Looking at the ALS fiber distribution box:

• The external PD is saturating, needs an ND filter with a factor of 2-4 attenuation.
• The internal PD gain is set at 0dB, but should be 30dB (this also requires adjustments to the lower limit and normalization).

We have 15-20mW into the fiber. With a ~7% splitting ratio (the losses through the AOM are about 3 times higher than specifications would indicate), we will have about 1mW in the SQZ LO fiber output. This is in line with the L1 measurement in alog 28961.

The first plot attached shows 24 hours of Corner3 and other CPS In1s along with Lock state signals.

The two large spikes in the first half of the trend shows saturating spikes on the Corner 3 CPS and concurrently the BS Watchdog tripping.  These glitches only show on C3 and are on all four C# CPSs.  Not shown are earlier locklosses and BS Trips that JimW in aLog 37499 attributed to Corner3 glitching.

First action Jim performed was power cycling the satellite racks at the chamber.  Jim did this by unplugging the rack power cord.

After the first LL on this log's plot, Jim un- and re-seated the gauge boards in the satellite racks.

After the second LL at ~1740pdt, I power cycled the C3 interface chassis and reseated and tightened the power cord of the chassis.

There is a final lockloss on this plot around 6am pdt but there is no to little spiking of the CPS channels and certainly nothing of the WD tripping saturations seen earlier.

So, the 6am lockloss was not the CPS but only time will tell if things are completely cleared up.

I still plan to look closer at the glitching and locklosses here so there may be more.

Attached are 3 day trends of the laser head flow rates in the PSL HPO.  So far everything looks to be holding steady; the jagged-looking nature of the signals at the beginning of the trends is the tail end of the flow rate tests performed by Peter and Jeff on Tuesday.  Will continue to keep an eye on these flows.

TITLE: 07/14 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Earthquake
OUTGOING OPERATOR: Travis
CURRENT ENVIRONMENT:
    Wind: 4mph Gusts, 2mph 5min avg
    Primary useism: 0.01 μm/s
    Secondary useism: 0.07 μm/s

Quiet at the moment with small useism & low winds.
QUICK SUMMARY:

Richard and Travis were troubleshooting IMC when I walked in.  Richard asked me to run the DOWN state (Guardian was in the INITIAL_ALIGNMENT state), and this locked up the IMC.  Will start an initial alignment from scratch now.

Note:  OBSERVATORY MODE is in the EARTHQUAKE state.  It was taken to OBSERVING last night.

TITLE: 07/14 Owl Shift: 07:00-15:00 UTC (00:00-08:00 PST), all times posted in UTC
STATE of H1: Earthquake
INCOMING OPERATOR: Corey
SHIFT SUMMARY:  After the lockloss (still not convinced it was due to the EQ), I started an IA since the AS spot looked iffy and DRMI was not locking.  When I got to Input_align, the IMC lost lock and would not relock.  I followed the procedure in the troubleshooting wiki and recovered flashing in the IMC, but it still would not lock.  I started looking around at the MC SUS screen and found a channel that had a huge value coming into it.  Thinking it might be electronics, I called in Richard to have a look.  At the moment, he is still poking through MEDM screens to try to find the source of the signal. 
LOG:  See previous aLogs.

I set the OPS_OBSERVATORY-MODE to Earthquake, but I'm not entirely certain that was the culprit.  We were at the peak of the motion from a 5.3M EQ in Chile, but the BLRMS top out at only ~0.1 um/s, so well below the typical lockloss threshold.  Jim's StripTool shows some wiggles in the BS CPS traces, but not what I'd call glitches.

For the SEI team, I attach screenshots of the USGS, Terramon, and SEISMON 5-EQ screens that show that SEISMON is not reporting the latest EQs (there have been several EQs over 4.0M with dates of 2017-07-14 that don't show up on SEISMON). 

Called LLO control room to verify they received the alert.  Jeremy at LLO verified they did, and logged back in to Teamspeak since their Teamspeak machine had rebooted.

