Today I rephased the AS WFS 45 segments.
To avoid lockloss, I transitioned the DHARD input matrix to AS WFS 45 A only so I could phase WFS B. Then, I switched DHARD to WFS B and phased WFS A. I switched by using the DHARD blend filters. Set the DHARD_P_B and DHARD_Y_B input matrix values to the desired value, then make sure the DHARD A and B blend filters have the same ramp time. Then, ramp the gain of one blend to zero and the gain of the other to one. I used a similar process here: 85774.
To get the phasing, I drove a DARM line at 255 Hz in DARM2 EXC, after engaging the EBS255 filter in DARM2 FM10. I was able to get good SNR with an excitation of 3000 counts.
I adjusted each segment to reduce the signal in I (red trace) and maximize in Q. I was able to reduce the line height by between a factor of 2 to a factor of 5 on each segment. The phase changes were between 3-6 degrees on each segment. The light pink and blue traces show the phasing before I started.
This change was SFDed in both safe and observe, shown in attachments below. For some reason, AS A RF45 is in the ASCIMC model but AS B RF45 is in the ASC model.
I attached a comparison of the before and after for each segment. (Note, seg2 on WFS 45A had a strange behavior where it appears the line in Q reduced with the phase change, but that reduction was present before I started changing the phase.)
I have reverted these phases since we don't see much benefit to using them right now, and there is a negative impact on the roll mode damping.
I tried testing these phases while relocking. We can still lock RF darm with no problems. However, it doesn't make any difference in our ability to engage DHARD right after RF darm is locked ( in other words, we still can't engage DHARD after RF darm).
Since this doesn't give us any benefit, and screws with our roll mode damping, I say we don't use these new phases.