Reports until 10:33, Sunday 26 March 2023
H1 ISC (TCS)
daniel.brown@LIGO.ORG - posted 10:33, Sunday 26 March 2023 - last comment - 16:27, Tuesday 18 April 2023(68165)
14.7 kHz parametric instability

Cao, Dan

During some ring heater changes last night two PI were the cause of 10 locklosses between 1363820600 and 1363877600. They both ring up quickly.

14759 Hz

14767 Hz

These seem to be from the X arm ring heater change. At the time ITMX was 0.44W and ETMX was between 0.6 - 0.8 W, dropping it back to 0.4W has stopped these PI ringing up.

We suspect these might be the second order second drumhead mode which Carl's document suggest is around 14700-14800 Hz.

Images attached to this report
Comments related to this report
victoriaa.xu@LIGO.ORG - 21:00, Tuesday 28 March 2023 (68230)

After Dan's ETMX ring heater change this evening (68215), we saw one of these PI's ring up briefly into the lock, from around 1364096099-1364096189, reaching max amplitude around 1364096149, see screenshot of the following PI-bandpass outputs.

Naoki and I have setup the PI damping infrastructure to target these modes:

  • Mode 15 = 14.768 kHz BP --> I/E-TMX Ch. 6
  • Mode 16 = 14.760 kHz BP --> I/E-TMX Ch. 7

We haven't successfully moved these modes yet and can't yet say whether they're I- or E- TMX modes. But, it seems our filters can catch these modes, so we can begin to work on damping.

Unlike the other forest of PI modes around 10.4kHz which are always there (which we can damp if needed 67966), these 14.76-7kHz modes are different in that they're at a very clean part of the spectrum with no existing lines otherwise, making them difficult to damp normally.

Images attached to this comment
victoriaa.xu@LIGO.ORG - 23:35, Tuesday 28 March 2023 (68233)

Both modes are ETMX (14760 and 14768)!

Can confirm Mode 14768 is on ETMX. I can excite it using PI-damping sent to ETMX, with e.g. H1:SUS-PI_PROC_COMPUTE_MODE15_DAMP_GAIN = 8000. Mode is not excited when driving itmx. The mode frequency doesn't appear that that stable (drifts by ~1Hz over seconds). And, the phase I use to drive the mode changes its frequency somehow? Are there multiple modes here maybe, and our PI ESD drive to LL/UR is somehow ringing both in strange ways?

For the ETM PI ESD damping, the models seem only setup to send ESD drives to LL/UR, unlike ITMs where we can send individual drives to all 4 (UL/LL/LR/UR).

Driven in diff LL/UR = 1/-1, Phase -180, Mode @14767 mag~1.5e-7

Driven in diff LL/UR = 1/-1, Phase +180, Mode @ 14769, mag~1.5e-7

Driven in common LL/UR = 1/1, Phase +180, Mode @ 14768, mag~6e-7

Driven in common LL/UR = 1/1, Phase 0,    Mode @ 14770, mag~7e-7

Driven in common LL/UR = 1/1, Phase -180, Mode @ 14768, mag~6e-7

------

Mode 14760 is on ETMX too. Similar to above, I can ring up this mode driving ETMX but not itmx, the same gain works, ie H1:SUS-PI_PROC_COMPUTE_MODE16_DAMP_GAIN = 8000. As above, the mode frequency drifts around a bit, this one by several Hz over minutes. And, driving in common LL/UR excites it to higher magnitudes for the same drive, and the mode frequency also seems to depend on the drive's PLL phasing.

Images attached to this comment
victoriaa.xu@LIGO.ORG - 23:44, Tuesday 28 March 2023 (68237)

Lockloss around 1364104380 as I was driving both modes. DCPD doesn't see either of these 14.7k mode heights increase as we lost lock, so I think it's still possible but also possibly unlikely these PI modes broke the lock as I was driving them.

Images attached to this comment
vladimir.bossilkov@LIGO.ORG - 20:33, Thursday 30 March 2023 (68303)

Hey Cao, I'm worried you might be misreading Carl's documentation. Carl in-fact tuned his simulations having seen glaring discrepancies, to frequencies that are observable and there should be nothing in the band 14.4 kHz to 15.02 kHz. It could straight up be testmass-to-testmass difference we're seeing - but in the 4 test masses at LLO, we have observed and been able to excite the "Drumhead O2v" mode in the 2 new ETMs (with the ITM ones being identified before my time), they indeed lie around 15020 Hz and can all be seen as thermal noise peaks in the OMC spectra. At LLO we have observed a very problematic 14.9 kHz PI which cannot be seen in the OMC spectrum until it makes use lose lock [aLog]. I implore you to consider it is a mode similar to what we are facing: and that it is ?likely? an aliased down mode. If you are in the unfortunate position where you know you are about to face this PI: please look at the 524kHz OMC output before down-sampling and see if you can identify whether or not it really is at some frequency ~50kHz, that would be really informative.

vladimir.bossilkov@LIGO.ORG - 20:19, Monday 10 April 2023 (68572)

Lateish follow-up: over the weekend our 14.9 kHz PI was identified at 80437 Hz (2 aliases away). So have a good far look when, and if, you get this PI again [LLO:64284].

victoriaa.xu@LIGO.ORG - 16:27, Tuesday 18 April 2023 (68803)

Even later follow-up: this 14.7 kHz is indeed an aliased-down version of the real 80.3 kHz PI on ETMX, see LHO:68760.