All appears to be well.

TITLE: 07/14 Owl Shift: 07:00-15:00 UTC (00:00-08:00 PST), all times posted in UTC
STATE of H1: Observing at 53Mpc
OUTGOING OPERATOR: Patrick
CURRENT ENVIRONMENT:
    Wind: 5mph Gusts, 3mph 5min avg
    Primary useism: 0.01 μm/s
    Secondary useism: 0.06 μm/s
QUICK SUMMARY:  We are finally back to Observing even with the range still a bit lower than we'd like.  I'll keep an eye on the BS ISI CPS for glitches if we have any locklosses.

TITLE: 07/13 Eve Shift: 23:00-07:00 UTC (16:00-00:00 PST), all times posted in UTC
STATE of H1: Observing at 54Mpc
INCOMING OPERATOR: Travis
SHIFT SUMMARY:

I spent the beginning of the shift trying to help Jeff and Hugh damp violin modes. I worked on adjusting the gains for the first fundamental modes 2, 4, 6 and 8 of ETMY. Increasing the gain on modes 4 and 6 did not seem to have any effect on the ~508.219 Hz mode. Increasing the gain on modes 2 and 8 may have helped damp the ~507.992 Hz and ~508.66 Hz modes respectively, or they could have just been coming down regardless, it was unclear. Thomas, Pep and Jenne worked on moving the beam spot positions but reverted their changes before they left. We had another lock loss attributed to the BS ISI CP glitching. Hugh made another change to try and address it (see his alog). We can now make it to NLN without intervening in the violin mode damping and without skipping any states. After consulting with Vern we set the intent bit to observing despite the range of ~54 MPc.

LOG:

23:55 UTC GRB verbal alarm. LLO not on teamspeak. Ignoring.
00:42 UTC Lock loss. BS ISI CP glitch. Thomas and Pep reverting beam spot moving ASC changes.
00:44 UTC Hugh to CER to powercycle BS interface chassis.
00:48 UTC Untripped BS ISI. Hugh back to CER to tighten screws.
01:00 UTC Thomas and Pep done. Relocking.
01:36 UTC NLN.
02:01 UTC Observing after consulting with Vern.

Damping of the baffle has been shown to have successfully reduced noise in DARM from local vibration injections around the input beam tube (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=36979), but an improvement for site-wide vibration has not been shown until now.

While we have been down for some time following the Montana quake, there was enough data before the quake that I could look into indicators of site-wide coupling. The evidence here suggests that the baffle damping also reduced coupling of global 10-30 Hz vibrations and thus supports the hypothesis that the baffle was the dominant coupling location on site for this band.

The rush hour range drop was thought to be produced by scattering from the Swiss cheese baffle (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=35735), and it was thought that damping might reduce this nearly daily drop in range.  Each of the panels in Figure 1 show the 10-30 Hz seismic band and the inspiral range. The rush hour traffic features are the broad ~gaussian peaks (from many cars) at the beginning and end of the work days. The panels from April and last December show that the range dropped during each rush hour (unless the range was already low), and that this has been happening for the entire run. The recent plot does not show similar range drops, suggesting that site-wide coupling in this band has been substantially reduced (as an aside, we do still often loose lock during the morning rush hour, even though the range doesn’t drop, and this bears further investigation).

To test for a global effect when the traffic noise was not elevated, I used Jenne’s jitter subtraction code and compared non-rush hour times before and after the vent. I took 1024 second stretches that were matched in original inspiral range and applied Jenne’s code. Figure 2 shows that the range with subtraction was, for each pair, higher after the vent. The difference was only a few percent but is significant (at p<0.05 ). Because the scattering noise is highly non-stationary, damping may also have improved stationarity (compare minute-scale range variation in Figure 1).

After consulting with Vern, we (Jeff, Hugh and I) set the intent bit to observing at 02:01 UTC. The range is around 53 MPc.

