Sheila, Alexa, Dan, Evan

Summary

Tonight on dc readout, we started to reduce the coupling of alignment fluctuations into DARM by closing some more ASC loops. We focused mainly on the input and power-recycling optics.

During this work, the lowest value of REFL_A_LF observed was 0.4 ct (compared with 21 ct on resonance), meaning the interferometer visibility is 98%. The highest arm power observed was 1150× single-arm buildup, meaning the interferometer recycling gain is 34 W/W.

Details

List of loops, and things tried:

• We were able to close REFLB9I → IM4 pitch with an OK bandwidth (> 0.1 Hz, based on step response). Here we have a 1/f loop shape everywhere (FM2+FM3 in INP1), and a gain of −1.0 ct/ct. We may want to use REFLA9I − REFLB9I instead (see below). Closing the loop had an immediate improvement on the mode seen on the REFL camera.
• For IM4 yaw, we noticed that the combination REFLA9I − REFLB9I gave a sensible error signal (as was the experience at Livingston [LLO#16254]). But the lock broke while trying to close the loop, and we did not pursue it further. The most egregeous alignment fluctuations in the corner appear to be in pitch anyway.
• At one point were able to close REFLA9I →PRM pitch, but it seems this error signal is only good intermittently. Possibly it requires the input pointing into the PRC to be good already (i.e., we may need to close it after the IM4 loops).
• We looked for a while for a good signal for the common ETM motion, but none of the responses in REFL9 or REFL45 seemed immediately good.
• The situation with AS36 → BS is still mysterious; neither ASA36Q nor ASB36Q seemed to give good error signals.
• The differential ETM loops ("DHARD", but not actually DHARD) seem to have more optical gain now that the SR2 position has been moved. Good filter module gains are 8 ct/ct for pitch and 80 ct/ct for yaw; these are possibly conservative for the 2.8 W lock. For the 8 W lock they may need to be reduced a bit further still. Otherwise they continue to work fine.