I've made a script that uses a la mode to double check the mode matching that Jason and Ryan are working on using Jammt , I'm posting it here so that people can use it in the future if they want to.  It is saved in sheila.dwyer/PSL/modematching  along with a copy of the a la mode code (alm).

I took the beam profile measurements that they saved in /ligo/gitcommon/labutils/beam_scans/Data/NPRO_a_21Nov2024.txt and used them for fitting.   In this text file the distances are a rail location, I used the information in 80895 and compared it to what I got by fitting the data in NPRO_26Oct2024.txt to find the offset between their rail distances and their coordinate system. 

From their data I get a horizontal waist at 162.1um at -0.097 meters, and a vertical waist of 182.1 um at -0.176 meters (see first attachment).  This is more astigmatism than the laser 1661 had.  If no mode matching had been done at all with this NPRO swap we would end up with an overlap of 87% to amp1 (sqrt(X overlap* y overlap)).

I used the average of the vertical and horizontal waist locations, and asked a la mode to optimize lens 2 + 3 for that average waist (as Jason and Ryan were only planning to move lens 2 + 3 to avoid changing the waist through the EOM).  Then using that lens solution to propagate the fitted x and y waists, the overlap given by sqrt(Xoverlap * Y overlap) for that solution is 95% for lens locations of:

    label    z (m)     type    parameters         
    -----    -----     ----    ----------         
    L1       0.1020    lens    focalLength: 0.2220
    L2       1.1511    lens    focalLength: -0.334
    L3       1.2787    lens    focalLength: 0.2220

This is in pretty good agrement with the photo that Jason sent me of their Jammt solution. I also plugged in their solution, and get an overlap of 94%:
 
    label    z (m)     type    parameters         
    -----    -----     ----    ----------         
    L1       0.1020    lens    focalLength: 0.2220
    L2       1.1440    lens    focalLength: -0.334
    L3       1.2780    lens    focalLength: 0.2220

Using the lockloss page, I picked out some times when the tag "IMC" or "FSS_OSSCILATION" was triggered. Then, I just made some simple time series comparison plots between several PSL channels and PSL PEM channels. In particular, the PSL channels I used were:

H1:PSL-FSS_FAST_MON_OUT_DQ, H1:PSL-FSS_TPD_DC_OUT_DQ, H1:IMC-MC2_TRANS_SUM_IN1_DQ, H1:PSL-PWR_NPRO_OUT_DQ, H1:PSL-PMC_MIXER_OUT_DQ, H1:PSL-PMC_HV_MON_OUT_DQ. I compared those to the following PSL PEM channels: 

H1:PEM-CS_ACC_PSL_PERISCOPE_X_DQ, H1:PEM-CS_ACC_PSL_PERISCOPE_Y_DQ, H1:PEM-CS_ACC_PSL_TABLE1_X_DQ, H1:PEM-CS_ACC_PSL_TABLE1_Y_DQ, H1:PEM-CS_ACC_PSL_TABLE1_Z_DQ, H1:PEM-CS_ACC_PSL_TABLE2_Z_DQ, H1:PEM-CS_MIC_PSL_CENTER_DQ. For this analysis, I've only looked at 3 time periods but will do more. Below are some things I've seen so far:

1. I definitely do see correlations between the IMC/FSS (IMC_MC2_TRANS & FSS_FAST_MON)  channels and the PEM PSL sensors. In particular, it seems like the most consistent correlations are with the PSL_PERISCOPE_X/Y and PSL_TABLE1_X/Y/Z channels. In some cases, there was a visible glitch in PSL_TABLE2_Z, but the amplitude is much less than in the other PEM channels.
2. Some PEM glitches correlate with the glitching in PMC_HV_MON, but this was only the case for 2 of the 3 times I looked at.
3. Some PEM glitches correlate with PMC_MIXER, but I wonder if that is just a coincidence (see slide 9 for an example)
4. I have seen no obvious correlation with the PEM PSL sensor glitches and the NPRO_PWR channel.

From my very limited dataset so far, it seems like the most interesting area is where the PSL_PERISCOPE_X/Y and PSL_TABLE1_X/Y/Z channels are located.

The stem wall pour is completed, forms are stripped and it has been entirely back-filled. Weather has prevented the pour of the slab this week so Jake has shifted his attention to erecting the metal framework of the building. The forecast is not favorable for a slab pour tomorrow, so its likely to take place early next week.

T. Guidry

[Debasmita, Anamaria]

We wanted to check if there is any microseismic coupling at LHO, similar to that of LLO. So far, for LLO our understanding is that the microseismic ground motion (0.1-0.5 Hz) is being amplified by a potential scatterer which has a transfer function as shown in the first image. This transfer function was modeled based on some other information and seems to explain the LLO noise quiet well (LLO aLOG 73845). 

For LHO, we amplified the ground motion by the same transfer function and simulated the scatter shelf. If we assume that the scattered light amplitude is same as LLO (4e-10), given the amount of microseismic ground motion at LHO the scatter shelf falls below the current DARM spectra of LHO (second image). Even if we double the light amplitude, the scatter shelf does not show up above the present noise background (third image) but it comes quiet close to the noise floor.

But, if we double the microseismic ground motion and then further amplify it by the model transfer function, the scatter shelf shows up above the present DARM spectra (fourth image). This shows that, the same source which is producing the microseismic scattering at LLO can produce similar noise at LHO, but it will only be visible if the IFO is locked during a somewhat high microseismic motion. Even if the amount of light is half of what it is at LLO, the scatter shelf would still be visible if the ground motion is higher (fifth image). 

I used the ground motion as it was during 2023-12-06 11:00:00 UTC at LHO. This is one of the high microseismic days of LHO, the median of rms ground motion in 0.1-0.3 Hz was 689.66 nm/s and in 0.3-1.0 Hz was 95.02 nm/s. 

In the last image, I plot all the simulated traces together for a better visualization of how the light amplitude and amount of motion can modify the scatter shelf.  

WP12214

Marc, Ryan C, Dave:

h1susex has been fenced from Dolphin and powered down in preparation to re-establishing the LIGO DAC 28ao32 as the driver of the ETMX L1/L2 and ESD channels, essentially undoing what was done on Tuesday.

Fil is borrowing one of the PEM adc channels I was borrow for the Guralp 3T huddle testing. He unplugged one of the horizontal readbacks from the Guralp on ADC4, leaving the other 2 dofs plugged in. Channel he connected to is ADC 4 28, H1:PEM-CS_ADC_4_28_2K_OUT_DQ.

Thu Nov 21 10:11:56 2024 INFO: Fill completed in 11min 53secs

FAMIS26472

All trends look relatively stable, but in the last week EX is starting to see some small control output and a minor trend downward in pressure. Definitely worth keeping an eye on this week.

The PSL NPRO swap has begun per work permit 12210 and I have changed the OPS mode to CORRECTIVE MAINTENANCE.

TITLE: 11/21 Day Shift: 1530-0030 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Microseism
OUTGOING OPERATOR: Oli
CURRENT ENVIRONMENT:
    SEI_ENV state: USEISM
    Wind: 3mph Gusts, 1mph 3min avg
    Primary useism: 0.03 μm/s
    Secondary useism: 0.57 μm/s
QUICK SUMMARY:

• The PMC and FSS are locked as I arrive
  • Running the PSL_weekly script I see the PMC stayed locked all night 14.5 hours whereas the FSS has only been locked for 40 minutes
• 2ndary microseism continutes to decrease, its just getting under the 90th percentile
• Low wind
• More NPRO/PSL work to be done today, potentially a laser swap

The attached plots show the PMC mixer error signal and the ISS PDB readout. The later implements a fair amount of whitening so tends to fluctuate at a slow rate compared to the glitches.

The PMC error signal goes as high as 1.25Vpp during a PDH sweep. The channel recorded by the DAQ has a lot of gain and clips around +/-80mV, but it is calibrated correctly. So, the calibration is 1.25V/1.19MHz = 1.05V/MHz.

First plot shows a train of glitches, going as high as ~70mVpk (or ~70kHz).

The second plot shows a zoomed in version of a glitch of about the same size. The PMC servo drives the PZT with a 3.3K series resistor forming a ~1kHz low pass with the PZT capacitance of ~45nF. This yields a chracteristic time constant of ~150us. The glitches as seen by the PMC mixer are at least a few times faster than this.