Pep C. Thomas V. Jenne D.

After heroic efforts to successfully dampen the violin modes and get to NLN, the BNS range sat around 55 MPC so Jenne thought it would be a good to optimize the beam position in the arm cavities to match the pre-Montana-Earthquake.

This is done by first changing the A2L gain coefficients to what they were on June 17th 02:00:00 when Sheila was able to achieve better BNS range, and then turning on the dither scripts that show the length to darm coupling found in userapps/isc/common/scripts/decoup/setA2lines.py.  We then adjust the CSOFT and DSOFT loop offsets to minimize the dither resonances, which effectively moves the spot position back to what Sheila's settings were in mid-June.  After doing this, there seems still be the same amount of jitter noise at low freq ~10-80Hz.

However, it seems that the input pointing from the IMC has also shifted since the earthquake according to Cheryl's ALOG and we were messing with the IMC loop DOFs to adjust the WIT sensors back to their pre-EQ values, this proved to hurt our sensitivity a bit so we might need to spend more time fine-tuning this approach.

Because we haven't really improved the sensitivity, we changed the A2L gains back to their values on July 12th 01:06:22 and going to a more stable lock state.

Still drilling into the fine print but this still holds although there may be more details to uncover.

After this glitch, I power cycled the sensor chassis in the CER--this did not include the T240s.  After doing so I noticed the power cable to the Corner3 chassis was not screwed in and the cable was somewhat tight and the connector did not seem square/fully seated.  I pulled a bit of slack in, seated and screwed down the connector--we'll see...

C. Gray, J. Kissel, R. McCarthy, H. Radkins, P. Thomas

We now believe we've confirmed the functionality of all fundamental violin mode damping for all 32 modes of the 4 test masses. As such, I've reconciled the VIOLIN_MODE_DAMPING guardian state, and updated the Violin Mode Table such that all agree with each other and what has been confirmed successful. I couldn't keep track of every mode that had changed, but I'd say it was about 5 modes out of 32 that have changed its settings (and not restricted to any particular test mass, the ITMX has been relatively trouble free). Our latest theory is that the fiber mode shapes and/or their coupling have changed as a result of the earthquake, and thus they require new actuation magnitude / phase.
 
While we only have statistics of 2 lock acquisition sequences, we've been able to successfully turn on automated turn-on of these damping loops as per normal.

We're still having trouble completely squashing all fundamentals (namely EX MODE4 at 505.805, and EY modes 3-6 around 508.2 are still interesting), but they're at least stable.

In our spare time, we're slowly crushing the 2nd harmonics (stuff that's not coded into guardian because it's not typically problematic, but still rung up from the EQ), and updating the reference table accordingly. This is limited by the number of filter banks and our patience, but we're otherwise successful. After O2, we should definitely increase the number of filter banks on each suspension.

Notes:
- this further work on violin modes has not improved the sensitivity. 
- this work has not been impacted by the IFO or IMC beam spot restoration / moves / tests that occurred concurrently.
- just about all lock losses today are from the Beam Splitter ISI's CPS glitching.

As was alluded to earlier by Corey, it looks like the CPSs on BSC2 started glitching this morning and caused the ISI to trip, breaking several locks. Attached plot shows several hours around the locklosses this morning. The windows where the ST1 WD was at state 4 are when the WD tripped. The long fuzzy period on the left of the CPS plots are the earthquake this morning, only ST1 H3 & V3. It looks like the CPS started glitching some time after the earthquake. I went out, pulled and reconnected the power on the satellite racks on the chamber. I've had a striptool running of the CPS since power-cycling and it looks like the glitching has stopped. Given the glitches start before getting bad enough to trip the ISI, maybe we can come up with some of diagnosing before this becomes a problem, so we at least know what we have to fix. I don't know what that would be though. We could at least get longer trends (an hour or two) of CPS if a "mysterious" ISI trip is suspected of causing a lock loss.