After we lost lock aligning the reference cavity, 49305, we had had two relocking attempts so far.  

Because Georgia and Niko had moved the ADS phase around last night, and Georgia looked at the HWS and saw that the beam was moved veritcally on ITMY, I stopped in DC readout to check all the ADS phases.  After ADS converged, I opened the loops, and moved the A2L gains to see if the response was in the I phase.  This was the case for P4, P5 and Y3 (ETMX P, ETMY P, and ITMY Y) but I changed the phases for the other loops.  After changing the phases, I checked that the old A2L gain setting still corresponded to a zero crossing, for two of the loops it did not so I found an A2L gain that gave us a zero crossing for the current alignment.  

We tried to power up this way, but there radiation pressure soft instability showed up at lower power, and in the X arm only (screenshot).  Since tuesday it has been a common soft instability.  As the 0.5Hz was ringing up, I tried to increase the CSOFT gain, which is normally 15, first to 30 then to 60.  It's not clear if this helped.  

We looked at the TMS QPDs, and saw that for TMSX we are nearly falling off the QPDs.  While sitting at DC readout, we moved TMSX pitch down by 8.6 urad to center the positions on the QPDs.  (We also turned off the offsets in the QPDs, which are no longer used at DC and therefore don't need offsets).  Then we lost lock because of the earthquake.  

We have replaced the TMSX test offset which is used by the dither alignment script to point the green beam to the baffle PDs (-34.1) with our offset of -8.6 urad.  Keita, Georgia and Jeff B will try engaging this in the next lock before powering up to see if this helps the QPD loops do a better job of suppressing the instability.  

We need to spend some time moving pico motors on TMS, because we are also nearly falling off the green QPDs which are used as our initial alignment reference.  That is a longer project though so we will leave it for another time.