Broadband PCAL measurement taken this afternoon starting at:
PDT: 2024-04-05 15:19:02 PDT
UTC: 2024-04-05 22:19:02 UTC
GPS: 1396390760
Measurement ran for ~5 minutes.
FAMIS 26238
I added the PDWD to this script after the first time I ran this.
Laser Status:
NPRO output power is 1.811W (nominal ~2W)
AMP1 output power is 66.96W (nominal ~70W)
AMP2 output power is 137.3W (nominal 135-140W)
NPRO watchdog is GREEN
AMP1 watchdog is GREEN
AMP2 watchdog is GREEN
PDWD watchdog is GREEN
PMC:
It has been locked 3 days, 3 hr 59 minutes
Reflected power = 17.61W
Transmitted power = 108.3W
PowerSum = 125.9W
FSS:
It has been locked for 0 days 4 hr and 45 min
TPD[V] = 0.8019V
ISS:
The diffracted power is around 2.3%
Last saturation event was 0 days 1 hours and 29 minutes ago
Possible Issues: None reported
We've added a new path to the CAMERA_SERVO guardian which will now be the default, where we set the camera offsets to fixed values stored in lscparams rather than reseting it each lock based on the ADS set points. This is because we found last night that the build ups were better at a different offset for the BS camera 76961
We caused one lockloss doing this switch, which we think is because I changed the order of doing the switch. We've edited the guardian in a way that we think should not have a large transient.
Eric, Camilla, Naoki
Attached are some plots showing several FOM for the impact of changing the ZM4 and ZM5 PSAMS on the squeezing seen in DARM. The axes of these plots show the strain gauge readouts for the PSAMs since the piezos are known to have issues with hysteresis. Our nominal setting is (7.2, -0.72).
Initially we had relatively little reason to believe that the current settings were optimal. Both PSAMS were centered in their range in terms of voltage after offloading during the commissioning break. We decided to start with a broad scan (see PSAMS.jpg) and chose to prioritize regions not covered by the PSAM scan in 76507. We Note, however, that the scans in 76507 were completed prior to the recomissioning of the SQZ alignment optimization scripts which may impact those results to some degree. We learned that the current point is actually not too far from optimal.
We then narrowed in on points closer to (7.2, -0.72) (see PSAMS_zoom.jpg). It looks like there may be a point which produces slightly higher levels of ASQZ. We would like to explore the area around this PSAM setting more carefully.
A few comments:
We've added a script to pull data to do this analyisis to opt/rtcds/userapps/release/sqz/h1/scripts/PSAMS_data/pull_PSAMS_data.py.
It looks at channels = ['H1:AWC-ZM4_PSAMS_STRAIN_VOLTAGE','H1:AWC-ZM5_PSAMS_STRAIN_VOLTAGE', 'H1:SQZ-CLF_REFL_RF6_PHASE_PHASEDEG', 'H1:SQZ-DCPD_RATIO_1_DB_MON', 'H1:SQZ-DCPD_RATIO_3_DB_MON', 'H1:SQZ-DCPD_RATIO_4_DB_MON', 'H1:SQZ-DCPD_RATIO_5_DB_MON', 'H1:SQZ-OMC_TRANS_RF3_DEMOD_RFMON'] which is BLRMS at 85Hz, 350Hz, broad 1kHz, 1.7kHz.
You need to feed it a list of ASQZ and SQZ gpstimes as a text file, example attached, we used data from 76949 and 76925.
Results attached. As expected they mainly agree with Eric's.
First plot zoom_heatmap.png is with reduced color bars to show the details of which of the better PSAMS positions are best, heatmap.png has full color bars showing how bad the minimum PSAMS settings are.
I took some sqz and no sqz quiet times from yesterday to try and make a plot of the excess noise agaion as in this entry. As Vicky thinks this correlated (non-quantum) noise is getting better.
For the first image I compared calibrated DARM from this week with the same measurement from March and from 04a.
Grey is from 04a, purple is from previous slide with March data, yellow and green are from this week.
What we can get from this plot is that this non-quantum noise is different from lock to lock and we are not sure why.
Between 80 and 190 Hz the excess noise seems to now be better relative to March and 04a.
But the fact the noise without squeezing changes so much in a week is confusing.
In the second plot I have just compared three no squeeze quiet times from this week and it easier to see that the noise is not consistent below 200Hz.
Fri Apr 05 10:05:56 2024 INFO: Fill completed in 5min 53secs
Gerardo confirmed a good fill curbside. TCtrip raised to -100C while this cold spell lasts.
