After the watchdogs were reset a lightening strike caused a power glitch. GC UPS reported going into and off of battery power.

h1susauxh2 and h1susauxex went offline, turned out they were rebooting themselves and came back after a few minutes.

PSL is in a bad way, Ibrahim is calling for support.

The lights in my house flickered at the time I saw the sus-aux machines go down. We have been having a storm roll over us for the past hour, moving from Oregon northwards.

SWWDs for ETMY, ITMY and ITMX have tripped.

HWWD for ITMY has just tripped, its RMS is significantly higher than the others.

Strangely ETMX SWWD and HWWD have not come close to tripping.

Lockloss due to an EQ that hit before EQ mode could activate. Seems to be either very local or very large - not on USGS yet.

This afternoon a 6.4 mag eq in Fiji provided another good opportunity to test the high asc gain eq mode. This time peak velocities were about 3.6 micron/s on peakmon, around 2 micron/s on the .03-.1hz blrms on the wall fom. We were out of observe for ~1hr, but we probably would have been down for at least 2hrs without the transition. Definitely one of the largest eqs we've ridden out, ever.

On Sept 4th we had a longer commissioning period to allow us to heat up OM2.  The main goal was to use this to characterize the mode matching of the arm cavities to the OMC, but we also made some other interesting measurements.

Summary:  heating up OM2 now costs us about 1% of optical gain, while it used to cost us 2%.  Heating up OM2 also changed the SRCL offset needed to get flat squeezing significantly, and reduced the amount of squeezing that we could get (without adjusting psams).

The interferometer unlocked (86727) while Jennie Wright was running the DARM offset step script with OM2 cold, to get a measurement of HAM6 throughput and look at OMC refl before heating up OM2.  As that link says, there wasn't an obvious connection between the DARM offset script which was in it's final steps (nearly back to normal) when the lockloss happened. 

We turned on the OM2 heater while relocking after some back and forth, then had an commissioning caused lockloss while trying to recover.  By the time we were relocked, OM2 was heated up according to thermistor 2, which is the thermistor who's timescale matched the timescale of optical gain changes in the past (see screenshot of June 2023 example).  We also had a large earthquake while relocking, so we paused after power up and before going to nominal low noise, so we do not have tracking of the optical gain using pcal in the first 30 minutes of this thermalization. 

The optical gain in this thermalization seemed to be fairly similar to the previous lock where OM2 was cold in the first hour.  The change in optical gain between OM2 hot and cold was much smaller this time around, so we needed to see the full thermalization in order to see what the gain change was.  We set the OM2 heater off 4 hours and 5 minutes after the power up, the attached screenshot shows a trend of optical gain during the previous thermalization while OM2 was cold, with a vertical cursor 4 hours 5 minutes into the lock.  The horizontal cursors show where the optical gain was at 4 hours and once the thermalization was complete, the optical gain continued to increase by 0.2% after the first 4 hours. The next screenshot shows vertical cursors also at the time of power up and 4 hours later, and the horizontal cursors are the same as on the previous screenshot (OM2 cold).   It seems that the optical gain was about 1.1% lower with OM2 hot than cold, although fitting these two thermalizations to an exponential might bring them closer by as much as 0.2%. Our current ring heater settings are 0W on ITMY, 1.5 W/segment on ETMY and 0.44W/segment on ITMX and 1W/segment on ETMX.

Since this is a different result than in the past times of OM2 changes, I've gone back to look at old times when we did this change.  One possible explanation for the difference could be ring heater settings being different. 

• 12/18/2023, cooled off while unlocked, optical gain increased 1.8% ETMX +ETMY RH 1W, ITMX 0.44W
• 9/19/2023, heated while unlocked, optical gain dropped 1.9%   (this time is confusing, it seems like the thermistors which had been disconnected were plugged in and looked to be thermalizing, but the set power was unchanged).
• 7/19/2023, heated in steps, 71484 ifo unlocked and relocked several times, perhaps calibration was reset making optical gain comparison confusing. (it did seem to drop by about 2%, but this seems more like a calibration reset) ETMX + ETMY RH 1W/segment, ITMX RH 0.44W/segment
• 7/12/2023 OM2 cooled off in lock, lockloss happened in the early part of the cooling (2% optical gain increase), ETMX + ETMY RH 1W/segment, ITMX RH 0.44W/segment
• 6/27/2023 OM2 heated while in lock, optical gain dropped by 2%, ETMX + ETMY RH 1W, ITMX RH 0.44W/segment, 0 on ITMY.

Once the OM2 heater was turned back on, the optical gain increased by nearly 1%, but the IFO lost lock before that thermalization finished.  There was a 2% decrease in POP18 during the cool off and a 3 urad shift in SRM top mass. 

