With the help of the SUS folks, I increased by a factor of 3 the gain on the quad yaw and vertical (both M0 and R0 on ITMY and ETMY). I also added 0.45Hz boosts to L and T for the transmon damping. All of these changes are aimed at making sure that the ISI does not see sharp resonant features from the suspensions.
I started the evening with the PLL unlocked, and the cavity badly aligned. The cavity alignment problem came from the HEPI, which had tripped and thus let go of its 1M count RZ offset. It turns out to be surprising difficult to restart HEPI without the watchdog tripping;
Or maybe we should just consider making the WD less sensitive.
The PLL was not working because the CM board was not set properly. The compensation filter was off, and the boost was 1. I guess these should be {on, 0}, or at least it works like this. Similarly the PDH CM board had the common option on, which made it not work either. We should really find a better solution.
After fixing these problems, and tweaking the alignment with the ITM and ETM, the cavity locked and was very stable. We should check the state of the ISIs during this time.
Plot requested by Jeff K.
boost=1 and no comp for PLL means that something/somebody attempted to copy the PDH CM setting to PLL.
But PLL was left locked when I went home. There's no good reason the setting should/could have changed. Puzzled.