georgia.mansell@LIGO.ORG - posted 18:19, Wednesday 27 July 2022 - last comment - 11:17, Thursday 11 August 2022(64181)
More work on switching OMC to dither alignment
Today I phased the OMC dither loops. I stepped the offsets in the 4 OMC ASC loops (POS_X POS_Y ANG_X ANG_Y) while the dither lines were on, and looked at the P1, P2, Y1, Y2 I and Q signals. When the offset is added we want to only see a response in I. I ended up setting the phases to minimize the response in Q.
To see a step response I put these offsets in the DOFs:
POS_X: 3
POS_Y: 1.5
ANG_X: 4
ANG_Y: 6
The phases I set were:
P1 120 (was 100)
Y1 130 (was 115)
P2 150 (was 130)
Y2 -20 (was -55)
While I was adding offsets I noticed that Kappa_C *may* have been slightly higher for an ANG_X offset of ~3. Perhaps we want to explore some different yaw offsets on the OMC QPDs.
I started stepping the QPD offsets to minimize the P1 P2 Y1 Y2 error signals and the OMC suspension started hitting its rails. It would be good it we could offload the OMC alignment onto the OMs so that we can take the OMC alignment to where the dither alignment says it should be.
I think I found the time when this test was performed. The time when all four dither error signals are closest to zero, around 22:30:00 UTC, shows large positive offsets being pushed to OMC suspension coils T2 and T3 (the pink and the brown trends in the attachment are totally overlapped). This corresponds to a combination of V and R in the OMC's suspension basis, but we would think of more usually as corresponding to pitch/height of the beam relative to the OMC. In other words, we are trying to use the relatively short lever arm of the OMC suspension to achieve a vertical/pitch beam alignment that is much more efficiently achieved with the longer lever arm of the OMs.
Since OM3 is already being used for OMC alignment, offloading the OMC suspension alignment control requires touching some combination of OM1 and OM2. In that case, it seems more direct to just change the actuation matrix to feed back to OM1,2,3 and leave the OMC suspension out of it, which is the Livingston strategy. Their actuation matrix could probably be imported wholesale (Hanford/Livingston comparison attached). This would mean giving up centering on one of the AS WFSs. Since WFS B is not being used for anything, it seems like the natural choice (also a Livingston strategy, and their AS centering matrix could probably be imported too).
Worth noting that way back when the OMC alignment was first being commissioned in 2015, pushing counts to the OMC suspension was known to cause scattering from 40 to 200 Hz (LHO:17264, LHO:17273, LHO:17290).
The squeezer alignment would be using both WFS, no?