While OM2 was hot, we did a few tests.

• We had a large change in the SRCL offset needed to reduce the squeezing rotation: 86739  This is interesting, in the past we have tried to change SRM alignment to see an impact on this number and it was a much smaller change.
• No squeezing time for a senstitivty comparison (not much change): 86788
• quick squeezer data set (before SRCL offset updated: 86736 with SRCL offset updated: 86739 ) , to see a comparison with OM2 cold see Camilla's plot from this morning
• Jennie Wright did DARM offset steps with OM2 hot, and cold.

TITLE: 09/08 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Earthquake
INCOMING OPERATOR: Ibrahim
SHIFT SUMMARY: We stayed locked the whole shift, 14 hours. We dropped obs to ride though a 6.4, we're still waiting to leave EQ mode.
LOG:                                                                                                                                                                     

• 15:20 UTC Superevent S250908y
• 15:43 - 18:32 UTC Commissioning
• 22:12 UTC Earthquake mode from a 6.4 from Vanuatu
• 22:28 UTC We dropped Observing after I pressed the ASC_hi_gain button
  • I transitioned back at 23:30 UTC
  • 23:31 UTC Observing

TITLE: 09/08 Eve Shift: 2330-0500 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Earthquake
OUTGOING OPERATOR: Ryan C
CURRENT ENVIRONMENT:
    SEI_ENV state: EARTHQUAKE
    Wind: 14mph Gusts, 8mph 3min avg
    Primary useism: 0.42 μm/s
    Secondary useism: 0.09 μm/s
QUICK SUMMARY:

IFO is in NLN as of 10:30 UTC (14 hr lock)

We are temporarily out of absorbing as we try to ride out a 6.4 EQ using ASC High Gain (new feature), alog 85740

After SEISMIC goes back to CALM, we'll turn off the high gain and go back to OBSERVING

Link to report here.

• Average observing time: 71.2%
• Range was generally between 145-160 Mpc throughout the week, except for a drop at 1100-1200 on Wednesday Sept 3 when it was around 135 Mpc, mostly likely due to SQZ issues
• EQs and anthropogenic ground motion may contribute to increased glitch rates and noise in h(t) several times throughout the week
• Tuesday had quite strong violin modes (esp. 500 Hz) after relocking
• Magnetic corner station plots for Thursday Sept 4 are blank
• Common Hveto channels throughout the week: H1:ASC-Y_TR_A_NSUM_OUT_DQ, H1:ASC-Y_TR_B_NSUM_OUT_DQ, H1:ASC-Y_TR_B_YAW_OUT_DQ, H1:ASC-Y_TR_A_PIT_OUT_DQ, H1:ASC-Y_TR_A_YAW_OUT_DQ, H1:ASC-CSOFT_Y_OUT_DQ, H1:ASC-DSOFT_Y_OUT_DQ
• Fscan-Observing lines stayed around 2700-3000 throughout the week

In the past, when heating up the OM2 heater, we have noticed that the low frequency sensitivity improves, for example alog 78652.

I wanted to make this comparison again, but we have seen that the squeezing performance can be variable, so I first chose to compare a no squeezing time with OM2 hot and cold.

It's also important to note that the calibration has likely changed with the hot OM2. Some of that can be corrected with the TDCFs. However, we saw that the required SRCL offset to minimize SRC detuning with OM2 hot was significantly different, alog 86737. This means that the sensing function is probably significantly different, causing some change in the calibration that is not compensated. We didn't take any calibration measurements with hot OM2, so we have no idea how much or how to compensate the calibration change.

With OM2 hot, the coherence of DARM with SRCL and MICH increased, although SRCL can explain most of the excess noise between the two times. Interestingly, the coherence of DARM with jitter decreased with OM2 hot. There is no significant change in the coherence with PRCL.

Here is a comparison of the cold and hot OM2 times, using GDS CALIB STRAIN CLEAN.

I performed a simple coherence subtraction of SRCL alone, and SRCL and MICH together. I checked that there is no significant coherence between SRCL and MICH.

Given the possible issue with the calibration, I am not certain how comparable the low frequency noise is. There is a small increase in the noise up to 30 Hz with OM2 hot, even with the above subtractions applied.

H1 ISI CPS Sensor Noise Spectra Check - Weekly Famis 26655

BSC high freq noise is elevated for these sensor(s)!!!
    
ITMY_ST1_CPSINF_H3
 

In 86589, Jeff points out mysterious noise seen in SRC2 Y CTRL between 1 and 3 Hz that is only seen when the SR3 estimator is fully on with the L to P contribution included.

