We are clipping on IM4 trans. I trended the position of IMs 1-3 over the past few months and it appears they moved significantly from their position before and after the vent. The t1 marker on these plots marks the approximate time that the positions changed which corresponds to the end of O4a and the start of the vent.

As a note, this means our PRG channel is untrustworthy right now.

TITLE: 03/11 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Planned Engineering
OUTGOING OPERATOR: None
CURRENT ENVIRONMENT:
    SEI_ENV state: USEISM
    Wind: 14mph Gusts, 12mph 5min avg
    Primary useism: 0.04 μm/s
    Secondary useism: 0.73 μm/s
QUICK SUMMARY:

• Currently running through an initial alignment
• High winds seem to be calming down
• Microseism is elevated

Last week I scanned the SHG temperature for more data points to confirm the dip in the middle of the sinc function. A rough fit to the side lobes suggests we should have about 160mW of green output. A few days before we improved the PMC alignment we measured 230mW of 1064 input into the SHG. 160 mW would be 69.5% conversion efficiency. This number isn't too far off from the conversion efficiency measurement we took when we first installed the SHG 6 years ago. We have known for sometime that we are missing a lot of green power for the amount of 1064 we put in. This plot makes the picture a bit more clear that we have been indeed losing more than half the green power. The cause is unclear. We have been suspecting the gray tracking in the SHG cryatal to be the cause even though we have a low green finess in the SHG. 

The equation I used to create the fit was A*(np.sinc((T-T0)/width))**2 

Where A is the input 1064 power, T0 is 34.9 C, and the 'width' of the sinc function is arbituary set to 2.5. 

Sun Mar 10 10:11:07 2024 INFO: Fill completed in 11min 3secs

We reran the DARM offset step test around GPS 1394062645, similar to alogs 71913, 68870, 64974.

Our current measured contrast defect is slightly higher than in alog 71913, but overall pretty similar.
Matt will comment with the calibration of X0 offset cts into picometers.


Contrast Defect: 2.1 mW
Nominal Total Amps from DCPDs: 40 mA
Responsivity = e λ / c h = 0.858 A/W
Nominal Total Power on DCPDs: 46.6 mW
Nominal Homodyne Angle: 12.2 degrees

Using alogs 47217 and 45734 as a guide, we measured/calculated a new OMC-DCPD_MATRIX.

Note: this might have to be redone if we update the anti-TIA filtes for the DCPDs, see alog 76228.

We measured the coherent signal amplitude ratio g = PD_A/PD_B above 30Hz (i.e. above the TIA transfer function imbalance), so the matrix is optimized where we actually care.

g=1.0471

We measure the shot noise signal amplitude ratio h = PD_A/PD_B above 400Hz to avoid thermal noise and correlations from the feedback.

h=1.0247

This produced the new output matrix (see secript)

0.99861 1.00139
0.97725 -1.02328

For reference, old matrix:

0.99963 1.00040
0.98198 -1.0184

The new matrix was loaded, but not yet tested - the IFO lost lock due to 43kts wind gust.

Georgia, Trent, Elenna

We lost lock on the interferometer (most likely from 50mph gusts of wind) and had to align the arms. The y-arm proved difficult and we could not get the flashes on ALS-C_TRY_A_LF_OUT_DQ above 0.6 counts. To help the arm along in locking we changed the pdh auto-locker threshold to 0.55. We then decided to run an initial alignment.

During the initial alignment we got red dust warnings for the corner station.

After the intial alignment, the y-arm green flashes reaches 0.73 and we didn't need to change the pdh auto-locker again.

Today we saw in CAL-DELTAL_EXTERNAL_DQ some kinda weird hump in the DARM spectrum around 100 Hz.  
It made us curious about the calibration, so we measured the DARM OLG and sensing functions (PDF one).

1. The nominal DARM open loop gain unity gain frequency is around 80 Hz.  The loop suppression has a factor of 3 gain around 100 Hz.  This is higher than I recall, seems like we're over the phase bubble for sure.  
We lowered LSC-DARM1_GAIN from 400 to 285 to make DARM sit at the top of its phase bubble at 60 Hz.  This affects the way that CAL-DELTAL_EXTERNAL_DQ looks, which is expected, but is overall not impacting our GDS sensitivity significantly (PNG one and two).

2. The DARM sensing function seems to exhibit significant detuning, with an anti-spring kicking in lower than 12 Hz. 

3. We have 10^10 W/m DARM sensitivity.  That's super high compared to one year ago, increased by the increased amount of OMC DCPD light.  Seems to indicate the current OMC is pretty healthy.

Meashured the transfer function between DCPD_A and DCPD_B usinng a broadband injection. There is a slight frequency-dependent disagreement (0.2dB) between the two diodes, right around 25Hz, where the TIA has its complex poles.

It looks like  one or both Anti-TIA filters are slightly mismateched. We won't tweak it today because we don't know with one is the culprit.

Sat Mar 09 10:12:00 2024 INFO: Fill completed in 11min 56secs

Naoki, Vicky, Nutsinee, Matt

We recovered the FDS and got 4.5dB squeezing in IFO with 40 times more CLF power as shown in the attached figure. 

First we restored the ZM1/2/3 and FC1/2 OSEM positions to the values in O4a. Then we found the FC green trans in camera and PD. We aligned FC1 and FC2 and successfully locked the FC green. We went to SQZT8 and centered the green camera and green QPD.

Since we aligned the green QPD, the green QPD offset value is not valid anymore so we removed the beam spot control from FC ASC. We made a flag for the beam spot control in sqzparams and it is set to False now. We also set the ADF servo flag to False since the ADF demod phase might not be correct. After we figure out the optimal green QPD offset and ADF demod phase, we should revert them.

We reduced the FC IR gain from -3.5 to -0.1 and reduced the FC ASC gain from 0.1 to 0.005.

We have not done any PSAMS scan so we will do it next week.

Since we have a new OMC, we need to find the best optical gain spot on the new OMC.

To that end I closed the OMC ASC dither loops iwth a simple diagonal matrix (OM3-->OM3 and OM1-->OMC_SUS). It did converge with very low gain, but to a spot with not much headroom on the OMC_SUS (we are at 115000ct out of 131000cts on T2 and T3).

In that setting I reset the H1:ASC-OMC_[A/B]_[PIT/YAW]_OFFSET offsets, so right now we can lock to that spot with either dither or QPD.

Next: hunt for best optical gain around this spot.

I'm having a hard time deciding if the ITMY violin mode 5/6 damping settings are decent. Stefan has been moving around the OMC alignment which seems to have some large effect on the monitor filters. However, the alignment has converged at some value and I have engaged a 0.01 gain with FMs 6, 7 and 10 and so far the monitors show both modes decreasing. The nominal gain is 0.02, so I think if we lose lock and relock that gain setting will be ok. We are stuck in the OMC_WHITENING state and this mode pair is the only problem- all other modes are low or are decreasing very well in their settings.

Edit: nominal low noise achieved. These damping settings are good.

The pressures:
HAM7: ~1.9E-7 Torr
HAM8: ~4.7E-7 Torr
Corner: ~4.4E-8 Torr
EX: ~4.9E-9 Torr

Today's activities:

- HAM8 Turbo was stopped, the cart was disconnected
- Relay tube - HAM7 - HAM8 was valved in to the main volume - RV1; FCV2; FCV3; FCV4; FCV8 are now opened, so the system is back to normal

As the vacuum works now ended, this was the last page of the vent vacuum diary.

Congratulations to the vacuum team, and to all who helped in the vacuum tasks, it was a great effort, and it ended up being a success.
If anyone is interested in the vacuum results and tasks more in-depth, here is a summary document: T2400085-v1

We've had a large crew in the control room today working on locking:

• Oli ran initial alignment, manual BS adjustment was needed for ALS.
• I added back some of our DRMI ASC in the guardian now that we've recovered IFO alignment, I added back everything except PRC1, we'll see if this works and check the set point for the POP QPD.
• OMC whitening gain was 45dB, it should be 0dB, Stefan reset the dark offsets and accepted in SDF.
• I set the weights in the IDLE_ALS and SHUTTER_ALS edges so that we will again shutter ALS.
• There was a large 1.5Hz (ish) motion due to ETMY ISI, Jim says this was a bad L4C which he has now taken out of the loop, separate alog.  summary page
• While Jim was working on the ISI, Jennie and I checked the OMC ASC step responses (76208)
• Elenna checked that the POP offset she set yesterday (76190) in full lock look good for DRMI, so she added it PRC1 back to the DRMI ASC and now our DRMI ASC should be back to normal.
• We lost lock at RF DARM transition a couple of times, we changed the RF DARM normalization back to the X arm (It was changed to Y arm earlier this week) and increased the wait time between transitioning the input matrix and changing the filters to 2 seconds (we previously had a 1 second wait for a 2 second ramp). 
• We had two locklosses trying to transition to DC readout, this was because I reverted the change to the RF DARM normalization, but Jenne reminded us that we also had to make a change in DC_READOUT_TRANSITION for this change.
• We've put SDF revert back in the normal path of ISC_LOCK guardian.
• We had to rerun initial alignment, then checked the DARM UGF while powering up.  We have done most of the steps to nominal low noise except for injecting squeezing and OMC whitening. 
• Jennie Wright is moving the OMC ASC offsets to increase kappa c.

Louis, Vicki, Craig

Today we revived the correlated noise budget code located at https://git.ligo.org/aligo_commissioning/correlated_noise, first made in June (alog 71333).
This code is a standalone singular script which grabs the raw DCPD data, grabs the DARM calibration OLG and sensing function info, does some math from Kiwamu's DCC T1700131, and produces a correlated noisebudget.

Formerly, I had used my own noise_recorder white noise injections to get the DARM calibration information.  But since I am gone, it would be better if we used a maintained piece of software to get the calibration.
Today, Louis helped me to use a pydarm .npz file to get the DARM OLG and sensing functions.  
The code we developed today is in /ligo/gitcommon/correlated_noise/code/plot_cross_correlation_pydarm_npz_calibration.py

Attached are the correlated noisebudget results for GPS start 1387130434, over 600 seconds of data.  
I posit that we don't really need a ton of time to integrate down, both because the shot noise floor is not that far away from the correlated noise floor, and we can use logbinning averaging to achieve a higher effect number of averages in a shorter amount of time.

Results
I've used the Dec 18 noisebudget traces made by Camilla in /ligo/gitcommon/NoiseBudget/aligoNB/out/H1/lho_darm_noisebudget/lho_darm_noisebudget.hdf5.
Overall, the correlated noise results seem decently sensible, especially at high frequency.
I note that the correlated noise limit we are hitting at 1 kHz seems to be laser frequency noise (third plot attached).

At 500 Hz, we are running into our "thermal noise floor".  
Whether or not this is really thermal noise is anyone's guess, since it is above our official estimate, but the story for O4a remains the same as from O3.

At 20 Hz, the correlated noise seems to overestimate what is actually in DARM, which cannot be correct.  
This is something Sheila has been concerned with in the past (alog 70978).  
This could be chalked up to a small change in the actual vs pydarm DARM OLG, unclear what else it could possibly be.

It is worth noting that the DARM OLG cannot really affect the correlated noise results massively above 500 Hz, just because the magnitude of the DARM OLG is around 0.1 there and falling quickly.  

Future Work
The next steps is for Vicki to import this code to the aligoNB, removing all calls to my nds2utils library (avoiding a dependency there).
Louis also wants to add some functionality to aligoNB so that the noisebudget can call pydarm at a GPS time and get the most recent calibration.  (Craig loves this).
Louis will clone the aligoNB env to verify everything still works before making a "pull request" to the actual aligoNB repo.

This log follows up some previous work that's been done to understand the LSC, in particular MICH, noise coupling that we have observed. The most recent work done on this front was by Dana, in alogs 74477 and 74787.  Notably, the MICH coupling follows an interesting shape here, especially below 20 Hz. For MICH, we expect a flat coupling depending only on the finesse, Gm = pi/2*F. Evan Hall made some great measurements of this coupling years ago and you can see the results in Fig 2.17 of his thesis(link to thesis)- without any feedforward the magnitude of the coupling is flat in frequency down to 10 Hz. However, Dana's measurements clearly show a frequency dependent coupling below 20 Hz. Also, Dana mentions that the MICH coupling is about 40% too high than we expect based on finesse.

I'll add here a quick note that currently the UGF of MICH is about 8 Hz and the UGF of SRCL is about 11 Hz, so I think it's safe to assume that we are not seeing any significant closed loop effects from either of these loops down to about 12 Hz in these measurements.

Something to remember here is that when we measure the MICH coupling we drive the beamsplitter which does induce a change in the MICH length, but it also changes the SRCL and PRCL length. Therefore, we should expect to pick up SRCL coupling along with the MICH coupling when we measure. Evan's thesis also includes the form of the SRCL coupling in Eq 2.29: Gs = 0.012 [m/m] * Pa/750 [kW] * dL/10 [pm] * F/450 * (10 [Hz] / f)^2 (I am only including the first term that dominates at low frequency). The SRCL coupling results from radiation pressure in the SRC due to the DARM offset (dL), power in the arms (Pa), and finesse (F) and follows a 1/f^2 slope (again, I am ignoring the second term here that rises like f^2 at high frequency). We are operating at much higher power and double the DARM offset, Pa=370 kW and dL = 20 pm (and F = 440).

Regarding the 40% excess MICH coupling, if you divide Dana's result by an additional sqrt(2), it lines up exactly how you would expect. Is there a chance the MICH calibration is missing a rt2 factor related to the fact that we measure from the beamsplitter?

Assuming I'm correct about the sqrt(2) factor, and plotting the magnitude of the expected MICH and SRCL coupling along with Dana's calibrated measurement, our expected O4 SRCL coupling lines up almost exactly with the response at 20 Hz and below (see first attached plot).

I think this effect has always been there, but not easily observable because the SRCL coupling was much lower due to lower operating power and smaller DARM offset. Comparing Evan's MICH coupling measurement (Fig 2.17) with his SRCL coupling measurement (Fig 2.18), the SRCL coupling at 20 Hz was close to an order of magnitude lower than the MICH coupling.

I think this understanding of the coupling better explains why we need to put so much work into the LSC feedforward. Specifically, we measure the coupling for MICH and SRCL, and then usually tune MICH first and then SRCL. Then, to do our iterative tuning, we redo the MICH coupling. It's likely we could do a better job first just tuning SRCL and then measuring and tuning MICH, since we need to subtract the SRCL coupling well enough to get a "true" measure of the MICH coupling.

Furthermore, the SRCL coupling is dependent on our DARM offset. When we make changes to OM2, for example, we change the mode matching at the output. Since we servo to keep the amount of light on the DCPDs constant, we could be changing the required length to achieve this amount of light slightly. In principle, this should only effect the SRCL coupling, but since SRCL shows up in our MICH coupling measurement, it effects both.

Finally, it appears in Dana's measurement that there is a rising trend above 30 Hz, which I have no explanation for. I don't think that can be explained by the SRCL coupling. A look at that second term in the SRCL coupling equation: Gs = 3e-5 [m/m] * phi_s/10 [deg] * dL/10 [pm] * F/450 * (f / 100 [Hz])^2 (Evan's thesis Eq 2.29), I estimate that with a SRCL detuning of about 1 deg, around 30 Hz the SRCL coupling term with an f^2 slope is about 5.3e-7 m/m. This is both too small and the wrong slope to explain that feature.