FAMIS 19972
pH of PSL chiller water was measured to be just above 10.0 according to the color of the test strip.
S. Dwyer, R. Short
During H1's locking troubles this morning, we decided at one point to start an initial alignment from the start since DRMI was not locking. While the ACQUIRE_XARM_IR state was running during the input alignment steps, Sheila noticed the LSC-TR_X signal was far off from zero with no light, meaning there was a dark offset that needed to be fixed. I took the mode cleaner offline so we could observe the dark signal on the POP_A_RF45 photodiode and applied the negative value of that to the offset for both the I and Q segments. This allowed XARM IR to catch quickly, and input alignment completed without issue (contrary to its behavior in recent weeks). Sheila has updated the hard-coded offsets in ALIGN_IFO's PREP_ARM_IR state generator (lines 770-771) and included a note that these need to be updated periodically when input alignment has trouble.
This should fix the troubles we've been seeing with input alignment and serves as a reminder to check for any erroneous changes to dark offsets if we notice issues with alignment steps.
https://ldas-jobs.ligo.caltech.edu/~gabriele.vajente/bruco_1396341217_GDS_CALIB_CLEAN/
DHARD_Y is pretty bad, much worse than usual. Probably need a retuning of A2L
Jennie, Sheila, Camilla
We expect DHARD yaw to be worse. In the seconds half of last night's lock the CAM_YAW1 offset was changed by ~7 counts, see 76961. It is expected that A2L would need to be retuned, we are still working on adding these 76961 changes to the guardian.
TITLE: 04/05 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Earthquake
OUTGOING OPERATOR: TJ
CURRENT ENVIRONMENT:
SEI_ENV state: CALM
Wind: 4mph Gusts, 3mph 5min avg
Primary useism: 0.04 μm/s
Secondary useism: 0.32 μm/s
QUICK SUMMARY: H1 is currently running and initial alignment following downtime for about 5 hours from an earthquake (mentioned in the owl shift alog).
The IFO lost lock at around 1015UTC, then while running an alignment, a 6.8M earthquake from the pacific rolled through, tripping ST1&2 ITMY ISI. This also put SEI_ENV into error for requesting the EARTHQUAKE state from SEI_CONF, which for reasons is actually EARTH_QUAKE. I'm waiting for the ground motion to come down a bit before finishing the alignment and starting to relock.
After waiting a bit for the earth to calm down, I tried finishing up the alingment, which was a non-issue. When I tried locking, I was able to lock PRMI but not DRMI. The AS air camera spot looked off, but no matter how much I adjusted SRM, I couldn't get it to lock or get the spot to look much better, I ran another alignment, but skipped green and input align. At this point another earthquake rolled through and made SRC a bit tricky, but it eventually locked. Second time around and I was back in the same spot. PRMI would lock, though it perhaps looked shaky, but I couldn't get DRMI to lock. It would have many "baby" locks, but I couldn't seem to increase the POP flashes. I tried changing the trigger thresholds for SRCL and MICH, but this didn't seem to help either.
I ended up giving up and running a full alignment. It's currently finishing up SRC and I'm handing this off to Ryan.
TITLE: 04/05 Eve Shift: 23:00-07:00 UTC (16:00-00:00 PST), all times posted in UTC
STATE of H1: Observing at 150Mpc
INCOMING OPERATOR: TJ
SHIFT SUMMARY: We've been locked for 23:18 as of 07:00UTC
Lock#1
LOG:
Start Time | System | Name | Location | Lazer_Haz | Task | Time End |
---|---|---|---|---|---|---|
23:42 | CAL | Francisco | PCAL lab | LOCAL | PCAL work | 00:15 |
We briefly dropped from Observing due to the SQZer losing lock from 00:58 till 01:14 UTC. I reloaded SEI_ENV during this time which caused it to fail due to an index error, so Dave and I reverted the latest changes and restarted the node.
We've been locked and observing since, lock time of 19:20 as of 03:00UTC
EX Cryobaffle
In January we damped the EX cryobaffle through a viewport (75553) because it had not been available for damping during the O3-O4 break and because the scattering noise from it increased with beam power (70808). The baffle has a 4 Hz resonance that matched a 4 Hz mode of the isolation system of one of the HVAC fans, making it particularly noisy early in the run, until we switched fans. Figure 1 shows that, after damping, about five times more motion of the beam tube is required to produce about the same level of noise in DARM.
