Today locking was a bit of a struggle as the ring heaters cooled down. When we did acquire lock our buildups (particularly RF18) and LSC sensing matrix looked better than usual. The lock we had this afternoon was in a strange state - we had to reduce the PRCL gain, and the arm ASC was unhappy (and eventually caused a lockloss). I don't want to put our changes in guardian though since we're not sure how to reprocude this thermal state yet. Assuming the difference in our interferometer today vs previously is common ring heater, it seems like we might want to try operating with a little bit of common ring heater in the future.
Lock acquisition
After Austin did and initial alignment we were able to get to DC readout. We kept losing lock as we powered up to 5W, we noticed there was a ~130Hz ringup in the buildups and a peak in PRCL. It looekd like PRCL was gain peaking so we reduced the gain by half by LSC-PRC2_gain to 0.5. This allowed us to power up but hasn't been captured in the guardian.
Many violin modes were rung up, we had to sit at 10W and damp them for a bit. When we went to 50W Craig reduced the DARM offset so we didn't saturate the DCPDs. We rediscovered that ETMY mode1 needs a gain of +0.1 at 50W input power and -0.1 at 10W (or 2W) input power. ITMY mode6 is rung up and I don't think we have good settings to damp it right now.
Once we got to 50W we saw some angry ringing in the ASC, visible on the ASC FOM as well as the PR gain. We lowered the CHARD_Y from 25 to 20. This didn't hurt but we might need to do some more gain tweaking. Our eventual lockloss looked ASC-ish with a 0.44Hz CSOFT_P mode seen in the test mass oplevs.
LSC sensing matrix
Once we reached 50W input power we measured the LSC sensing matrix using Craig's script in labutils. The radar plot is attached and should be compared to the most recent 50W radar plot from April 13, also attached. Things look better - the magnitude of MICH and SRCL are stronger in POP (and this is with a factor of 2 gain reduction in MICH). MICH and SRCL are more orthogonal at POP45 compared to our previous measurement.
Thermalization comparison
I made a couple of ndscopes comparing the buildups during thermalization in today's lock vs the pre-ring heater lock on Friday. Flicking back and forth it looks like today:
- The PR gain is similar, maybe slightly worse (noisy in today's lock from the ASC)
- LSC refl is higher, going over 20 vs 14 in previous locks
- POPAIR_B RF18 thermalizes to around 60 - usually it's closer to 52
- POPAIR_B RF90 is similar
- The arm powers are lower, and start to decrease 20 mins into the thermalization where previously they just went up
- POP and REFL RF9 demods decrease rather than trending upwards (not sure how to reconcile this with the increase in popair rf18 going up?)
- POP RF45 demod is lower, REFL RF45 is very noisy
One thing I don't understand is, if I look at the buildups during Friday night's ring heater test I assumed I'd see POPAIR_B_RF18 and REFL_A increase and then decrease as we overshoot some ideal ring heater power. I don't see that at all - they seemed to drop as the RH was turned on.
Power up vid
For funsies I used Dan's gui to make a video of the ITMX HWS during today's powerup.
Ring heater test tonight
For the final lock tonight I had to lower the PRCL2 gain by hand (from 1 to 0.5). Craig reduced the DARM offset when the DCPDs started saturating at 50W input. To squash an ASC ringup I took the CHAR_Y gain from 25 to 15, the DHARD_Y gain from -60 to -50 (DHARD seemed more effective).
I've set a common ITM ring heater test for tonight. The ring heaters will go to 0.75 W per segment at 8pm (1334372395) for 8 hours. This script is runing on zotws18.
The ring heater test last night was a partial fail, we lost lock a few minutes after turning the ring heaters on (seems unrelated). Weirdly the interferometer reacquired lock on its own with the ITM ring heaters at 0.75W, presumably with the higher PRCL gain and ASC gains. Summarizing the attached screenshot, during the lock with ring heaters on:
- The PR gain was low, POP_LF was low
- REFL was also low (~12)
- POPAIR_B_RF18 was low (~50), RF90 was unganged
- Arm buildups were low
- POP_RF9_DEMOD and REFL_RF9_DEMOD were higher?
I don't understand what the thermal state of the interferometer was during our lock on Monday, I expected a bit of ring heater power would give similar buildups, maybe 0.75W was too much?
