elenna.capote@LIGO.ORG - posted 22:14, Tuesday 07 February 2023 - last comment - 14:45, Thursday 09 February 2023(67280)
Returned to NLN, IY Ring heater off
Evan H, Elenna
We have been able to return to NLN with the IY ring heater now turned off.
- We engaged PRC2 P with a 50 ct offset by hand. Evan reduced the offset slowly as we powered up, and at high power we don't need any offset. This is not in the guardian
- we reverted the DHARD P and Y gains to their nominal settings
- We reverted all phasing and intrix changes we made when the ring heater was on, as well as carm changes
- we reverted the CO2 annular powers
- We commented out the ADS dof 4 and 5 10 dB gain boosts that turn on in engage_soft_loops, but then we turned them back on by hand. They are still commented out in the guardian
- Evan has noticed that the REFL A 45 and POP X phasing is still bad. These should be rephased.
- There is some 2.2 Hz buzzing in CSOFT P that I don't understand. Raising the gain makes it worse, lowering the gain slightly improves it?
- I dropped the CHARD Y gain to 150, and everything is fine so far. This is not set in the guardian.
Yay! Nice work!
Just for fun we tried cutting the CO2 X annular heating in half (4 W to 2 W measured on the power meter). The POP18 buildup improved, the frequency noise coupling at 200 Hz and 4 kHz got worse, and the DARM antispring did not improve. We reverted this change.
And to elaborate on what Elenna said about PRC2 P, which currently uses POP X as a sensor and feeds back to PR2, it seems more stable if engaged at low power with the 50 ct offset, but this is less necessary as we increase the power. I also used the unused DC7 P infrastructure to monitor an alternative PRC2 P error signal, REFL A 9I − REFL B 9I, which seemed to the same overall spectrum as POP X in pitch (but not in yaw) and had a favorable dc setpoint that did not require the offset. I suggest that if PRC2 P continues to be troublesome that somebody try this alternate error signal, to be installed with input matrix gains of +0.125 for REFL A 9I and −0.125 for REFL B 9I to match the POP X gain. I was not able to easily find an error signal for PRC2 Y that used REFL diodes, but currently REFL 45 is entirely unused in yaw. See the attached ASC matrices.
Judging by a frequency line injection, all the REFL 9 and 45 diodes could use a bit of rephasing, but REFL A 45 in particular is probably off by tens of degrees. Judging by a PRCL injection, POP X is off by 45 degrees and I cannot see a reason for not rephasing it (i.e., the current POP X Q looks just as clean as POP X I).
Since reducing CHARD Y gain to 150 has been successful, I ran some quick loop measurements (so not as precise as Craig's) and determined that we can probably reduce the gain to 125. That's still higher than "nominal", which would be 75 based on my previous loop design, but 3 dB less than what it has been. I think we should continue to measure this loop and see if we can reduce the gain further. Adding to the guardian.
EDIT
2 Hz buzzing is back. I am increasing CHARD Y back to 150. Removing 125 from guardian.
The last holdout change from the IY ring heater adventure was this line from the original alog above: "We commented out the ADS dof 4 and 5 10 dB gain boosts that turn on in engage_soft_loops, but then we turned them back on by hand. They are still commented out in the guardian"
I have just uncommented those lines from the guardian code in engage_soft_loops. When we relocked last night, I had turned on this 10 dB gain filter by hand.
We are also able to lock without the PRC2 P offset, although it does improve the buildups at 2W to engage the offset along with the loops.