I checked to make sure the line and the BP matches. Changing the phase both on the main PI medm screen and on the damping filter doesn't make a different. The PLL green light is on but Shelia mentioned that PLL is not actually locked and I don't know how to make it locked (some other modes with PLL green light indicator also have red PLL lock status). I also tried to zero the gain as Sheila suggested.
Then I found that until Maintenance morning (1 Nov 18:26 UTC to be precise) this mode has been using OMC DCPD as its input signal. It was magically switched to be using TRY QPD. I put the gain back to where it was in the matrices. We'll see how it goes.
Okay doing that somehow kill the input signal alltogether. I let through the signal from TRY QPD again but zero the damping gain.
Nevermind. Reverted it. Things became bad real fast as soon as I let through signal from TRY QPD for Mode27.
And PI broke the lock. This is the 4th lockloss caused by PI tonight.
And 5th time. The time constant is about the same as previous locklosses so I guess switching input from TRY QPD to OMC DCPD doesn't matter. In fact I couldn't get any signal from OMC DCPD so that's even worse since I have zero chance of damping it. I'm going to try the original setting again but this time alter the gain (which I have also tried).
And 6th time.