[Jenne, Niko]
Last week I discovered that the MICH_BRIGHT_ALIGN state of initial alignment wasn't waiting long enough (or waiting at all) for convergence (alog 54802).
Today Niko changed the convergence thresholds from 10 to 0.5, and that indeed made the initial alignment guardian wait until the ADS servos were closer to ready before offloading.
However, I don't think that the BS ADS setpoints are actually where we want them. After initial alignment was complete, Niko went to MICH_DARK_LOCKED and hand adjusted the BS since it was clearly not well aligned. I then turned on the ADS dither lines (but not the servos), and saw the lines quite clearly in the H1:ASC-AS_A_DC_NSUM_OUT_DQ (although I did not check where they were in MICH Bright, since it was time to head back to Observing). The BS M2 Y2L gain is currently set to 11, and was last changed in ~August 2018 when we were doing TCS tests to check the contrast defect. Before then it had smaller but still nonzero values.
Next week during maintenance I'd like to revisit this by hand-aligning the BS in MICH_DARK_LOCKED, and then using that alignment in MICH_BRIGHT, and set the A2L gains for BS to minimize the lines in the ADS error signals. We may want to revert to the current NomLowNoise values of P2L=0, Y2L=11 in the DOWN state if those are indeed giving us the best A2L decoupling at full lock, but they seem to not be putting us in a good spot for acquisition.
This morning I went back and forth between MICH dark and MICH BRIGHT a couple of times to check the alignment that we get from MICH_BRIGHT, it seems like mich bright and mich dark do give us the same alignments.
The sensor correction was off while I was doing this, so the sopt is moving around and I cannot get as nice of a dark michelson as we normally see with sensor correction on. I started with the dark michelson, checked that it was well aligned. Then I locked the birght michelson and noticed that the ADS error point was around zero. Since the ADS demodulates the AS DC power for the BS alignment, the BS A2L gains don't have an impact on the signal (I double checked this by changing Y2L and saw that as expected it has no impact). I then went back to the dark michelson and the alignment still looked fairly good.