[Evan, Swadha]
We are trying to find a way to prevent PR3 angular motion from being impressed onto the test masses by the ASC. We had previosuly attempted unsuccessfully to diagonalize PR3 motion out of the REFL sensing matrix. Since that did not work, we are trying to find a potential error signal for suppressing PR3 motion.
We tried to close a loop around PR3 pitch using POP_A. We took the matlab plant model for HLTS M2 torque to M3 pitch, inverted it, and put it into the PRC1 filter bank (attached for pitch). We then turned on a 1/f filter. Driving M2 with this filter and turning up the gain, we found that the error signal was suppressed but microseism was being impressed onto PR3 as seen in the oplev spectrum (attached). Therefore, POP_A does not seem to be a good sensor.
The attachment shows the appearance of PR3 pitch motion in the REFL WFSs driven at 8 Hz with notches in the ASC loops. Since it shows up with opposite sign in the 9 MHz and 45 MHz signals, while cHard shows up with the same sign in all signals, we had thought we could tweak the cHard error signal (it uses all 4 of A9I, A45I, B9I, B45I) to remove PR3 motion. However, the resulting error signal had increased microseismic motion and motion above 10 Hz (presumably HAM1 motion). This was not placed in loop. There was no discernible reduction in the coherence between the proposed error signal (inserted into CHARD_P_B) and the PR3 oplev.
