keita.kawabe@LIGO.ORG - posted 00:01, Tuesday 08 January 2019 - last comment - 05:13, Tuesday 08 January 2019(46278)
Locking tonight - moving spot positions, updated green references, 10 minutes at NLN
Craig Georgia Jenne Sheila
Edit: this is Georgia, accidentally posting as Keita!
- Moving spot on TRY (and TRX a bit)
- As Sheila mentioned in alog-46727, we noticed that the spot positions on the ASC transmission QPDs had drifted over time, particularly on TR_B_PIT on the Y arm. This is a potential culprit for our scatter-shelf from ~20-60 Hz. We tried adjusting the spot position on the ETM by adjusting the A2L gain at the L2 stage, while the spot position was controlled by the dither loop.
- We used a reference time from a 2W lock on December 8, before the scatter shelf was problematic, and adjusted the P2L and Y2L gains to move the spot position back towards the old position. We found the spot moved more on TR_B than TR_A.
- Meanwhile Jenne realised that over time the TMS, particularly TMSY, had drifted according the the witness sensors. We adjusted the alignment sliders on TMSY to bring it closer to its position on December 8.
- First attachment shows:
- First column: Y and X QPDs during a lock on December 8
- Second column: same QPDs during today's lock where we adjusted the A2L gains and the TMS
- Third column: The P2L and Y2L gains on ETMY and ETMX
- Fouth column: TMSY Pitch alignment, and PR gain
- In the end we chose a spot position which gave slightly better PR gain rather than trying to return to exactly the same spot.
- We got to nominal low noise (03:40:00UTC) and, at least qualitatively, the scatter shelf was still there. I tried to sit at NLN for a while so we could get a nice spectrogram, but we lost lock 10 mins later.
- I brought the A2L gains back to where they were.
- Updated Green references
- In the next lock, while we were locked at 2W with all the ASC loops on, Craig un-shuttered ALS and updated the QPD offsets using Kiwamu's script. We also updated the camera references though they were not very different.
- We noticed a perplexing 0.5 Hz glitch on the Y arm transmission, when the alignment was bad (second attachment).
- After some soft loops weirdness, (lockloss after increasing CSOFT and DSOFT gain in ENGAGE_SOFT_LOOPS) we finally got to NLN again only to lose lock after 10 minutes again! Attachment 3 is a screenshot of some ASC signal at the lockloss.
Images attached to this report
Comments related to this report
- Attached old and new green QPD offsets. Had to realign TMSY to relock the Y-arm after the last lockloss.
- Tried to measure CSOFT P loop using template in IscSVN/path/to/CSOFT, this killed the lock.
- Stepped through ENGAGE_SOFT_LOOPS by hand, while skipping over the turn off of FM1 for CSOFT P and DSOFT P, which is just a -20dB filter module. (this killed the lock last time with a slow ringup.)
-- Got to the end of the state with no problems, then slowly increased the gain of CSOFT P and DSOFT P by factors of 2 until I was able to engage FM1 for both with no problem. Unclear why this is an issue earlier in the state.
- We've lost lock three times in DHARD_WFS from large yawing as soon as the state kicks on. The last time I tried engaging DHARD with 5 times lower gain, still lost lock. Seems that the arms are yawed badly relative to one another.
-- Tried lowering DHARD P and Y to -1 gains from -10 and -15 respectively. Still lost lock... This is odd because it appeared that DHARD Y was turning back around and was gonna make it this time.
-- DARM appears to be running away around 10 seconds after the arms become better aligned.
-- Tried putting the A2L back back (i.e. to the gains Georgia found earlier). This did not help. These were
ETMX P2L ETMX Y2L ETMY P2L ETMY Y2L
Current Gains 2.65 2.7 3.4 1.95
Old Gains 2.65 3.3 2.6 1.0
- POP18 looks more ratty than usual, probably from the exceptionally high microseism (~0.7 um/s).
- Bumped LSC dithers by a factor of three based on info from Australia (thanks Dan). He claims that MICH gain may be going UP during thermalization, which is just crazy.
Leaving the IFO in DRMI.
Images attached to this comment
I had used the wrong OSC demod channel when looking at MICH previously, what I had seen was the MICH coupling into POP45_I (not Q) increasing by ~2. This had a low SNR at 2W input so we wanted to increase the line amplitudes a bit, but that isn't needed now so I set them back to how they previously were.
Plots attached is showing how the PRCL/SRCL/MICH excitations couple back into their respective DOFs for one of todays 30W locks. The large drop in PRCL can be seen, SRCL and MICH also have small decreases at 30W.
Images attached to this comment
