For noise hunting reference, during the times below the IFO was locked at 20 W in "low noise"

 1223225342 - 1223225544

[Gabriele, Jenne]

I've noticed a few times today that when we reduce the 45MHz modulation depth, we see more buildup in both the carrier and the 9MHz in the PRC. 

We expect that, if we start with a modulation depth of 0.2 and reduce by 3dB, we should see about 0.5% 1% more carrier.  But, it looks like we're getting 2% more.  Also, I don't think we should expect any more actual 9MHz to be coming out of the PSL, since we're not changing that modulation depth. 

So, I wonder if we have an offset in one of our length loops that is using the 45MHz sideband, and when we reduce the modulation depth, we reduce the size of the offset.  Perhaps it comes from RAM, perhaps it comes from elsewhere?

Figure 1 shows coherence between the IMC DC WFS and DARM, with photos of the optic mounts that are producing the jitter peaks, based on this log: https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=42551 . The Qs are fairly high so damping should help. One tool that would be useful, is a small tuned mass damper, variable across the 200-800 Hz band, that can easily attach to a U200-A mount with a little clamp or epoxy. A possible clamping scheme is visible in the picture on the bottom left of Figure 1: I have used two commercial spring clamps near the adjustment screws to broad-band damp (via the mass and the red rubber grips) the U200-A mount. The spectrum shows that this and other damping has reduced the Q relative to the other two mounts, but a tuned mass damper on a clamp like this might be useful on the other mounts and possibly on this one. The jitter spectrum also has some fairly high-Q U100-A mounts (see above log entry).

Figure-2 compares DARM, IMC DC WFSs, coherence and table motion for Monday night and for O2. This is a bit premature since DARM and the WFSs are not yet well calibrated, and so I can’t make a definitive comparison of jitter coupling. But in this preliminary look, the change seems consistent with the table motion reduction being the biggest factor. The ~344 Hz peak seems to be a lot lower, and the table motion has decreased by 5 or more in that band. But the 483 Hz peak is in a band where the table motion has not really decreased (it was already low in O2 in this band, compared to nearby bands) and the 483 Hz peak does not seem to have improved by much in DARM. But, again, this is a first look; calibrations to come.

The second page of Figure 2 includes accelerometer signals when the water was off, showing that we could still improve by a factor of a few in the band around 500 Hz. We are planning to improve some water circuits during the vent. It might also be possible to reduce some flow rates.

I managed to run BruCo on one of the locks in "low noise" (GPS 1223100750 + 180 s). Results can be found here:

https://ldas-jobs.ligo.caltech.edu/~gabriele.vajente/bruco_1223100750/

Some comments (work in progress):

• As expected, coherence with MICH / SRCL / PRCL between 10 and 70 Hz (fig. 1-2-3)
• As expected again, some coherence with ASC loops, for example CHARD_Y (fig. 4)
• At 57 Hz, coherence with EY magnetometers (fig. 5). Also coherence at 114 Hz (harmonic?)
• The bumps at ~50 Hz and ~100 Hz are coherent with REFL_A_RF9 (fig. 6) and OMC_A_PIT (fig. 7)
• ...

Note: due to a recent "upgrade" of the clusters, the "pylal" library (that BruCo used to access data frames) does not exist anymore. ldas-pcdev12.ligo.caltech.edu has been "downgraded" by reinstalling the library there, so that's the only machine where you can run BruCo for the moment being. I tried to port BruCo to gwpy, and failed: the processes eat up all the host memory and make the machine unresponsive.

E. Goetz, J. Kissel,

 Evan and I installed many front-end model changes in support of O3-level calibration pipeline infrastructure. Apologies for the delay in the actual "we did it!" log.
This was briefly touched on in Dave's CDS report yesterday, (LHO aLOG 44440), and documented in Evan's Monday aLOG in prep for it, (LHO aLOG 44418), and mine before that (44319). The changes are re-stated (with one addition and a few edits) from Evan's aLOG listed below. 

We are actively and furiously drawing MEDM screens (most of yesterday afternoon and evening, and will continue today) in order to reflect the changes so we can make the new infrastructure functional, so please have patience with us while we work through that to find bugs and errors.

Yes, we have changed several things including the DAQ channel list, and calculations of time dependent correction factors so we necessarily expect that down stream products will be broken for a bit (e.g. LHO aLOG 44454) until we fully integrate these new changes.

We'll be regularly updating G1801594 and T1700106 in order to reflect the changes we've made as we've been hit in the face with the reality of actually implementing all this new stuff.

 (most repeated here for your convenience, with some additions):

  1) Since LHO alog 44319, in CAL-CS the Library Parts -- the meat of the changes -- we have:
/opt/rtcds/userapps/release/cal/common/models/CAL_CS_MASTER.mdl
/opt/rtcds/userapps/release/cal/common/models/CAL_LINE_MONITOR_MASTER.mdl
     - completed the front end computation of the detuned SRC spring frequency and Q (f_s and Q) because we changed the tracking of PUM and UIM stage from a common time dependent scalar to be separately tracked as two time dependent scalars
     - implemented parallel path calculation for on-the-fly compensation for a time-dependent cavity pole frequency
     - added excitation point in QUAD model parallel to the calibration lines so that we make excitations for sweeps at the same point as for the calibration lines that are injected. This allows for easier removal of time-dependent changes in actuation gains so that we can search for unknown systematic errors

  2) CAL-CS top level model changes:
/opt/rtcds/userapps/release/cal/h1/models/h1calcs.mdl
     - Removed DARM_ERR_WHITENED and DARM_CTRL_WHITENED since these are no longer actually whitened and instead are double precision channels
     - Added an IPC receiver from each end station TX PD to the top-level CAL-CS model and piped it in to the CS library part
     - Added new CFTD_DELTAL channels to the frames (CFTD = compensated for time dependence) as parallel channels to the standard DELTAL channels
     - Deleted DARM hardware injections in top level CAL-CS. These were unused
     (Note removing whitened channels and DARM hardware injections saves at least five 16k channels from being saved to frames. We added only two 16k channels and four 4k channels)

  2) PCAL library changes:
/opt/rtcds/userapps/release/cal/common/models/PCAL_MASTER.mdl
     - Removed oscillator line filters because they were unused
     - Removed RX and TX_PD_VOLTS filters and replaced with EPICS and testpoints
     - Implemented computation of optical efficiencies and force coefficients as laid out in T1800046
     - Allowed for different ADC gains and different analog AA filter gains because we care about this in the h(t) calibration at the <0.1% level

  4) In top level PCALEX / PCALEY model:
/opt/rtcds/userapps/release/cal/h1/models/h1calex.mdl
/opt/rtcds/userapps/release/cal/h1/models/h1caley.mdl
     - added IPC senders of TX PD signals to go to the CAL-CS model

  5) In the QUAD_MASTER and QUAD_ITM_MASTER.mdl parts, and associated L1 & L2 library parts,
/opt/rtcds/userapps/release/sus/common/models/QUAD_MASTER.mdl (for the L3 stage)
/opt/rtcds/userapps/release/sus/common/models/QUAD_ITM_MASTER.mdl (for the L3 stage)
/opt/rtcds/userapps/release/sus/common/models/FOUROSEM_STAGE_MASTER_OPLEV_TIDAL.mdl (for the L1 stage)
/opt/rtcds/userapps/release/sus/common/models/FOUROSEM_DAMPED_STAGE_MASTER_WITH_DAMP_MODE.mdl  (for the L2 stage)
     - added excitation point right next to where the calibration line is injected to facilitate compensating frequency-dependent sweeps with time-dependent correction factors.


Again -- presentation given to the calibration group (see G1801594) will be updated to reflect these changes.

Craig, Sheila, Jenne, Gabriele

We made enough progress tonight that we should be ready for calibration sweeps tomorrow.  The guardian as the graph is now should be able to take you to LOWNOISE_ESD_ETMX.  

• Violin modes
  • During the maintence period today the violin modes were rung up.  I edited the guardian to slowly increase their gains as they damp if the gain setting is lower than the max_gain listed in LSC params.  
  • ETMY mode 6 had the gain set to 0 in the guardian, but it was rung up tonight and a gain of -6 was fine, so I've put this in the guardian. 
• The first time we transitioned to the ITMX ESD, a mode at 6049.73Hz rang up and broke the lock.  We modified the notch in DARM for the 4.7kHz violins to also include a notch around 6050Hz.  Craig made a monitor for this mode in ITMX_L2_DAMP_MODE10, it rang down very quickly on it's own so it's not a violin mode (spectrum of 64kHz DCPD channel is attached).
• The 45 MHz reduction that Jenne worked on yesterday is in the guardian path now, before we increase the power. 
• We had several locklosses in a row caused by rough engagement of SRC2.  Craig reordered the SRC_ASC state so that the gain is ramped on. 
• We attempted once to go to 25W of input power, we still had a large drop in POP18 power, and after a few minutes at 25W we lost lock (You can find this in the new lockloss tool which is still being developed here: lockloss).  Yesterday we reduced the POP45 whitening gain by 12dB, I reduced it further after that lockloss, but that may be unnecessary.  
• We switched back to the ETMX ESD in low voltage mode.  We used a gain of -40 in the drivealign filter bank to roughly match the gain of the ITM ESD.  The second attached plot shows a comparison of the response of DARM to excitation at drivealign for both ITMX and ETMX after switching to the low noise driver.  The third attachment shows a quick DARM OLG measurement. 
• We edited the guardian state ESD_ITMX (renamed LOWNOISE_ESD_ETMX) so that it transitions DARM control to ITMX, switches the EX ESD to low voltage mode, and transitions DARM back to ETMX ESD, leaving L1+L2 on ETMY. There were several edits to PREP_FOR_LOCKING and ALS_DIFF needed to be able to relock after this. 
• I unlocked the interferometer using the MICH_OLG_NOISE_FULL_LOCK template at 2018-10-10 07:23:45 UTC

[M. Wade, A. Viets]

I restarted the primary and redundant calibration pipelines on DMT0 and DMT2 at roughly GPS second 1223179096. This restart was necessary because of a change in channel names in the raw frames. The EPICS channels (which contain reference values needed to compute time-dependent correction factors) were renamed.

In order for data flow to be restored, the LHO_DMTDQ process needs to be restarted as well. None of the the kappas are being computed, so these channels which are usually present are not there:

• GDS-CALIB_KAPPA_TST_REAL
• GDS-CALIB_KAPPA_TST_IMAGINARY
• GDS-CALIB_KAPPA_TST_REAL_NOGATE
• GDS-CALIB_KAPPA_TST_IMAGINARY_NOGATE
• GDS-CALIB_KAPPA_PU_REAL
• GDS-CALIB_KAPPA_PU_IMAGINARY
• GDS-CALIB_KAPPA_PU_REAL_NOGATE
• GDS-CALIB_KAPPA_PU_IMAGINARY_NOGATE
• GDS-CALIB_KAPPA_C
• GDS-CALIB_KAPPA_C_NOGATE
• GDS-CALIB_F_CC
• GDS-CALIB_F_CC_NOGATE
• GDS-CALIB_F_S
• GDS-CALIB_F_S_NOGATE
• GDS-CALIB_SRC_Q_INVERSE
• GDS-CALIB_SRC_Q_INVERSE_NOGATE

For now, it is not possible to compute the time-dependent correction factors (kappas), so the pipeline was restarted with the calculations turned off. A few updates will be necessary in order to resume calculating the kappas:

• The new channel names need to be added to the low latency raw frames.
• The software on the production DMTs needs to be updated to gstlal-calibration-1.2.1, after testing of the software on the testing DMT is complete.
• The new calibration line frequencies need to be chosen and the lines turned on.

I wanted to try adding the "jel" packs that Arnaud and Eyal have been using at LLO (for instance, here ), because it seems to help the low frequency noise. Jason had some left from the oplevs (excess beyond 3IFO needs), so I tried adding some the ETMY BRS, see attached photo. In the process, I must have bumped the "mass adjuster" in the process, because when I checked the BRS status before leaving the end station, it was dead, out of range. So, I set about rebalancing, and found that the rdesktop program (which I had aliased) with the new zotac workstation was laggy and basically useless. After talking to Carlos, I tried using remmina, which was much better, but still laggier than rdesktop was on the old mac/debian workstations. When I left the BRS was still rung up and not ready to use, but was slowly settling down. With the low microseism, this shouldn't interfere with commissioning, so I don't want to risk trying to settle it by hand, yet. If it's still angry tomorrow, I can try soothing it.

We had at least one lockloss tonight when a violin mode damping filter was reset without having the gain be 0.   Looking at the gaurdian logs, there were both ezca connection errors and same state redirects, although the ezca connection error seems to be the culprit here. 

The log is attached, the reason we lost lock is that the filter states for ITMX_MODE7 were reset in the main state (with the bandpass filter momentarily turned off, which is fine if it happens only once when the gain is set to zero) (lines 47-52).  Then the gain is ramped on line 76.  On line 181 there is an ezca connection error, this is the kind of error we have been having frequently the last few months, bugzilla 1122.

The main state then repeats but isn't idempotent which unlocks the IFO. 

Craig edited the code so that the bandpass filters are no longer momentarily turned off, so this wouldn't be as bad if it happens again.  

The daq was restarted around 5:30pm to pick up a change to h1calcs.

ITMX bio has the same problem we had before. 44046

TJ, TVo

Daniel suggested that we try swapping the cameras with any spares we have in order to see if the bad pixel problem goes away.  From the user manual, it is not clear what happens when a pixel "malfunctions' on the CCD so this may require some more testing.

TJ and I found two spare cameras in the cabinet, one is labeled SN 008 and the other is unlabeled.  The original installed camera for ITMX HWS was SN009.  We first installed the unlabeled camera and the noise was worse with more bad pixels, then we installed SN008 and saw that there were less bad pixels so we left that one in. 

The stationary (no transient heating) noise in the spherical power seems to have decreased with this new installation with the variance being about 3 udiopters over the course of half an hour (time series attached).  Also attached is a gradient plot overlapped with the raw CCD images with the plates on and off. There are one or two gradient vectors which stand out in the upper half of the images so it still behooves us to try and implement a bad pixel finder which eliminates these bad data points before going into the HWS code which fits the spherical power. 

Marc, Daniel, Nutsinee

We have to much gain at the output of the LO common mode board. The box has -41.5 dB gain up to 10Hz and -57dB gain above 60Hz. The box was installed last week. Combine with -30dB gain at the common path slide bar gives us UGF at 1kHz.

TITLE: 10/09 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Preventive Maintenance
INCOMING OPERATOR: None
SHIFT SUMMARY:

• Maintenance started roughly at 13:15 (for PSL) & ended around 19:30utc (i.e. 6:15am-12:30pm).
• BRSy was bumped & it looks like it will need time to damp out on its own, so we have the ISI CONFIG in the "Windy No BRSy" state.

Ops LOG: (text file is attached)

IFO locked at 10 W, moved CSOFT and DSOFT to maximize the recycling gain. Moved also SRM and SR2 to keep good sideband buildups. New offsets in the TRANS QPD:

caput H1:ASC-X_TR_A_YAW_OFFSET -0.237102
caput H1:ASC-X_TR_B_YAW_OFFSET -0.133515
caput H1:ASC-Y_TR_A_YAW_OFFSET +0.279309
caput H1:ASC-Y_TR_B_YAW_OFFSET +0.292131
caput H1:ASC-X_TR_A_PIT_OFFSET -0.019295
caput H1:ASC-X_TR_B_PIT_OFFSET -0.022531
caput H1:ASC-Y_TR_A_PIT_OFFSET +0.061096
caput H1:ASC-Y_TR_B_PIT_OFFSET +0.149773

Peter King, Marc Pirello

In an effort to reduce RF noise in the PSL racks, we modified the AOM VCO Amplifier (D1201423) by removing the anodization on the front and rear panels to create a better RF "seal" on this chassis.  Though not spectacular, they were measurable.  79MHz RF leakage magnitude near this panel was -37.8dB prior to the installation (S1203924), and post installation (S1203923) we recorded -44dB, roughly 6.2dB difference.

Tuned up the alignment into the reference cavity.  The transmission improved from ~2.2V to ~3V.


    Removed the DC block from the AOM (see attached photo for the aftermath).

    Measured the transfer function with the current gain slider settings.  The UGF was ~452 kHz
with about 60 degrees of phase margin.  Two measurements ar attached, one out to 2 MHz and the
other out to 10 MHz.  No big changes were observed.

Craig, Sheila

We have glitches in CARM associated with the discrete stepping of our analog gain sliders while we power up.  These are unavoidable, since we must compensate for the increase in optical gain by decreasing the analog gain by an equal amount, and we cannot use digital power normalization like for other LSC.
In order to reduce the size of the glitches, we have moved the stepping stage downstream, from LSC-REFL_SERVO_IN1GAIN to IMC-REFL_SERVO_FASTGAIN and LSC-MCL_GAIN.  

Our current gain settings are not optimal, but the glitches are lower than before: Sheila will post a follow-up confirming this.

We did this by changing what sliders the ISC_library.py function IMC_power_adjust_func() adjusts.  I also adjusted IMC_LOCK so the IMC locks with IMC IN1 at 25 dB now (this is increased by 3 dB from usual, another alog is in the works).  In any case, the guardian works now for both locking the IMC and powering up.

There are a total of 7 relevant gain sliders in the analog CARM path, along with two compensation filters (from the awiki pole/zero -- IMC: 40 Hz / 17 kHz, LSC: 40 Hz / 4 kHz ) and one boost (1 kHz / 20 kHz).  
Assuming these filters have a gain of 1 at AC, then at DC we get gains of:
IMC Comp DC Gain = 52.56 dB
IMC Boost Gain = 26.0 dB
LSC Comp DC Gain = 40 dB


CARM Gain Slider Values at 2 watts (at the end of Guardian states RESONANCE, CARM_TO_ANALOG, and ADJUST_POWER [at 10 watts])

ISC_LOCK State    SUM NODE A IN2    LSC IN1      LSC MCL GAIN    LSC FAST    IMC IN1    IMC IN2    IMC FAST   |  CARM FAST OVERALL    CARM SLOW OVERALL 
-------------------------------------------------------------------------------------------------------------------------------------------------------
RESONANCE                   2 dB     -16 dB     240 (47.6 dB)        7 dB      25 dB      -6 dB        0 dB   |             -13 dB              33.6 dB
CARM_TO_ANALOG            -12 dB      11 dB    53.8 (34.6 dB)        0 dB      25 dB     -12 dB        0 dB   |             -13 dB              33.6 dB
ADJUST_POWER [10 W]       -12 dB      11 dB    10.2 (20.2 dB)        0 dB      25 dB     -12 dB      -14 dB   |             -27 dB              19.2 dB


CARM FAST OVERALL = SUM NODE A IN2 + LSC IN1 + LSC FAST + IMC IN2 + IMC FAST 
CARM SLOW OVERALL = SUM NODE A IN2 + LSC IN1 + LSC MCL GAIN 
Neither "OVERALL" gain includes the gain of the filters.