Tonight we again spent some time on ASC, which was causing us some problems staying locked: SRC1, PRC2 (REFL WFFS), and CHARD.
- SRC1- changing the RF9 modulation depth definitely does change the sensing for SRM. I've moved the modulation depth reduction to happen before we switch from the dither loop to AS36. I also retuned the matrix we use to combine segments for AS 36AI again tonight. (Last time this was done was in alog 30759) I've changed the seg 3 element for pitch to -0.487, and for yaw it is -0.205. This seems to be working. Earlier tonight we had a lockloss that might have been due to the modulation depth change, or could have been something else like SRM ASC (or REFL WFS controlling PR3).
- PRC2- we measured both the pitch and yaw loops on POP X and after transitioning to RELF WFS, and found that both gains were too low once we transitioned. Yaw was 12dB too low, while pitch was only about 3dB too low. For this lock I had added the extra gains in the filter modules, but I edited the guardian so that they should go into the input matrix next lock. Screenshots of the measurements taken after the gain adjustment are attached. Once the gains were increased the sideband powers no longer fluctuate as much when we are controlling PR3 using REFL WFS.
- After about an hour, we had another example of the 0.5 Hz instability ringing up, which seemed to be stopped by turning off the cut off on CHARD P and increasing the gain from -0.14 to -0.2. I took a long measurement of CHARD P (with the higher gain and cut off off), which broke the lock as it finished (3rd screenshot). Based on this I modified Stefan's 25W filter to avoid having a ugf at around 0.7 Hz by moving the first set of poles down from 0.75 to 0.7Hz (screenshot attached of filter change)
This all seemed fine, but we just lost lock after trying to transition to the REFL WFS again. For tonight we will leave the gaurdian set up to skip REFL WFS again.
Other things:
This afternoon we had a long stretch of down time mostly spent doing a couple of inital alingments. One thing that didn't go well was locking the XARM in IR, which has been a long standing problem. After Corey got it to lock, I measured the loop and had a look at the filters. I made a new filter called servo and one called boost that I think should be able to replace all the filters that the guardian currently uses (boost should be triggered as the boosts currently are). Next time we do an inital alingment and there is a commissioner around, please try these out, they might make life easier. There is a template to measure the loop in /ligo/home/sheila.dwyer/LSC/XARM_IR
Corey and I were working on accepting things in SDF, and saw something strange that we should look into tomorow. In our lock acquisition that finished around 7:40 UTC on the 17th, in the REDUCE_RF9_MODULATION_DEPTH state, it seems like the guardian skipped the gain increase for ASC-REFL_A_RF9_I1, as shown in the attached screenshot. We manually entered it. Looking at the code, this is hard to explain since this segment is the ninth element in a list, and the gains before and after it in the list were all changed correctly. This didn't happen in last night's lock.
I edited this code today to round off the values that we use so SDF doen't get confused.
for pd in pd9MHzlist:
