REMOVE SEARCH FILTER
SEARCH AGAIN
Search criteria
Section: H1
Task: AOS
[Ryan C, Elenna, Matilda, Camilla, Louis]
We have locked each arm on green. Looks like the X arm build up on ALS TRX is about 0.6, and Y is about 1.2. The COMM beatnote is -3 dBm, and the DIFF beatnote is -20 dBm.
Going to LOCK SLOW GWFS ETM TMS on Y arm does not engage the WFS, but works, which means the slow controls do work on y arm. I guess the WFS are disengaged for now.
However, it appears the X arm slow controls are still not working. Going to no slow with green wfs on X arm requires going through a state that does engage the slow controls, so we're still sitting in "locking" on X arm.
Camilla and I are about to head to the table to touch up the DIFF beatnote.
We figured out how to engage the X arm WFS by hand and they are improving the buildup.
****
We (Camilla, Louis, Elenna) have aligned the DIFF beatnote to -11 dBm. The biggest move was ensuring the beam made it to the DIFF PD- Camilla found both edges of the PD using the steering mirror and centered the beam on the PD.
As promised, this morning I took a new measurement of the JAC reflected beam on wavefront sensors A and B. This time, we unlocked the JAC, which provided a cleaner beam for centering in both pitch and yaw. The results were much better, with the relative differences between segments reduced to below about 10%.
As a sanity check, we compared the measured and calculated G1/G4 ratios for WFS A and WFS B, as shown in the attached table. For WFS A, the discrepancy is about 10%, while for WFS B, it is only about 2%. We think the remaining error may come from beam clipping when we attempted to recenter the beam between the two halves, the approximately 1 mm gap between the quadrants, or pitch-yaw cross-coupling.
Overall, this measurement gives us an estimate of the uncertainty associated with this technique for measuring the relative segment gains.
Per WP 13415 I finished migrating the dust monitors and the dewpoint monitor to the IOC cluster (service-host 0,1,2). With this I was able to turn off the old h0epics server. These will now auto start after power outages as well as there are no manual steps in the startup sequence.
One note.
The dewpoint monitor held its files in /ligo/lho/h0/target. As a limit in the container I am running these in it wants the command that is executed to be under /opt/rtcds. So I copied the target area over to /opt/rtcds/lho/h1/target and put a symlink in on /ligo/lho/h0/target/dewpoint.
Fil, Betsy
Today Fil and I did a quick repeat of T2200048 HAM7 ground loops check just before doors are going on.
Last checks were done:
Feb this year by Camilla/Sheila https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=89048
Feb 2022 Fil/me https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=61738
Keita had done them prior, hence the document.
We basically found the same results as Camilla/Sheila namely:
All the rest were "open".
Wanda V., Reinhardt R. On Friday (08/07) we connected both fibres which measure the arms to the Febus interrogator, with the X arm on Ch1 and the Y arm on Ch2. This frees up the Sintela interrogator for Reinhardts experiments. The Febus interrogator is now measuring at 2500 Hz with a downsample factor of 10 before saving, coming to a recorded sampling rate of 250 Hz. The gauge length was 6m. In case there is a crash or the measurement needs to be restarted, the pipleline to load is LIGO_20260807_DS_SR_Writer_both250hz_v2. Filesize is ca 50 MB/min/channel Monday August 10th: We measured at 10 m gaugelength over the weekend and it was changed to 6m at 11:00 am (PST) on Monday. The filesize is roughly 100 MB/min for the 6m gaugelength (so 200 MB/min for both channels) and roughly 60 MB/min per channel for 10m GL. If this measurement should be restarted, the pipeline is LIGO_20260807_DS_SR_Writer_both250hz_v3. In principle, the interrogator can now run "as is" through IR1, but we are still free to change any of the settings.
This has been wrong for forever, I fixed it at one point a long long time ago (alog 29798) but apparently didn't put it in SDF.
Anyway, FM4 of H1:PSL-ISS_SECONDLOOP_QPD_SEG filters named 0.4:40 was OFF, it should be ON all the time like in the attached screenshot. When it's wrong the signal looks noisier than it actually is and is annoying even though it's not used for control and doesn't matter during observing.
I accepted the change in the SDF.
Keita Kawabe, Jennie Wright, Masayuki Nakano, Khanh Vu
Below is a summary of how we closed the JAC ASC loops, including our measurements of the sensing and input matrices, the open-loop gain measurements, our work on the IOT1 table, and the final closing of the loops.
We attempted to measure the sensing matrix three times. During the first attempt, we observed strong cross-coupling between the pitch and yaw channels. However, we did not account for this coupling correctly because we inverted the full 4x4 sensing matrix, including the pitch-yaw cross terms, and then extracted the relevant values from the inverted matrix.
The second attempt was more successful. We removed the cross terms and reduced the 4x4 matrix, which would require 16 inputs, into two separate 2x2 matrices for pitch and yaw, requiring only eight inputs in total. We then inverted the two smaller matrices separately. There was also some confusion about the matrix labeling, but Masayuki noticed that the input and output ordering changes when the sensing matrix is inverted, so the rows and columns needed to be labeled accordingly. This proved to be correct. However, the newly calculated matrices were still not good enough because the PZT and JM1 responses in yaw were too similar. This was problematic because we wanted their signals on the two wavefront sensors to be well separated.
Therefore, we went to the IOT1 table to check the alignment. Masayuki made an important observation that the beam was being clipped at the shutter, with approximately half of the beam blocked on the left side. This explained why we were having more trouble with yaw than pitch, since the clipping affected yaw much more strongly.
After aligning the shutter, we measured the sensing matrix for a third time. The result was much better: the PZT and JM1 yaw response vectors were separated by approximately 45 degrees in the WFS basis, while the pitch response vectors were separated by approximately 124 degrees. Although these values are not at the ideal 90-degree separation, we determined that they were good enough and entered the new input matrices.
We then performed open-loop gain measurements and confirmed that all of the loops were behaving properly. Finally, we closed the loops and slowly increased the WFS gains from 0.1 to 1. All four loops closed successfully and worked well.
I added the ASC engagement to Guardian. A new ASC_ENGAGING state simply ramps JAC-WFS_GAIN from 0 to 1 over 1 second. It is a minimal implementation.
The CDS conda environment has been updated.
New versions of ndscope and diaggui can connect to an arrakis data source.
Also included is the arrakis-python client library. "include arrakis" to use in your python scripts.
Many packages have been moved up to the versions used in the IGWN environment.
A detailed list of changes can be found here:
https://git.ligo.org/cds/packaging/cds-conda-distribution/-/wikis/Environments#version-2026-08-04-01
You can run a command outside the environment with "noconda "
You can exit conda within a terminal window with "killconda".
You can activate other conda environments with "conda activate " or "conda activate ". Some additional configuration of your account may be needed to get "conda activate" to work.
The read me for the arrakis python module has a quick start
https://git.ligo.org/ngdd/arrakis-python/-/blob/main/README.md?ref_type=heads
Documentation is here:
Keita, Louis, Elenna
Today, we proceeded to look for the input alignment, given all the changes that occurred in HAM2 during the ISS install, 90545. Although we have been able to lock PRX, we have not yet seen any beam on ASC-POP_A or B QPDs (they capture the forward POP beam).
To start, we locked the JAC and IMC, and aligned PRM. We could see the usual beam on the ISCT1 REFL camera. We noted that the beam position on IM4 trans QPD was far from center, about 0.26 in pitch and 0.66 in yaw. The IM4 trans NSUM was about 1.825 W, which is similar to a pre-vent value of 1.842. At this time the IM3 sliders were P: 40.3 and Y: 655.6. We confirmed that the IM sliders were where we expected them to be set due to the vent work.
I moved IM3 sliders to center the beam on IM4 trans QPD, so new slider values were P: 103.3 and Y:472.6. Then, following Keita's direction, I proceeded to measure a transfer function of IM1 P and Y and IM3 P and Y to the IM4 trans NSUM as a way to quantify possible clipping in the input path. IM1 coupling should indicate clipping somewhere along the IFI, which includes some baffles. IM3 coupling should indicate clipping on the baffle between IM3 and 4.
The first four screenshots show those results, which I obtained by driving a 30 ct excitation from the test bank of each suspension dof at 8 Hz. IM1 P, IM1 Y, IM3 P, IM3 Y
Next, Keita turned off the IMC ASC and moved JM3 to bring the beam to a position on MC2 trans that recreated the beam position he found during the vent. He set a yaw offset of -0.64 on MC2 trans and then re-engaged the IMC ASC. After the ASC converged, I recentered the beam on IM4 trans using IM3 again. We confirmed that the NSUM on IM4 trans returned to 1.825. I reran the same coupling test above and we saw that for all four dofs, the coupling to IM4 trans NSUM increased. This is evident in both increased coherence and coupling value at 8 Hz. IM1 P, IM1 Y, IM3 P, IM3 Y
Then, Keita flipped the sign of the yaw offset to +0.64, since there is some confusion about which way the sign goes on that QPD. I recentered the beam again on IM4 trans, confirmed the NSUM came back to 1.825 , and reran the coupling measurement. The results show that for all four dofs, the coupling is decreased to below the starting value, shown in both the coherence and coupling value. IM1 P, IM1 Y, IM3 P, IM3 Y
It's possible that we are confused about the sign on MC2 trans, so that the first offset Keita tried actually went to the wrong position.
Another confusing point is that we see coupling in both pitch and yaw, and beam movement in both pitch and yaw. However, the beam translation we did was yaw only, to relieve yaw clipping.
Since we had found what we believe was a better beam position in HAM2, we checked the REFL alignment. With PRM aligned in this alignment, the beam is on the edge of the ISCT1 refl camera, so not great. There was also very little beam on the REFL WFS, and trying to run the REFL WFS centering servos rails RM2. We are now concerned about the REFL path and possible clipping on the REFL baffle in HAM2 as well.
To finish, Keita turned off the MC2 trans offset, and I reverted the IM3 sliders to the start position. However, the final screenshot here shows the final IM3 position we found where we think we had very little clipping in the input alignment.
None of theae alignments helped us find the beam on ASC POP A, but that's not surprising since we didn't make any IM4 moves. As we finished, Keita set up a long raster of IM3 and IM4 to look for a beam on those QPDs.
> transfer function of IM1 P and Y and IM3 P and Y to the IM4 trans NSUM as a way to quantify possible clipping in the input path. IM1 coupling should indicate clipping somewhere along the IFI, which includes some baffles. IM3 coupling should indicate clipping on the baffle between IM3 and 4.
The point is that IM3 to IM4_TRANS TF excludes the clipping between IM1 and IM3. Of course the TF from IM1 to IM4_TRANS can also show the clipping downstream of IM3 (such as baffles between IM3 and IM4).
> Next, Keita turned off the IMC ASC and moved JM3 to bring the beam to a position on MC2 trans that recreated the beam position he found during the vent.
Turned off the IMC ASC in a hope that MC1/2/3 were all hanging at the same angle as they used to during the in-air work. (Moving JM3 won't change the beam position on MC2 when ASC was not working, it just improves the matching into IMC.) Without ASC, MC2_TRANS YAW was 0.64.
I did this because of my recollection that we intentionally off-centered the MC2 in YAW, but that was wrong, what actually happened was that we did center the beam spot on MC2 because it was initially off in YAW but not in PIT, the only thing was that people including myself were somewhat suspicious about the MC2_TRANS path at the time, so marked the horizontal beam spot position in HAM3 using vertical hard edge put on the ISI surface and then measured the distance from the edge to the neighboring screw holes on the ISI surface.
I trended MC2 TRANS back to the time when Rahul and I finished centering the IM4_TRANS path using IMC flashes (alog 90536). Unfortunately we cannot see fast channels for individual segments but some flashes were long enough (short but lasted for ~3 clock cycles for 2kHz system) to produce meaningful PIT and YAW signals in MC2_TRANS, see Screenshot2026-08-05003532.png. It was 0.68 in YAW, not that different from 0.64.
Anyway I just put an offset of -0.64 to MC2_TRANS YAW and re-engaged IMC ASC to the beam spot on MC2 won't move in YAW.
I also tested the other side of MC2_TRANS.
As Elenna showed, the coupling from IM1/3 dither to IM4_TRANS changed with the MC2_TRANS YAW offset. Positive offset was better than zero offset which was better than negative offset.
As of now I cannot tell if this level of coupling is significant enough or not, we'll need beam propagation math to be able to say anything.
> Since we had found what we believe was a better beam position in HAM2, we checked the REFL alignment. With PRM aligned in this alignment, the beam is on the edge of the ISCT1 refl camera, so not great.
This is not surprising because we caused non-negligible change in the beam going into PRM and the beam was not retro-reflecting. On top of that, since IMC alignment change will change the alignment of the IFO REFL beam going into HAM1 even if PRM retroreflects.
However, with zero offset in MC2 YAW, without much care/attention to IMs,
That's already a sign of reasonable alignment. If we'll have to do a major rework of HAM2 to accomodate MC2_TRANS YAW offset, that seems to be a sign that the IMC is all in all different from where it was in-air desipite the MC2_TRANS YAW position of in-air flashes.
As a side project, I'll let Elenna and/or Louis measure the MC2 beam position offset by a2l without MC2_TRANS YAW offset as the intent was to center MC2.
I scanned IM3 (+-200urad, 0.043Hz) and IM4 (+-200urad, 0.053Hz) at the same time in YAW to find beam on POP_A and/or POP_B. With these frequencies, one scan cycle is exactly 1000seconds.
Laser power was increased to 35W. Whitening gain was nominal 12dB without any whitening filters, and there is -12dB gain in the digital to compensate.
PRM transmission is ~3%, PR2 transmission is ~230ppm and there's 90:10 splitter in the POP A/B path that throws away 90% of the power. Each QPD receive roughly half of that, i.e. 35W*3%*230ppm*0.1/2~ 12uW, which is not large but large enough so we can clearly see something if we believe the POP segments calibration (1 ADC count = 0.19 uW with 12db whitening gain and -12dB digital gain).
But I don't see anything, not 12uW, not even 1uW, really nothing. Attached is the trend for 3000 seconds. Are POP QPDs working? Connected?
I lowered the power to 2W and stopped excitation after the scan was done.
The plots I attached to my original alog are kind of impossible to parse, so I remade them comparing each dof at each offset. I also added in the calibration to the xml file, IM4 trans NSUM is calibrated into W and the damp ins of each suspension is ideally calibrated into urad. Now the transfer functions show real units, so they are more physically meaningful.
As a reminder, I drove the exact same excitation strength for each injection, 30 ct, and the line injected was almost 2 orders of magnitude above the noise in each damp inmon. Each measurement was run for 20 averages, 8 s BW with 50% overlap.
If you want to read the exact values in the file, you can find the xml template in my home directory (ligo/home/elenna.capote) as IM4_trans_coupling_calibrated.xml
Masayuki Nakano, Khanh Vu We successfully closed all of the JAC ASC loops. Attached below are the outputs from the two wavefront sensors. All of the signals converge to zero within approximately 30 seconds, indicating that the loops are working beautifully. More details will be reported tomorrow.
Jennie Wright, Masayuki Nakano, Khanh Vu This morning we worked on several tasks on the IOT1 table, including installing the camera and shutter, profiling the beam, and calibrating the DC power. We identified a new location for the camera using the beam transmitted through JACR_M5. During this process, Masayuki noticed that the beam was being clipped by the shutter. We suspect that the beam may have been clipped for some time. We then installed the camera in its new location, and Masayuki aligned the shutter on the table. Next, we profiled the beam for the wavefront sensors. We found that the Gouy phase separation between the two WFSs is approximately 70 degrees. We decided to leave the current configuration as it is since the separation is good enough. Masayuki will make a plot and perform a more detailed calculation later. We also maximized the laser power in the REFL path by optimizing the waveplate angle. When JAC is unlocked, the measured power on the RFPD is 5.1 mW, and the trigger PD voltage is 0.28 V. When JAC is locked, the measured power decreases to 0.4 mW, and the trigger PD voltage is 0.02 V. Since the beam is split evenly, each WFS receives approximately 2.55 mW of optical power. Finally, we calibrated the DC readout of RFPD by converting counts to mW. Before performing the calibration, Masayuki checked the alignment and recentered the RFPD. He then recorded two sets of measurements, each averaged over 10 seconds: Measurement #1 * JAC_REFL_A_LF_INMON: 1072.3966186523437 counts * DC power: 4.4 mW Measurement #2 * JAC_REFL_A_LF_INMON: 1073.5228637 counts * DC power: 4.5 mW After the calibration, we updated filter number 10 with the new coefficients.
The beam profile between the beamsplitter and the JAC WFS was measured. Here I fit a Gaussian beam to those measured beam sizes and convert the WFS locations into Gouy phase.
The measured beam diameters were fit independently in x and y with the standard Gaussian beam model, w(z) = w0 * sqrt(1 + ((z - z0)/zR)^2) with zR = pi * w0^2 / lambda and lambda = 1064 nm:
| w0 [um] | z0 [m from JACR_BS4] | zR [cm] | |
|---|---|---|---|
| x | 141.2 | 0.377 | 5.89 |
| y | 151.9 | 0.370 | 6.81 |
The beam is slightly astigmatic, so x and y are treated separately throughout.