Input Arm
During the break, we doubled the angle of the MC baffles in the input arm, so that there would be no specular reflection paths between optics, we added more damping to the MC baffles, added nozzle baffles to each of the nozzles with blank-off flanges, and added an unplanned baffle for a glint (75726).
Figure 2 shows a frequency sweep, injected by a speaker, and the resulting noise in DARM before and immediately after the break. The noise produced by injections above 15 Hz, is greatly reduced immediately after the break. However, we had also hoped to better reduce the noise produced by injections below 15 Hz. I spent last week looking into glints that I could see in the chamber and other potential sources of the residual coupling below 15 Hz. I found that I could greatly reduce the residual coupling by changing the yaw of the ITMY compensation plate by 50 microradians. The last pair of spectrograms in Fig. 1 shows that this CP angling further reduced the coupling to levels that we had hoped to achieve.
Figure 3 shows the elimination of noise for a 13.1 Hz injection after angling the ITMY compensation plate in yaw. The second page shows that there is repeatable structure at about 20 microradians in the coupling, suggesting that there are multiple ghost beams or, more likely, multiple retroreflecting structures. The third page shows a bright light on the wall of a spool piece that bears further investigation.
TITLE: 04/04 Eve Shift: 23:00-07:00 UTC (16:00-00:00 PST), all times posted in UTC
STATE of H1: Commissioning
OUTGOING OPERATOR: Corey
CURRENT ENVIRONMENT:
SEI_ENV state: CALM
Wind: 11mph Gusts, 8mph 5min avg
Primary useism: 0.03 μm/s
Secondary useism: 0.28 μm/s
QUICK SUMMARY:
Back into Observing at 23:00 UTC
SQZ_PMC is giving notifications of "PMC PZT volts low", the node returns this notification if the SQZ-PMC_VOLTS channel is less than 5. Its decayed from 80 to where it is currently, hovering around 5 over the past day.
SQZ unlocked dropping us out of observing at 00:58UTC, I also took the moment to reload SEI_ENV which caused it to lose FAIL/lose connection? I'm going to call Dave
Dave and I reverted the most recent code changes from Jim and I restarted the node with "guardctrl restart SEI_ENV", the codes changes included some states becoming unrequestable and a few other small changes, none of which made any sense for the "Index error" that the LOG had. Verbal alarms is now constantly throwing a TypeError for SEI_ENV "NoneType is not iterable". I went back into Observing at 01:14 after the SQZer relocked itself and then the MANAGER prompted me to "Return to FDS" so I requested us back to FDS
Attachment 1 - Unsqueezed (left) vs. squeezed (right) DARMs in ER16, with respect to O4a. Like Gabriele's plots, the ratios on the bottom show a given trace / O4a reference, so ratio > 1 means more noise now than O4a.
Colors: Rainbow colors are ordered past to present: e.g. Red = O4a, ... Purple = now.
Attachment 2 - SQZ dBs after subtracting non-quantum noise, with various recent PSAMS settings. Upper subplot: Anti-SQZ, Lower subplot: SQZ. Lines are a moving average of the data. Some caveats for the subtraction: here I am using the same H1 unsqueezed quantum noise model for all subtractions; this may not be the most accurate, but I think it should be fairly close.
Reference alogs wth times that I used to grab GDS_STRAIN_CLEAN:
Adding FDS times from the DARM comparison in 76955, and only including times with sqz/no-sqz comparable to those recent darm comparisons.
Attachment 1 - Unsqueezed (left) vs. squeezed (right) DARMs in ER16, with respect to O4a. Like Gabriele's plots, the ratios are trace / O4a reference, so ratio > 1 means more noise now than O4a.
Attachment 2 - SQZ dBs after subtracting non-quantum noise. Analysis with non-quantum noise subtraction for this week's PSAMS data in 76925 and 76949 ongoing.
Attachment 3 - Comparison to a very basic gwinc noise model (pulled from H1 NB) for squeezed darm. For example the dashed purple "Unsqueezed QN model" is (almost) what I am using for subtracting non-quantum noises.
Colors are rainbow-ordered time-wise. Red = O4a reference. Purple = 4/4 times from 76955 for FDS DARM comparisons.
Note: here the PSAMS settings are referenced in terms of strain voltages. Just pointing out that LLO has a guardian to servo PSAMS voltage to meet the "_TARGET" strain gauge voltage LLO:63403.
Reference alogs wth times used to grab GDS_STRAIN_CLEAN: