It seems like Detchar must already be aware of this, but we are having huge glitches durring this lock long lock stretch, which show up in auxiallary degrees of freedom. We were wondering if anyone has checked for things like DAC crossings, or saturations.
Note, we're on a 20 hour lock stretch!
And the previous lock was also long at around 18 hrs.
The relock time between these 2 long stretches took ~1 hour.
Nice.
(Now we just need a trendable range channel...)
Since this was a feature we had in iLIGO, why did we devolve?
I have set the Intent Bit to Commissioning for some measurements by Elli.
Sudarshan reports that a PCal line was turned off sometime last night. He is turning it back on at 17:53.
Darkhan, Sudarshan
Pcal Lines got turned off last night during a pcal sweep measurements (LHO alog 20734 and 20732) because the optical follower servo got unlocked. We turned two lines one at 36.7 Hz and the other one at 331.9 Hz back on. We ramped it slowly to avoid any lock loss but we still saw some drop in the range. We left the higher frequency line at 1083.7 Hz off for now. We will turn this back on during the comissioning period or next lockloss opportunity.
Attached is a trend of Optical Follower Servo error signal showing when the lines got turned off. (around 2015-08-21 07:20:00 UTC)
Are we meant to be able to see the PCal lines in the normalised spectrogram of DARM? You can see them disappear and turn on again at about the times you mention, see the first plot (this is GDS strain). Also PCal End Y doesn't look so happy, see second plot. Plots were taken from the PCal part of the summary pages
Yes, Pcal line are supposed to appear in h(t).
Also, the third line at 1083.7 Hz is turned back on after the lockloss.
What's the best way for Operators to confirm whether PCal (and DARM) Cal Lines are present? (seeing Excitation on CDS Overview? looking for lines on DARM spectra? Navigating to Calibration Line medms?)
Let us know and we can put this in checklists for operators to check.
I made a DTT template which has all the calibration lines on it. May be we can arrange to display this on the screen (A screenshot is attached.). The template sits on the following location.
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/ER8/H1/Scripts/Templates/ Calibration_line_template.xml
The other way is to head to PCAL medm screen and look at the OFSPD plot on the screen. If there is no excitation this plot should be flat.
With Vern's approval, Commissioners have opened some shutters to return to a closer state to yesterday's. This resulted in a very brief (couple of minute) Intent Bit switch at 17:40. It is back to Undisturbed.
ER8 Day 4
No restarts reported
At 17:07, I accidentally clicked a Guardian state request while doing some Dataviewer work. While this did not effect the state of the IFO (since there is no path out of Low Noise), it did trip the Intent Bit out of Undisturbed. During this brief downtime, Betsy and Stefan took the opportunity to update a few SDF diffs. We are back to Undisturbed as of 17:27. This is a good reminder of the power of a stray mouse click.
(Keita writing)
While Betsy was checking SDF status she found that some ASC signals were routed to ASC-DC5 only in YAW while output matrix elements for DC5 are all zero.
It seems to me that somebody used DC5 matrix elements to write down the previous SRC1 sensing matrix when they had to change it, knowing that DC5 is not used at LHO, and forgot to delete the "memo". Sensing matrix doesn't look like a good place to store this kind of stuff.
It is a very convenient method for looking at different linear combinations of WFSs in real time (e.g., with striptool) when trying to pick out a new error signal. But certainly we should clean up after ourselves when done.
8/21 OWL Shift: 7:00-15:00UTC (00:00-8:00PDT), all times posted in UTC
Handed an H1 (~55Mpc) which has been locked for 7.5+hrs from TJ with H1 Undisturbed & Observing.
NOTES (Beginning of shift with Locked H1) :
Activity Timeline:
Went through the SDF to see if I could make sense of the differences on there. I did not make anything GREEN (i.e. I did not ACCEPT or REVERT anything because I'm not sure whether we want Operators to do this or Detector Engineers to do that).