The positions of WFS A and WFS B were measured with a ruler from the same reference as the profile scan: WFS A at z = 0.325 m, WFS B at z = 0.410 m. The corresponding Gouy phases, psi(z) = arctan((z - z0)/zR), are:
| WFS A [deg] | WFS B [deg] | Separation [deg] | |
|---|---|---|---|
| x | -41.3 | +29.5 | 70.8 |
| y | -33.5 | +30.4 | 63.9 |
The separation is 71 deg in x and 64 deg in y, not the optimal 90 deg. This is not optimal, but it is not terrible either: the two WFS remain well separated in Gouy phase and the sensing matrix would not be close to degenerate. Given the time available we did not optimize the layout.
If we want to optimize it later, the fix is straightforward: moving WFS A upstream (toward the BS) by about 5 cm in x / 7 cm in y, i.e. from z = 0.325 m to roughly z = 0.27 m, brings the separation to 90 deg. WFS B does not need to move.
Additional context for the work described above: The motivation for the table work came from the difficulties we had with the sensing and input matrices of the JAC ASC loops. In pitch, the PZT and JM1 signals are well separated, but their responses in yaw are too similar. This is problematic because we need the two wavefront sensors to distinguish between the motions of the two actuators. While identifying a new location for the camera, Masayuki noticed that the shutter was clipping approximately half of the beam on the left side. We suspect that the beam may have been clipped for some time and that this may be related to the yaw issue, since the clipping affects yaw more strongly than pitch.
Today we installed the iris to block the ghost beam on the JAC REFL path with an iris, and to re-measure the beam profile with it in place. The ghost beam was produced by the laser window which picks off the partial power of the JAC reflection beam (~0.4%). Since this laser window doesn't have the wedge on it, the AR reflection is not well separated. We observed this interference during the original beam profile measurement in this thread.
And now, the iris dumps the ghost beam, and we made a new the beam profile measurement. I made a good JAC REFL optical model which obtained by the fitting the beam profile measurement. We will use this model for the WFS signal calibration.
An iris was placed on the REFL path, between the first pick-off mirror and the first lens. To position it, the beam profiler was set just after the beam shutter, and the iris was closed while watching the profile, until the ghost was blocked and the main beam was left untouched. Actually, Since the ghost beam is very close, the main beam is partially blocked as shown in the attached pics. We will see if it would have any effect on our WFS signals.
1/e2 diameters along the REFL path, with JACR_MB4 as the origin:
| z [inch from JACR_MB4] | -21 | 3.5 | 5.5 | 7.5 | 9.5 | 11.5 |
|---|---|---|---|---|---|---|
| horizontal [μm] | 4440 | 1360 | 1140 | 895 | 696 | 500 |
| vertical [μm] | 4540 | 1360 | 1111 | 873 | 661 | 461 |
On-table distances were also measured: JACR_L1 to JACR_MB4 = 24", JACR_MB4 to WFS A = 12.5", JACR_MB4 to WFS B = 15.5".
I made a model of the beam propagation of the JAC REFL path from PSL to JAC and IOT1. This model was fitted to the new profile, with the PMC waist as origin.
Taken as known. The positions of lenses in PSL (IO_MB_L1/L2/L3) and of the JAC waist are the design values, i.e. the PSL bench to HAM1 relative distance is trusted.
Taken as unknown. The design placed the IOT1 table only loosely, and the periscope that matches the HAM1 beam height to the table height was estimated roughly. The distance from the JAC input to the IOT1 table is therefore the principal free parameter, allowed ±30 cm; it enters the calculation as the position of JACR_L1 measured from the PMC waist. The profile measurement was referenced to JACR_MB4, which carries its own error, so the JACR_MB4-JACR_L1 distance is a second free parameter.
The profile carries astigmatism, so a yaw tilt was allowed on each of the four lenses (three on the PSL bench, one on IOT1). This is deliberately over-parameterised: the individual tilts should not be read as physical alignment errors. However, the aim of this analysis is not to measure how each lens sits but a model accurate enough for the following calculation. So as long as the aquired prameters are physically reasonable, we can use these numbers as the following calculations.
One note:
The HAM1 periscope (JAC_M1/JAC_M2) rotates the beam 90 degrees about its axis, so the transverse planes swap on the way to JAC: bench x (YAW) descends from the upstream sagittal channel and bench y (PIT) from the tangential one. A tilt therefore gives astigmatism of opposite sign depending on which side of the periscope the lens sits. This is the reason why the x/y beam size flipped at periscope in the attached plot.
| parameter | fitted | vs design |
|---|---|---|
| JACR_L1 position from PMC waist | 11.4768 m | -21.5 cm |
| JACR_MB4 to JACR_L1 | 0.6004 m | -0.36" |
| yaw tilt, IO_MB_L1 / L2 / L3 | -16.9° / -7.1° / -1.8° | - |
| yaw tilt, JACR_L1 | -3.2° | - |
The fitted path from the JAC input to the IOT1 table comes out about 21.5 cm shorter than design, which is the scale of looseness that was expected there.
Astigmatism at the WFS planes
Expressed as the difference in accumulated Gouy phase between the two transverse axes:
| model | from the measured profile alone | |
|---|---|---|
| WFS A | -4.10° | -4.96° |
| WFS B | -10.61° | -10.32° |
Taking the cavity eigenmode as the reference, the fitted injection-lens tilts imply a mismatch of 0.48 % (tangential) and 0.43 % (sagittal), 0.91 % combined. Small enough not to conflict with the measured mode matching (~1%).
PMC eigen mode
| axis | w0 [μm] | zR [mm] |
|---|---|---|
| tangential (u) | 546.312 | 881.232 |
| sagittal (v) | 549.028 | 890.016 |
The waist sits at z = 0 in both axes. Downstream of the periscope the tangential channel becomes bench y (PIT) and the sagittal channel bench x (YAW).
z is given from the PMC waist (the model's own origin) and from JACR_MB4 (the origin the bench profile was measured against).
| element | z from PMC waist [m] | yaw tilt [deg] | source |
|---|---|---|---|
| PMC waist | 0.000000 | - | origin |
| IO_MB_L1 | 0.900000 | -16.86 | design / tilt fitted |
| IO_MB_L2 | 0.960000 | -7.12 | design / tilt fitted |
| IO_MB_M4 (PZT) | 2.812000 | - | design |
| IO_MB_L3 | 2.900000 | -1.75 | design / tilt fitted |
| HAM1 periscope (JAC_M2) | 7.040000 | - | design; x/y swap |
| JM1 | 7.268000 | - | design |
| JAC input mirror | 7.576000 | - | design |
| JAC waist | 7.826000 | - | design |
| JACR_L1 | 11.476799 | -3.15 | fitted |
| JACR_MB4 | 12.077174 | - | fitted (via JACR_L1 distance) |
| WFS A | 12.394674 | - | measured from JACR_MB4 |
| WFS B | 12.470874 | - | measured from JACR_MB4 |
| parameter | fitted value | design | allowed range |
|---|---|---|---|
| JACR_L1 from PMC waist | 11.476799 m | 11.691967 m | ±30 cm |
| JACR_L1 to JACR_MB4 | 0.600375 m | 0.609600 m | ±1" |
| yaw tilt, IO_MB_L1 | -16.8588° | 0 | ±20° |
| yaw tilt, IO_MB_L2 | -7.1210° | 0 | ±20° |
| yaw tilt, IO_MB_L3 | -1.7534° | 0 | ±20° |
| yaw tilt, JACR_L1 | -3.1527° | 0 | ±20° |
P. Thomas, F. Clara, R. Short, T. Sanchez, R. McCarthy WP 13441. The laser safety interlock code is now running (temporarily) on the CX2040-0155 machine labeled 'testing' in the MSR at IP address 10.105.0.113. It is using commit 86f4fb088fbed6ce3c0e47afcdfc756547e70416 on the branch labeled 'scripting' in the lho-laser-safety gitlab repository: https://git.ligo.org/cds/ifo/beckhoff/lho-laser-safety/-/commit/86f4fb088fbed6ce3c0e47afcdfc756547e70416 The original intention was to use the machine that was already running the previous version of the code, a C5210-0020 in the MSR labeled h1safety0 at IP address 10.105.0.10. However, after reimaging this machine with the Beckhoff service tool using the IN-0406-0112-03-0-2021-21-0002H_1.TIB, IN-0406-0112-03-0-2021-21-0002H_2.TIB, and IN-0406-0112-03-0-2021-21-0002H_3.TIB images, the computer went into a cycle where it showed the Windows square with a spinning busy indicator, then went blank and restarted again and again. I then tried using the IN-0303-0010-03-0-2020-11-0001V_1.TIB, IN-0303-0010-03-0-2020-11-0001V_2.TIB, IN-0303-0010-03-0-2020-11-0001V_3.TIB, and IN-0303-0010-03-0-2020-11-0001V_4.TIB images. This started off more promising, but then went to a completely black and unresponsive screen. I contacted Beckhoff technical support and they said these were the wrong images for this machine. They sent me IN-0406-0112-03-0-2024-00-00043_1.TIB, IN-0406-0112-03-0-2024-00-00043_2.TIB, and IN-0406-0112-03-0-2024-00-00043_3.TIB. I tried these but they sent the computer into a boot cycle like the first images. They then suggested updating the code on the service tool. At this point it was getting late in the day, so the decision was made to use the test machine to get things back up and running. I reimaged the test machine with the CX1800-0511-1009v2.4a_1.TIB, CX1800-0511-1009v2.4a_2.TIB, and CX1800-0511-1009v2.4a_3.TIB images and proceeded with the rest of the work permit. Filiberto, Ryan, Tony and I verified the following: IOT1 doors, LVEA Exit ESTOP, LVEA Entrance ESTOP, ISCT1 doors, IOT2 doors, PSL ESTOP, TCSY doors, TCSY ESTOP, CHETAY ESTOP, SQZ ESTOP, SQZT7 doors, SQZT0 doors, SQZ Table ESTOP, HWS doors, HWS ESTOP, TCSX doors, TCSX ESTOP, HIGH BAY Exit ESTOP, HIGH BAY Entrance ESTOP, CHETAX ESTOP, EY VEA Entrance ESTOP, EY VEA Exit ESTOP, EY ALS ESTOP, EY ALS doors, EY PCAL Enclosure, EX VEA Entrance ESTOP, EY VEA Exit ESTOP, EX ALS ESTOP, EX ALS doors, EX PCAL Enclosure, FCES ESTOP, FCES doors. The EX PCAL enclosure only trips off the EX and EY PCAL lasers. The EY PCAL enclosure only trips off the EX and EY PCAL lasers. The CHETA doors tripped off all of the lasers, which turned out to be the wrong thing to do. Filiberto made a hardware change so that this wouldn't happen, but the code needs to be eventually changed to fix this.
For reference, the Beckhoff restore tool reads: USB Image C9900-I901 v2.1.9.47 Computer Name: BST-000f7zrg
WP 13459
The outputs of the EP1957-0022 terminal are used to enable the CHETAY, CHETAX, and the CRS lasers. The CHETA enclosures are not installed. This required the inputs to the EP1957 for the panel/doors to be shorted. Verified enable outputs were present when system was nominal. Verified enable outputs went to 0V when an e-stop was pushed. Same was done for the CRS.
To clear the error on the EL2904, a phoenix safety relay was installed. This satisfied the minimum required load.
The branch labeled 'scripted' has been merged and deleted. The link still works however (https://git.ligo.org/cds/ifo/beckhoff/lho-laser-safety/-/commit/86f4fb088fbed6ce3c0e47afcdfc756547e70416).
Whatever happened to IM1, Rahul did make it better again. Reasons for improvement is unknown but we proceeded with the alignment and we're mostly done.
There is a mystery scattering or maybe clipping somewhere close to the IFI output but not on the output baffle nor DKDP baffle. It could be a scatter from the CWP surface, or a ghost beam somewhere, or something else. It will take a LONG time to diagnose this, and quite likely this existed for a long time. I'm tempted to leave it at this time. But I'll try to take some more pictures.
There's also something weird about the 1st pico mirror PIT actuator for ISS array.
We'd still like to take pictures/measurements here and there on Tuesday.
Following the morning work (alog 90525), REFL ASC censors were centered using RM1 and RM2. About ~30% of DAC range was used for RM2 (RMs_happy_again.png). FYI, using flash peak, [P,Y]=[-0.003, 0.015] for REFL_A and [-0.020, 0.001] for REFL_B. (Doesn't matter how close these are to zero as far as they're within +-0.1 or so and the SUM is decent, but it feels better to be able to get close to zero.)
Then we looked at the ISS array and the QPD was not centered. A quick adjustment of the second pico that is closer to the array was all it needed (keita_ISS_happy1.png). [P, Y]=[-0.03, 0.05].
We looked at the IFO REFL baffle (HA13) in front of the HAM2-HAM3 septum window and it was too high even though it was not clipping, so we lowered it by a couple mm. Before: lower edge height = 105.5mm; upper edge = 206.8mm. After: lower edge = 103.6mm; upper edge = 204.5mm.
This baffle was already moved by a couple mm in +Y direction last week (because the beam was closer to +Y edge). See IFOREFL_baffle_before_relocation.jpg and IFOREFL_baffle_after_relocation.jpg though it might be difficult to see the difference from this picture.
IFO REFL beam looked like IFOREFL_HA13_baffle_after.mov after the height adjustment.
We proceeded to check the IFI output baffle and I was bothered to find that something that looked like clipping was visible close to the left edge of the baffle using the IR camera. See IFI_OUT_clipped_720p.mov, this is a view from +Y door. But this was less frequent than the flashing itself. In a retrospect, this was probably the reflection from the PRM when the beam was swinging to the left of the video, but anyway we did various things:
See IFI_out_another_view.mov, this is after the beam quieted down enough and after PRM PIT was changed. The bright thing at the left side of the baffle hole is not visible any more, but you can still see bright-ish scattering of some sort inside the baffle aperture which was there even when Rahul blocked the beam between PRM and IM4. IFI_out_another_view.jpg shows the same thing but with more useful exposure. This is concerning.
DKDP baffle behind the IFI output cwp baffle looked OK (IFI_DKDP_baffle.mp4). BTW, as was reported before, it looked to us that the IFI output baffle has an offset in -X+Y direction relative to DKDP. In the video, the beam on DKDP is slightly biased to the right on average because we tried to split the difference between two baffles.
We moved IM1 in YAW by +-200urad while observing the IFI output by IR viewer to see if there is a better beam position on IFI output. It seemed to me that actually we don't have much space here.
Look at IFI.png to see how the beam is routed through DKDP, output CWP and then passes by the parking beam dump pick-off.
We'll be better once we're in vacuum because things will be quieter and the MC alignment will be better, so no beam motion and no HOM transmission, but I have to say that the clearance here looks to be unnecessarily narrow. I will NOT touch IFI itself so the only option for mitigation will be to move the parking BD pickoff, but it will be tedious to align that pickoff to steer the beam into the beam dump on top of HAM2. Given the limited time available I'll leave it as is.
We don't know what this scattering is, maybe it's the AR reflection of CWP or DKDP hitting something, maybe it's the surface scatter of CWP. I'll try to take the video from the back of the IFI output beamdump/CWP using a big inspection mirror.
Rahul used pico mirror to roughly center IM4 TRANS while I was monitoring individual segments. IM4TRANS_ROUGHLY_BALANCED.png
As planned. The beam originally was offset in +X direction (IM4_HA12_before.jpeg shows the original location, IM4_baffle_HA12_before.mov shows the beam position), IM4 baffle was moved a bit in -X direction (IM4_HA12_after.jpg, IM4_baffle_HA12_after.mov).
Tuesday update:
We took new pico pictures today. All picos seem to have decent threads both ways, nothing is close to mechanical stops.
Potential issue I was worried about was that the pit actuator stop ring for iss array pico 1 might be directly contacting the aluminum frame of the mirror holder, which means that the ring got loose or maybe the ball end was lost when it was assembled. See ISS_array_pico1.jpg, the ring is circled in red. I tried to rotate the ring by finger while holding onto the actuator screw so the latter won't rotate, and couldn't move the former. It's not like the ring is loose. Also there seems to be a gap between the ring and the mirror holder frame. Also see ISS_array_pico1_zoom.jpg. These rings were manually removed and put back on during the initial assembly, so my guess is that this specific ring was set shallower than other actuators from the beginning. It's fine.
We confirmed that the beam position in front of PRM was pretty good without any adjustment of IMs.
See the screenshot of alignment sliders as of now (even though HAM2 suspensions are in safe mode now, slider values should be valid).
See alog 90549. I don't know what that is, but it is not the main beam clipping. I recommend to move on. See how Disha, Jennie, Rahul (and myself) feel.
Ground check in HAM2/3 for IO/PSL/ISC.
(Added later: Forgot to attach the photo of the retroreflection check iris, so here it is. retro_check_iris.jpg. Each time the IM2-IM3 line changed the iris itself had to be recentered, and then the return beam should be centered on the iris using PRM.)
Note that the anti-whitening in FM4 of H1:PSL-ISS_SECONDLOOP_QPD_SEG filters were all OFF, though it should have been ON. (Explanation about this in alog 29798.)
This means that the signal was amplified by a factor of 10 for frequency higher than 4Hz.
It's still useful to look at the relative peak heights between segments but don't trust the absolute value. Same thing for H1:PSL-ISS_SECONDLOOP_QPD_SUM and NSUM.
I compared some power sensors like what RyanS did in alog91528 from our PRMI lock yesterday (08/19/26, GPS 1471225645) to a good lock at the end of 04 where we offloaded PRMI, then locked to NLN (11/16/25 GPS 1447346796).