The third plot shows a PDH scan.

The forth plot shows a typical trace when nothing happens.

TITLE: 11/20 Day Shift: 1530-0030 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Microseism
INCOMING OPERATOR: Ryan S
SHIFT SUMMARY: We spent the day in IDLE again investigating the PSL issues, we've had only the PMC locked for most of the day. At the end of the shift, the power supply for the PLS was swapped. Microseism has slowly starting to come down in the last ~5 hours. We are going to remain DOWN again tonight.
LOG:                                                                                                                                                                                                                                                                                                                  

• 15:00:26 UTC a minor frequency power glitch alog81376 earlier this morning tripped off the NPRO. The power discrepancy only lasted ~50ms.
• 17:15 UTC Jason brought the NPRO back
• 2ndary microseism seems to be decreasing, satellite images show the cyclone moving NW away from the WA coast today
• 17:55 UTC Jason and Marc turned off the PMC high voltage and unplugged the DB376 cable again to resume the test from yesterday (No IMC, FSS, or ISS, only PMC)
• I ran a spectra of the PMC_MIXER_OUT_DQ channel comparing times of similar range from the good lock we had on 11/19/24 to a lock back in 09/07/24

Yesterday I went to check on the chillers and found a bit of coolant on the CO2Y chiller air filter and on the bottom of the unit. The last time we saw this we had inconsistent supply temperatures, something we don't really see now but might explain our more frequent lock losses lately (alog 81362). We'll keep an eye on this and swap the chiller if necessary. It would be great if this could make it another 6 months and then we can send them in for service.

I replaced the air filter since it had a handful of dead bugs, coolant, and other junk on it. I used up our last filter so I'll have to order more.

Some of the "PSL story so far" is in 81193. Code saved to /camilla.compton/Documents/Locklosses/PSL_channels.ipynb

I've plotted H1:PSL-PWR_NPRO_OUT_DQ and H1:PSL-FSS_FAST_MON_OUT_DQ in the 30 seconds and 1 second before each locklosss tagged both IMC and FSS by the lockloss tool.

• Plots before the NPRO swap here and zoomed here.  There is large individual glitches, seen by both FSS and the NPRO PWR in an otherwise quiet signal before the NPRO swap.
• Plotted after the NPRO swap here and zoomed here. There is none of these individual glitches after the swap but there is the slower building oscillation in the FSS channel, which is overall noisier.

The FSS oscillations we are seeing now are similar to the FSS_OSCLATION locklosses we saw at the end of O3b (G2201762: O3a_O3b_summary.pdf). Note that the current NPRO is the same NPRO as we used in O3. Is it possible that we fixed the glitches issues but reinstalled a laser with similar issues as we saw in O3b? We could revisit that time period to see if we had similar IMC locking issues then too.

Reminder that the lockloss tool tags "IMC" if the IMC looses lock within 50ms of the arms (ASC_AS_A) and tags "FSS" if H1:PSL-FSS_FAST_MON_OUT_DQ with outside of +/- 3 within 1-5 seconds before the lockloss.

Since the secondary microseism has been very high this evening and preventing H1 from locking, we decided to leave just the PMC locked (no FSS, ISS, or IMC) for an extended time and watch for any glitches. At around 23:45 UTC, we unlocked the FSS, Richard turned off the high voltage supply for the FSS, and Jason and I unplugged the DB37 cable from the FSS fieldbox in the PSL-R1 rack in order to ensure no feedback from the FSS made it to the NPRO. Pictures of the DB37 cable's location are attached.

The first attachment shows the changes seen when the FSS was unlocked. Since then, I've seen several instances of groups of glitches come through, such as those shown in the second and third attachments. These glitches in the PMC_MIXER channel are smaller than ones seen previously that have unlocked the IMC (like in alog81228). There have also been times where the PMC_MIXER channel gets "fuzzier" for a bit and then calms down, shown in the fourth attachment; it's possible this is due to the NPRO frequency not being controlled so the PMC sees some frequency changes? Finally, I only recall one instance of the NPRO jumping in power like in the final attachment; the PMC doesn't seem to care much about this, only having one very small glitch at this time.

I'll leave the PSL in this configuration to collect more data overnight as the secondary microseism is still much too high for H1 to successfully lock.