was mid "INJECT_SQUEEZING;" this state is unrelated to this transient oscillation, as the latter takes ~minutes to ring up, but I mention it to inform outsiders that this is occurring relatively late in the sequence). The gain increase worked again, and seems to have saved this lock stretch. We've accepted a gain of 25.0 for the channel H1:ASC-CSOFT_P_GAIN in the h1asc model's OBSERVE.snap, but will leave it to others to programmatically do this in the guardian (or fix the problem "in hardware" by other means.)
I quickly took two different measurements of CSOFT P right after we locked (Oli was getting the filer cavity locked during this time so we were already out of observing), and it seems like CSOFT P is slowly gaining phase around 1 Hz as we thermalize. This is probably why we need a little extra gain at the start of the lock. Overall, this loop design is just not good, and needs to be fixed! I also don't think we want to always have this gain high once we thermalize.
For now, my hacky fix is this: I have ISC_LOCK set the CSOFT P gain to 25. Then, I added a 30 minute timer to the thermalization guardian. When that timer completes, the CSOFT P gain will be ramped back to 20. I have unmonitored this gain in the SDF for now. I am going to try to get this measurement plotted with a model and try to get this loop redesigned so we don't need this hacky fix. These guardians are loaded, and CSOFT P gain is SDFed back to 20.
Attachment 1: and incredibly mediocre measurement of CSOFT P. Dark blue ref is an old ref from 2023. Green ref was the first measurement I took today, red ref is the second measurement from today. You can see that we seem to gain phase at 1 Hz between the green and red measurements. However, we have a large amount of gain peaking near 1.5 Hz with this higher gain.
Attachment 2: unmonitoring CSOFT P gain.
The thermalization guardian never reduced the gain even though we are 52 minutes into the lock, so I probably made some mistake. However, all is well so far.