thomas.shaffer@LIGO.ORG - posted 12:22, Thursday 23 May 2024 (78005)
Back to Observing 1920 UTC
Wrapped up commissioning at 1920UTC and we are back to Observing. Locked for 23:39
Wrapped up commissioning at 1920UTC and we are back to Observing. Locked for 23:39
I ran the A2L script again today, accepted the gain diffs for the 4 test masses.
The script is still in ...userapps/isc/common/scripts/decoup/a2l_min_generic_LHO.py . Some room for improvement is that it leaves several SDF diffs (TRAMPs on the 4 quads, and settings in the PIT7 and YAW7 ADS loops), so we should add into the script to revert those.
Attached is the printout of what the script measured, as well as a screenshot of the accepted a2l gain diffs.
The attached screenshot shows that the A2L script worked well, the ASC coherences around 30 Hz are now low (the CHARD Y coherence below 15 Hz is there if things are well tuned at 25-35 Hz, this has been the case for a long time).
Otherwise, our sensitivity is still worse below 50Hz than it was in early May. The PRCL coherence may explain some of our noise in this region, but it is the same level of coherence as in early May.
Alex Ritzie, Ansel Neunzert
Key points
The 9.5 Hz comb triplet (LHO worst offender comb), 11.11 Hz comb, and 11.9 Hz comb, are present in {H1,L1}:PSL-ISS_PD{A,B}_REL_OUT_DQ. These combs were previously known to be common between the two sites (see LLO 70030, LLO 66036, git issue for 9.5 and 11.9 for more info). A near-30 Hz comb is also seen at LHO only.
Details
Plots of H1:PSL-ISS_PDA_REL_OUT_DQ and L1:PSL-ISS_PDA_REL_OUT_DQ are attached (figures 1 and 2). Note that the color coding differs between the two plots. The 9.5 Hz combs are much stronger at LLO than at LHO. The 11.9 Hz comb is almost not visible at LLO, but the first harmonic is faintly present (figure 3).
Note that the 9.5 Hz combs may not appear to line up exactly on this plot at higher frequencies, but they are known to shift around regularly and I'm just plotting the average O4a positions here. I'm confident that it's the same artifact because the characteristic triplet structure is clearly apparent at every peak (see figures 4 and 5).
Both of these plots are from May 6. This date is not special, it's just a convenient recent date with a reasonable amount of observing time, which avoids other recent comb contamination issues at LLO.
Methods & context
Alex has been building an improved version of the STAMP-PEM / Fscan comb comparison tool first mentioned in this alog. Essentially, this tool leverages the relatively low-resolution but much larger channel set of STAMP-PEM's coherence data to try to pick out possible new witness channels for combs identified by Fscans. Alex's results for Hanford December 30-31 2023 are attached in figure 6 (more recent STAMP-PEM data was not yet available). This reproduces prior results showing the combs in IMC WFS channels, and highlights several other channels. Following up with high-resolution Fscan spectra, I found that the PSL accelerometers which appear on the heatmap show some broad peaks in the vicinity of some comb frequencies, but are not a match to the combs at high resolution (probably unsurprisingly) so I moved on to looking at the channel mentioned above, and from there to looking at the corresponding LLO data.
I have opened a git issue here https://git.ligo.org/detchar/detchar-requests/-/issues/243 with notes on next steps & follow up; additional input welcome.
Thu May 23 10:09:29 2024 INFO: Fill completed in 9min 25secs
Similar to what we've been seeing:
Famis 26302 - HVAC Vibrometer HVAC fan check
There was a small stange looking increase on the May 19th at 16:00 UTC that lasted about 16 hours on H0:VAC-MR_FAN6_170_1_ACC_INCHSEC.
Ran a calibration broadband and simulines measurement at 1530UTC and coordinated with LLO and Virgo.
