To check if the lockloss reason is suspension or not, I kept IFO at 10W PRFPMI RF lock state and see how frequently suspensions are kicked.
Figure 1 shows the DARM error signals with 10-60 Hz bandpass filter.
Y cursor shows the rough threshold that causes the lockloss when IFO is OBSERVATION state.
The time distance between two T cursors is 1 hour, so large kick, which would cause the lockloss when OBSERVATION, happens four times in one hour.
Figure 2 shows the spectrum measured at the time including suspension kick.
In all cases, the 44.6 Hz peak was the largest, likely due to the ETMX resonance.
To mitigate the DARM residul motion due to the suspension kick, I implemented resonant gain at FM8 of DARM1 filter bank.
However, the lockloss happens due to the oscillation at 47Hz (fig3) roughly 3 minutes after engaging resonant gain.
I'm not so sure this oscillation comes from engaging resonant gain, it might not be a good way to engage resonant gains to mitigate the effect of suspension kick.
Note:
During this work, I modified the following filter banks and epics values.
FM8 of DARM1, FM1 of SRCL1, FM1 of ETMX_NBDAMP_V5, FM1 of SC-AA_DARM_ERR, FM1 of K1:LSC-AA_DARM_CTRL, and K1:VIS-ETMX_NBDAMP_INMTRX_15_36 (from 0 to 1).