Here are some of my notes of Differences we have by subsystem:
SUSITMY
SUSITMX
SUSETMX
SUSETMY
SUSSR3
LSC
ASC
OMC
CALCS
Lots of channels here. Seems like many can be ACCEPTED
As usual, the SDF stuff just needed some investigation. During the last few Intent drops, I cleared some SDF stuff in Corey's alog 20736. Usually we don't have to go into such detail, but since there were a lot of question marks by each, I am adding a few notes of why things were accepted below.
Here's what I have found after hunting things down and double checked a few things with TEAM-COMMISS:
ACCEPTED New violin stuff SUSITMY
ACCEPTED New violin stuff SUSETMX
ACCEPTED New violin stuff SUSETMY
ACCEPTED This is the revert to turn the CAGE servo OFF and the OLDAMP servo ON SUSSR3
ACCEPT Tramps are not a big deal. If they are, then they are written into the Guardian and therefore can be ignored anyways. LSC
CLEARED ASC
ACCEPTED See attached snap - an additional stage of whitening is ON (the 2nd of 3), Stefan says this will likely stay on. OMC
OMC DCPD - In fact, we have now just set SDF to NOT MON these switches since Evan/Stefan say they will be switching whitening states during different lock stretches as needed, in Guardian.
8/21 OWL mid-Shift: 7:00-15:00UTC (00:00-8:00PDT)
Handed an H1 in Observing Mode by TJ (10+hrs of lock during current stretch). The range has been hovering around 55Mpc. L1 is now done with Commissioning at are at 65Mpc. All is quiet.
7:02-7:40: Darkhan requested to run a Transfer Function (DARM Matlab model), and since L1 was down for commissioning, I decided to let him proceed. In hindsight, I should NOT have allowed this activity since it is outside of the 1pm-10pm LHO Commissioning period. It was a judgement call I made since L1 was down. Range dropped down briefly to 10mpc during his sweep.
10:07: ETMy glitch VocalAlarm and observed on DARM spectra
Took DARM open-loop transfer function measurement using template:
CalSVN/aligocalibration/trunk/Runs/ER8/H1/Measurements/DARMOLGTFs/2015-08-17_H1_DARM_OLGTFtemplate.xml
As opposed to measurement taken on 2015-07-17 (see LHO alog #20614), we did not turn off calibration lines during this measurement.
Measurement XLM file was committed to SVN (@ r1097):
CalSVN/aligocalibration/trunk/Runs/ER8/H1/Measurements/DARMOLGTF/DARMOLGTFs/2015-08-21_H1_DARM_OLGTF.xml
J. Kissel, R. Savage, S. Karki, K. Izumi After considering the front-end precision issues found in the DELTAL_EXTERNAL calibration yesterday by Stefan and Evan (LHO aLOG 20700), we've agreed to move the pole frequency which reconstructs the integrator (a pole at 0 [Hz]) in FM1 of the L1 (UIM) ETMY control signal chain from what was quickly installed yesterday -- 0.001 [Hz] -- up to 0.01 [Hz] to gain another factor of ten high-passing of the control singal, such that we don't suffer from floating point single precision issues. Rick and Sudarshan will follow up shortly, repeating the study on the correct whitening filter we should use. Note: The means that the DARM calibration is invalid below ~ 0.5 [Hz]. If you need to have a calibrated DARM spectrum below there, then make sure you (offline) compensate for this. The history is as follows: - In ER7, we ran with this integrator compensator OFF, accepting that the front-end calibration below ~1 Hz was incorrect. LHO aLOG 17528, or LHO aLOG 18248 - A few days ago, Kiwamu updated all of the digital control filters to make that actual suspension filter chain using an automated script, forgetting about this dynamic range issue. LHO aLOG 20617 - Yesterday, Stefan Evan and Sudarshan rediscovered the issue, and moved the pole frequency up to 0.001 [Hz], and added more stages of whitening on the DELTAL_CTRL signal. LHO aLOG 20700 Today on the calibration call, Joe reminded us that he moves the compensation filter's pole up to 0.1 [Hz]. After comparing our options (in course 1 mHz, 10 mHz, 100 mHz increments), we decided that 10 mHz caused the least loss of phase in the 1 - 10 [Hz] region (where we need to worry about accurately reconstructing the PUM / UIM and PUM / TST cross-overs), for the most amount of high-passing of the large DC control signal.