Simulines start:
PDT: 2024-05-23 08:37:24.653206 PDT
UTC: 2024-05-23 15:37:24.653206 UTC
GPS: 1400513862.653206
End:
2024-05-23 15:58:54,982 | INFO | File written out to: /ligo/groups/cal/H1/measurements/DARMOLG_SS/DARMOLG_SS_20240523
T153725Z.hdf5
2024-05-23 15:58:54,989 | INFO | File written out to: /ligo/groups/cal/H1/measurements/PCALY2DARM_SS/PCALY2DARM_SS_20
240523T153725Z.hdf5
2024-05-23 15:58:54,994 | INFO | File written out to: /ligo/groups/cal/H1/measurements/SUSETMX_L1_SS/SUSETMX_L1_SS_20
240523T153725Z.hdf5
2024-05-23 15:58:54,999 | INFO | File written out to: /ligo/groups/cal/H1/measurements/SUSETMX_L2_SS/SUSETMX_L2_SS_20
240523T153725Z.hdf5
2024-05-23 15:58:55,005 | INFO | File written out to: /ligo/groups/cal/H1/measurements/SUSETMX_L3_SS/SUSETMX_L3_SS_20
240523T153725Z.hdf5
ICE default IO error handler doing an exit(), pid = 977458, errno = 32
PDT: 2024-05-23 08:58:55.069485 PDT
UTC: 2024-05-23 15:58:55.069485 UTC
GPS: 1400515153.069485
Famis 26242 PSL Status Weekly Report
Laser Status:
NPRO output power is 1.819W (nominal ~2W)
AMP1 output power is 66.77W (nominal ~70W)
AMP2 output power is 137.6W (nominal 135-140W)
NPRO watchdog is GREEN
AMP1 watchdog is GREEN
AMP2 watchdog is GREEN
PDWD watchdog is GREEN
PMC:
It has been locked 50 days, 21 hr 23 minutes
Reflected power = 19.39W
Transmitted power = 107.2W
PowerSum = 126.5W
FSS:
It has been locked for 0 days 20 hr and 48 min
TPD[V] = 0.9001V
ISS:
The diffracted power is around 2.2%
Last saturation event was 0 days 20 hours and 48 minutes ago
Possible Issues:
PMC reflected power is high
Started well pump to replenish fire water tank. Pump will run for 4 hours.
[Jenne, Joe B] Jenne mentioned last night that the calibrated range seemed to jump last night, and was worried the line heights need to be increased. Turns out she was correct. We have a threshold of 0.01 on our uncertainty calculations for the front end uncertainty calculations which gate the GDS pipeline's accepting or rejecting current time dependent correction factors. Looking at the summary pages, for the last hour or so of May 22nd and the first half of May 23rd, I can see the PUM real and imaginary components as reported by GDS (red lines) being stuck at a value for a while, then the uncertainty drops briefly, then updates, then freezes again. Unfortunately, because UIM, PUM and TST are used to correct the kappa_C and F_CC (optical gain and cavity pole) calculations, those 2 calculations also stop updating, although they seem to be closer the raw values (grey lines) on those pages. I've included trends of the low frequency line uncertainties, and I can clearly see the PUM line crossing the 0.01 several times in the last 10 days or so. I also took a quick look at a quiet time spectrum versus a recent spectrum, and there may be a bump at 16.5 Hz causing a problem for the PUM line. At a minimum I'd suggest raising the PUM line amplitude by more than a factor of 2 (trend is nearly touching 0.02, factor of two would drop it only to 0.01). I'd probably suggest a factor of 4. UIM also looks like it is kinda close, so I'd recommend raising that by a factor of 2. The channels which control the suspension line heights are H1:SUS-ETMX_{L1,L2,L3}_CAL_LINE_CLKGAIN, where the usual L1=UIM, L2=PUM, L3=TST applies. We should consider setting up a monitor that alerts the operator when this kind of condition occurs, similar to what LLO has set up (although we've gone a step further where we automatically increase the line height in these kinds of conditions after an hour). See Vlad's LLO alog 71177
TITLE: 05/23 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Observing at 145Mpc
OUTGOING OPERATOR: Corey
CURRENT ENVIRONMENT:
SEI_ENV state: CALM
Wind: 2mph Gusts, 0mph 5min avg
Primary useism: 0.01 μm/s
Secondary useism: 0.11 μm/s
QUICK SUMMARY: Locked for 19 hours, range has been slowly decreasing over the last 10 hours. Commissioning time and a calibration measurement planned for today 830-12PT.
TITLE: 05/23 Eve Shift: 2300-0800 UTC (1600-0100 PST), all times posted in UTC
STATE of H1: Observing at 154Mpc
INCOMING OPERATOR: Corey
SHIFT SUMMARY: Very quiet shift tonight. H1 remained locked and observing the whole time, lock stretch currently up to 12.5 hours. Wind has died down and microseism is starting to drop.
LOG: No log for this shift.
State of H1: Observing at 155Mpc
Quiet evening here so far; H1 has been locked and observing for 8 hours.
TITLE: 05/22 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Observing at 153Mpc
INCOMING OPERATOR: Ryan S
SHIFT SUMMARY: 3 locklosses today, relocking was straightforward, we've been locked for ~4 hours.
14:20 UTC lockloss, there was a ~13 Hz oscillation in SRCL, PRCL, and DARM
16:02 UTC back to Observing after some SQZer aligning
16:16 UTC lockloss, similar as earlier, ~13 Hz oscillation right before LL
Sheila removed a DARM filter alog77983
17:20 UTC back to Observing
18:37 UTC lockloss
The alignment seems to be degrading after each LL, DRMI/PRMI flashes are worse each time. I had to move PRM 15 microradians in pitch to help PRMI seeing AS AIR camera clearly pitched. On to DRMI AS AIR looks yawed now but the flashes are much better post ASC
Sheila measured the PR2 spot once we got close to NLN alog77988
22:19 I saw a DIAG_MAIN message "PMC PZT volts high" and shortly after the SQZer lost lock and dropped us out of Observing
I ran an alignment and angle scan to optimize squeezing after it relocked itself (SDF diffs) since we were already out of observing and I was just thinking that SQZing wasn't looking great at high frequency, the trace on nuc33 didn't look much different after finishing and DARMs noise still looks elevated at high and low frequencies but the range seems to have increased by a few Mpcs.
