Sheila, Gabriele, Stefan, Alexa
Today we managed to close all DRMI ASC loops.
- REFLA9I - REFLB9I --> IM4 (BW: ~100mHz)
- POP A DC -->PRM (BW: ~10mHz)
- REFLA9I --> PR2 (BW: '100mHz)
- ASB36Q --> BS (BW: 3Hz)
- AS_C --> SR2 (BW: ~1mHz)
- ASB36I --> SRM (BW: ~100mHz)
The gain and filter settings are depicted in the first attachment. We also have decoupled AS_C from SRM with an adjustment in the output matrix (see second attachement). The BW are all rough estimates. Both POP A DC (P,Y=-0.03. 0.3) and ASC_C D (P,Y = -0.5, 0.1) have offsets that correspond to good buildups. So far we have been able to close these loops several times even with re-running initial alignment. This is all in the guardian now.
Originally we were only feeding back to the bottomg stage (except for BS which feeds back to M2). We have added offloading capabilities to the top stage for all the optics. As of now, we have successfully offloaded to the top stage for BS, PRM, and PR2.
In addition of what reported above, we also diagonalized the SR2/SRM actuation. With SRC1 and SRC2 ASC loop closed, we added an offset to the input of the SRC2 loop (which actuated only on SR2) and looked at how much the SRC1 loop moved the SRM to compensate for the cavity axis motion. We then added a SRC2 to SRM off diagonal driving term to reproduce this motion. The coefficient is -7.6 for pitch and 7.1 for yaw. We verified that with those new coefficients in the output matrix, an offset in the SRC2 loop does not move significantly the SRM.
This shoudl help us in increasing SRC1/SRC2 bandwidths, which are quite low right now.
In SRC2 filter banks we installed a compensator (HSTcomp, FM9) which has been designed to allow increasing the bandwidth up to few Hz. It's based on measurements of SR2 plant that we took in the straight beam configuration. In the same configuration we tested the loops and we could get a bandwidth of about 3 Hz in both pitch and yaw. Currently, we're not using such a high banwidth in DRMI configuration.
Attached is a screen shot of the SRC2 (AS_C to SR2) loops that we commissioned in single bounce this morning. We had to reduce the gain with DRMI locked, (perhaps because we had not done the diagonalization to SRM yet), but the loop shape should be the same.
Also, we have now offloaded all of the DRMI ASC loops to the top stage. The guardian engages all of these loops, slowly. (first BS, then INP1 and PRC2, the PRC1 (PRM), then SRC1, then SCR2) We have seen that this is repeatable for several different starting alignments. The guardian turns off INP1, PRC1, PRC2, SRC1, and SRC2 in the state CARM on TR. We think that LLO leaves the SRC loops on during the offset reduction, we want to see that the loops still seem reasonable in full lock before leaving them on.
We have manually requested the DRMI guardian to offload these offsets to the alignment sliders, this was rough and not very accurate but the lock survived.
Before starting this morning, we adjusted the dark offsets on the AS WFS and checked that the segments are wired correctly (I moved the beam onto one quadrant at a time and checked that all quadrants have about the same response)