To verify this, I plotted some other times with the estimators ON and OFF, and looking at those different times, I was able to confirm that the unknown increase in noise Jeff points out between 1 and 3 Hz is NOT caused by the estimator, but by some other noise source in passing.

Camilla, Sheila

We did a series of quick measurements related to filter cavity backscatter, a follow up with plots will come soon.

Injection for linear coupling (varies from day to day, we are repeating to see how much it changes, 1st screenshot):

• /ligo/home/sheila.dwyer/Noise_Budget_repos/excesspowerprojections/data_files/FC_backscatter.xml
• refs 32-34

Fringe wrapping (scattered amplitude seems slightly lower than in 2024 when moving ZM2, slightly higher when moving ZM5):

• userapps/sqz/h1/Templates/dtt/DARM_scatter_2025_0903.xml
• quiet refs: 33-37
• ZM2 test L excitation 100 counts at 1 Hz, ref 38-40
• ZM5 test L excitation 100 counts at 1 Hz, ref 41-43

Open loop gain (also varies from day to day):

• ref 12 Sept 3rd
• ref 18 (8% higher gain than on the 3rd)

FC2 M1/M3 cross over:

• userapps/sqz/h1/Templates/dtt/filter_cavity/FC2_M1_M3_crossover.xml
• cross over looks to be 1 Hz, with an instability at 1.3 Hz and less than 3dB of gain margin
• while doing this I moved the location of a filter in FC2, which had to be accepted in SDF and updated in the guardian.  This filter is set in the init state of the guardian, which should probably be moved to down instead of init.  I also copied over a boost filter, but after doing the measurements it is clear we can't use that boost as is, so it is off.

We tuned FIS around 100HZ and then changed the SQZ angle +/-5deg, plot attached. Shiela thinks this could help us constrain thermal noise in the future.

Plot attached of today's data (SRCL offset at nominal -382) compared to last weeks while OM2 was hot (with SRCL offset tuned at -235). Solid lines are cold OM2, dashed lines are hot OM2. FIS is better with cold OM2 as expected.

We noticed that mean SQZ was different <100Hz with the OM2 hot vs Cold. This persisted if I changed the averages so wasn't caused by a glitch and also seemed to be increased looking at Sheila's 86736 data too. Changed SRCL offset to see if this was the cause but it didn't appear to be. Plot attached, so the cause is OM2.

Mon Sep 08 10:07:09 2025 INFO: Fill completed in 7min 5secs

Gerardo confirmed a good fill curbside. TC-A became good again at 05:56 this morning and looks nominal in today's fill.

Gerardo inspected the TCs and suspects TC-A wiring had been moved by animal activity. Why it corrected itself this morning is a mystery.

J. Kissel

After today's H1 SUS SR3 Pitch Estimator's inclusion of the Sus. Point L to M1 P feed forward from the HAM5 ISI GS13s (where we had mistakenly only installed Sus. Point P to M1 P) -- see LHO:86567, I now compare the times of ASC signals (using 0.02 Hz binwidth, 64 sec FFT chunks, 30 averages, and a Hanning window with 50% overlap):
    2025-08-20 18:29 UTC - Both P and Y Estimators are OFF 
    2025-08-24 18:18 UTC - Both P and Y Estimators are ON, but for the P estimator, only the Sus. Point P to M1 P contribution is included in the GS13 FF
    2025-08-26 22:00 UTC - Both P and Y Estimators are ON, and for the P estimator, both P to P and L to P contributions are included in the GS13 FF.

I'm showing only the control signals, and only the ASC DOFs which are impacted by SR3: DHARD, MICH, SRC1, and SRC2; both pitch and yaw.

In All DOFs, the 0.64 and 0.75 Hz modes that had been made worse with only the P to P model of suspension point contribution have now been restored to no-estimator levels.
In some DOFs, the 1-10 Hz broadband motion is just barely improved, but enforces the conclusion that SR3 is only partially, if not 'not the dominate contribution' to the ASC signals. 
Here's a fun one for you -- look at SRC2 Y CTRL -- and look at how much the SRC2 *yaw* motion has changed from including the *longitudinal* suspension point contribution to M1 *pitch* top mass. #ThrowsHandsInAir

But -- overall -- with the improvements and all design intent included -- the SR3 P and Y estimators slightly improve the ASC noise from 1 to 10 Hz. Great!

As Oli notes in LHO:86551 comparing all of these different configurations from totally different days and times is dubious. We'll get "official" versions of all of these ON vs. OFF configurations on Thursday 8/28.

I also attach the local estimator metrics for pitch described in LHO:86553, comparing "only Sus Point P to M1 P modeled contribution" against "P to P and L to P sus point to M1 contribution." That similarly shows that L to P contribution forms a good fraction of the signal needed for damping.