Why would a 0.01 Hz pole make the calibration invalid below 0.5 Hz? At 0.5 Hz, the pole creates only a 2% amplitude difference. Maybe it's a typo, and you meant below 0.05 Hz?
The ASC control of the SRC hasn't been making sense. I did some calculations to see if the SRC might head towards an unstable condition if we run the IFO at high power (24W). We were interested in whether thermal lensing effects in the ITMs could cause a big enough gouy phase shift to cause us troubles.
At 24W there is 120kW of power in the arms and that the surface absorption of the ITMs is 0.5ppb (galaxy). The HR surface curvature of the ITMs changes by 1.06 udiopter/mW absorbed power (LLO alog 14634). This means we could expect a change in ITM curvature of 120e3*0.5e-6*1.06e3=65udiopters, or a change in the radius of curvature from 2000m to roughly 1750m as we go from cold state to 24W input power. I haven't considered substrate absorption, and G060155, slide16, indicates change in index of refraction due to bulk heating is roughly 10% of the surface absorption effect.
Putting this hot state ITM ROC into an aLaMode model of the interferometer (based on Lisa's model T1300960 with Dan's additions, alog 18915), the round trip gouy phase decreases from 32 to 22 degrees. Calculating the stability parameter for the SRC using the round trip transfer matrix, the cavity goes unstable. (A stable cavity occurs when -1<1/2*(A+D)<1. This parameter changes from 0.8 to 1.3.)
These numbers are just estimates, but the take away message is that as we increase the power, the SRC becomes less stable, and could be pushed into an unstable regime around about now. Turning on the ITM ring heaters would make the SRC more stable. (see comment) A ring heater on SR3 would also make the SRC more stable.
An additional note on transverse mode spacing: The SRC mode spacing is 240kHz in the cold state and 180kHz in the hot state. The cavity Finesse is 13.5, the free spectral range is 2.67MHz, so the full-width-half-maximum is 200kHz. So heating of the ITMs also pushes the SRC closer to degeneracy.
Paul F pointed out that I had neglected to look at the substrate lensing effect due to coating absorption. His calculations indicate this means that the SRC becomes more stable as we go to higher power. I will update my caculations too....
This is a follow up analysis of Robert's anthropogenic noise injections on August 12th. So far I don't see any convinving evidence that these activities were actually coupled into DARM. There were couple of injections ("human in optics lab" and "jumping in the change room") that seemed to coincide with DARM noise around 20-100Hz but it was unclear whether those injections were actually causing the noise. The noise *blobs* started prior to the injection and the PEM seismic sensor only coincided with one of them. Plus I would expect to see a coupling at much lower frequency if human jumping up and down the change room actually injects noise into DARM.
Note that the first injection time (15:05) is still unconfirmed. The time in the original table was wrong.
Adding SUS and SEI tags so I can locate entry. Seems like good news for the isolation crowd. (I note that just because a coupling only happens > 10 Hz does note mean SUS and SEI are off the hook. Robert and Anamaria have shown that loud drive at > 100 Hz can couple into HAM6 optics. So I would say this knocks SEISUS off the top of the list, but not off the list completely.)
Most injections lasted about 1s so the time series used for each tile should be about that long. These look like averages of time stretches almost an order of magnitude longer. I'll bet the signals will be more obvious if you zoom in in time on the individual injections instead of trying to see all 1s events in a singe 30 minute spectrogram.
I can see the injected signal clearly after I zoomed in. Below are the spectrograms for the truck horn injection. The signal can't be seen in the PEM-CS_MIC_LVEA_VERTEX channel. I'm working on the rest.
We're currently checking for DAC glitches, hopefully have an answer shortly
I checked for correlations of the glitches with the ASC error signals residual motion around zero, and with suspension witness sensors and optical levers. I couldn't find anything convincing.