22:32 UTC back to Observing
LOG:
| Start Time | System | Name | Location | Lazer_Haz | Task | Time End |
|---|---|---|---|---|---|---|
| 15:18 | FAC | Karen | Optics & VPW | N | Tech clean | 15:48 |
| 18:05 | FAC | Karen | Woodshop/Firepump | N | Tech clean | 18:19 |
TITLE: 05/22 Eve Shift: 2300-0800 UTC (1600-0100 PST), all times posted in UTC
STATE of H1: Observing at 155Mpc
OUTGOING OPERATOR: Ryan C
CURRENT ENVIRONMENT:
SEI_ENV state: CALM
Wind: 11mph Gusts, 8mph 5min avg
Primary useism: 0.02 μm/s
Secondary useism: 0.18 μm/s
QUICK SUMMARY: H1 has been locked and observing for 3.5 hours.
Since we are considering reverting the PR2 spot move, I measured the spot positions by adjusting A2L gains for PR2, PRM, and PR3. The first attached screenshot shows the current sliders and pico position, the second screenshot shows the A2L gains in SDF.
| A2L gain | ||
| PRM Y2L | 0.48 | |
| PRM P2L | 1.76 | |
| PR2 Y2L | -0.39 | |
| PR2 P2L | -3.0 | |
| PR3 Y2L | 1.5 | |
| PR3 P2L | -0.3 |
Here is the calculation of what these a2l coefficients imply in spot position (fixing the typo of swapping the PR2 P2L and Y2L coeffs):
| A2L gain | Spot position [mm] | |
| PRM Y2L | 0.48 | 1.0 |
| PRM P2L | 1.76 | 3.6 |
| PR2 Y2L | -3.0 | -6.0 |
| PR2 P2L | -0.39 | -0.8 |
I don't have handy the conversion for the larger PR3 triple suspension's A2L coeff to spot position, and Sheila also mentioned that the error bars on that one are a bit larger, so I'm not quoting a spot position for PR3.
| PRM P2L | 1.7 | |
| PRM Y2L | 0.52 | |
| PR2 P2L | -0.38 | |
| PR2 Y2L | -7.5 | |
| PR3 P2L | -1.5 | |
| PR3 Y2L | 1.6 |
In the measurements above, I didn't mean that the PR3 gains had more uncertainty than the others, but that the pitch ones had more uncertainty than yaw, because when I was tuning pitch I sometimes didn't see a clear difference. PR3 yaw seemed similar to the other yaw measurements.
I've accepted this in the OBSERVE and the safe.snap. Also attached is a screenshot of the sliders at the time of this measurement.
https://ldas-jobs.ligo-wa.caltech.edu/~lockloss/index.cgi?event=1400438280
Back into Observing at 20:10 UTC after Sheila finished measuring the PR2 spot and clearing the SDF diffs, I reverted these for the test masses
We need to reduce the DHARD rms to make our SRCL to darm coupling more stationary, (77729 77666) I ran a noise injection to measure the DHARD P loop this morning.
Elenna measured the hard loops and compared them to a model while we had 430kW in the arms in 69597, today's measurement is with 375kW. I used a template I found in userapps/h1/asc/templates/DHARD/DHARD_P_broadband_excitation.xml, which I think is the one that Elenna tuned in 69597.
Attached is a very simple script that uses dttxml to get the transfer function data from an xml file, getting the IN2/exc/IN1/exc unbiased transfer function. I also plotted uncertainties based on 10506, calculating an uncertainty for magnitude and phase from in1/exc and in2/exc separately and adding them in quadrature (this is not the correct way to add these uncertainties).
It seems like we have room to add a boost to this loop.
For modeling purposes, here's a fit to the DHARD_P plant obtained from this measurement, after factoring out the current controller.
Z = [-0.37656602]
P = [-0.00292406 +0.j ,-0.11496697 +6.49363092j,
-0.11496697 -6.49363092j,-0.19955184+16.53674456j,
-0.19955184-16.53674456j]
K = 2176.925725327256
Here's a possible design for a boost to improve suppression in the region below 0.7 Hz
zpk([-2.723526385480632+i*3.098947495963098;-2.723526385480632-i*3.098947495963098],[-0.6734376457906343+i*2.23044514115136;-0.6734376457906343-i*2.23044514115136],0.9990715840784479)
I have saved this new boost as GVboost in DHARD P FM8, but I haven't yet loaded the filter file.
