After that, we found that when the LSC_LOCK was at "MICH_3F_LOCKED" state, BS P oscillated with 0.65Hz. 

The attached fig is of at the timing of jump from HANDOVER_ALS_TO_IR to DOWN due to failing to pass the decorator check_als_xy. In the Guardian log, it seems CARM IN1GAIN can reach 13 (close to the end of increasing this gain...), but immediately after that, this decorator rises the false flag.

Seems 4 Hz oscillation in GRX TR?

Thank you for reporting several issues.

*GRX lockloss during initial alignment:
Since GR frequency is phase-locked to IR laser, which is locked to IMC resonance, GR laser frequency drifts and oscillates, which causes saturation of feedback signals.
So, if the GRX lock was lost during the initial alignment, please request ALIGNING_XARM_WITH_GRX again after confirming GRX is locked with TEM00 mode.

*BS oscillation during MICH_3F_LOCKED
At that time, MICH feedback signals are totally saturated (fig1) and that cases the saturation of local control of BS (both local control and IFO control are fedback to TM stage).
So, this oscillation was coming from MICH lock failure.
I will check mich lock check decolator in the guardian.

*Failure of HANDOVER_ALS_TO_IR
Since gain of each control filters are optimized two or three month ago, the IR laser power on each RF PDs can be changed because of the opening of several GVs (BS-ITMs and PRM-PR3).
Power normalization was done in the guardian but normalization factor is decided by IMC transmission power, so it cannot compensate the power increase afteer IMC.
Probably, gain optimization in the current condition is necessary to achieve stable handover.

*Bug in the guardian:
I fixed the bug in the guardian.
Now, DARM blending and CARM blending are engaged in series.

I found that I forgot to rotate back the REFL HWP after the PRC work yesterday.
So, this might be the reason that IFO was not stable in the morning.
Sorry for my mistake.

I implemented the check of HWP angle in the LSC_LOCK guardian (SET_LASER_POWER_NORM state).
If the HWP angle is far from nominal (95 degrees for FPMI), following notification appears:
'REFL HWP angle is not norminal value. Please rotate HWP to 95 degrees.'

So, please manually rotate the HWP if you find this message in the guardian.
HWP can be rotated from the HWP medm (sitemap -> Commissioning Dock -> HWP -> HWP motor calib or sitemap -> SYS -> HWP motor calib).
You need to set total(deg) value for IFO_REFL to 95.

I changed the CARM input gain from 13 to 11 for stable FPMI locking.
Figure 1 shows the OLTF of CARM loop.
UGF is 46.2 kHz and phase margin is 38 degrees, which are almost the same as before (klog22460).

This modification has been already implemented to the guardian.