Abstract:

I tried to recover the IFO and once LSC_LOCK guardian reached the OBSERVATION state by changing BS local control and BS ASC gain.
Since the stability is not good, further investigation is necessary.

Detail:

Since IFO cannot be locked after the long lockloss due to the earthquake, I strted investigate the reason.
Then, I realized BS yaw motion seems much larger than that before the earthquake (fig1: top: after earthquake, bottom: before earthquake).
Though I measured the OLTF of several loops, no critical loop was found.
So, one possibility is just large seismic motion shakes the suspension, so I changed TM_OLDAMP_Y filters to stronger one (fig2: before change, fig3: after change).

After that, I tweaked the alignment of IMMT2 and PR2 because IFO was still could not be locked.
However, no significant change can be seen, so I set the OpLev values of these suspensions to those when staring OBSERVING run on July 1.
I also tweaked ITMY BPC but no significant change could be observed, so I reverted the values of DRIVEALIGN matrix.

Then, I tried to lock with no ASC offsets and succeeded to lock PRFPMI_RF_LOCK though stability seems not good.
Since IFO can reach INCREASING_LAS_POWER state, I then reverted ASC offsets not to change the control parameters of ASC.
Soon after that, BS yaw started to oscillate maybe because of the local damping control change, so I set the gain of 0.8 (fig4) and then oscillation was stopped.

Then, IFO finally reached the OBSERVATION state though lock duration is just 1 minute.
Since I am not so sure what is the critical issues, further investigation is necessary to stabilize IFO.

Since IFO is not stable, I didn't accept the SDF diffs, and IFO won't go back to OBSERVING mode today.

Consideration:

There are several possibility, so I just list up what I think up to avoid forgetting.
1. Clipping inside PRC as discussed in klog34681.
2. Arm loss due to frosting.
3. Gain reduction of ASC due to PRG reduction.