We had some trouble with locking due to ASC this afternoon.
We had some locklosses when engaging ASC. We have been engaging all the QPD and WFS loops first, then adding the dithers after, but we have found that sometimes engaging all the loops other than the dithers brings the build ups down, until the dither loops are engaged. This afternoon Georgia and Jenne moved the DOF3 dither loop (ITMY-> PRM) engagement so that it comes on at the same time as the WFS loops. This seems to be working. I also increased the tolerance on some of the convergence checkers in the engage_ASC state, which has been taking ~10 minutes. It looks like there are some places where we are increasing gains after integrators in these states, which is why we need the long waits for offloading, but we could avoid this by moving the gains upstream of the intergrator.
We also lost lock twice in LOWNOISE ASC tonight, there is a large transient when we reduce the dither amplitudes and increase the dither loop gains. Georgia and I went through the dither loops and set them all up to be engaged with their final loops gains, so that we will no longer be increasing the gains of the loops in this state.
Jeff and I also had one lockloss where POP90 became large in final stage of the CARM offset reduction, which is something that I have seen happen before.
Georgia found that at 2W we can get higher POP18 build ups and lower POP90 with an offset in the SRC2Y SRC1Y loop.
Twice in a row we've lost lock during ENGAGE_ASC_FOR_FULL_IFO with the PRM ADS loop coming on at the same time. It seems mostly like a SRC problem - I could move SRM and get POP 90 down, and POP 18 and PRG up, but I'm just fighting the ASC, and if I stop clicking, the ASC pulls the SRM back to a bad place.
TJ and Ed are doing a wfsreliefpast to the time of my last lock this morning, which was right after I did an initial alignment.
We never offload our full IFO ASC to the suspensions because we have historically been concerned about wire heating, but perhaps with all the new baffling, we could offload our ASC (as LLO is talking about doing LLO-43581) after the SOFT loops have engaged and converged, before we increase power beyond 2W.