J. Kissel

We restarted the h1hwinj1 and h1hwinj2 injection machines today to check whether they were contributing to the recent EPICs freeze problems we've been seeing on control room work stations. The reboot did NOT help the problem. However, since they've now been restored, I figured I'd try my luck at using Eric's instructions on how to restart them (LHO aLOG 18831, T1400349). 

While I was able to log on to h1hwinj, find the appropriate directory, and start_psinject, after the 60 sec start up time, I did NOT see the injection start (looking at the excitation channel in the H1:CAL-INJ_CW_EXCMON or H1:CAL-INJ_HARDWARD_OUTMON), nor does the CAL-CS GDS_TP sceen indicate that there has been a test point selected, even though the interaction with start_psinject was as follows:
-bash-4.1$ ./bin/start_psinject 
Currently, we are running on: h1hwinj1.cds.ligo-wa.caltech.edu
Result of checking injection status: H1 injection is NOT RUNNING
  Automatic restart should occur within 10 minutes; no human action is required
H1 injection will be started with configuration ER7 if you proceed

For logging purposes, please enter your name     (or Ctrl-C to cancel)
Jeffrey Kissel
Type a BRIEF explanation (a few words) for why start_psinject is being executed
Hardware Injection Machines were restarted today to see if they were affecting EPICs slowdown seen in the control room. See LHO aLOG 19809.
Starting H1 injection into H1:CAL-INJ_CW_EXC on h1hwinj1.cds.ligo-wa.caltech.edu using ER7
exec /data/scirun/O1/HardwareInjection/Details/bin/exec_psinject_from_script /data/scirun/O1/HardwareInjection/Details/pulsar/ER7 /ligo/apps/sl6/gds-2.16.12 H1 H1:CAL-INJ_CW_EXC 1121573769
Injection will start at t=1121573769 (60 seconds from now)...
Injection starting now
-bash-4.1$ 


Also -- while the hardware injection machines were down, there was NO indication on the injection overview screen that the hardware injection machines were down / dead. In fact, since the EPICs records that *are* shown on the screen are all defined and stored in the front-end, they had been left in the ALL GREEN state making it deceivingly look like all is well. There should be a look-alive heartbeat EPICs variable sent from these machines, or at least a static alarm bit somewhere to indicate that these injection machines are down or up.

J. Kissel, B. Weaver, N. Kijbunchoo, T. Sadecki, E. Hall, S. Dwyer

Another heavily loaded Tuesday today. For what we planned to do, see LHO aLOG 19770. We have managed to complete the IFO recovery, on the same day, and before the sun has gone down!! Most of the failures and deviations from plan today were software problems, which didn't fix the problem they'd intended to fix, were slow to recovery, or had to be reverted because of new found bugs. Towards the tail end of recovery, there were other problems that were more related to half-finished or mistakes from commissioning last night, but thankfully by the time we had reached recovery the guilty commissioned had arrived on site and confessed their sins. In summary -- not so bad -- but there was little IFO systems' software actually restarted. 


(All times PDT)
07:00 Richard brings SEI platforms and SUS to SAFE/OFFLINE LHO aLOG 19778

07:30 Richard, Daniel and Fil swap the internal GPS of the timing master for the external GPS unit. LHO aLOG 19782
      This did *not* crash any front-ends, so the corner station models have stayed up all day, and were otherwise unaffected by the rest of maintenance. This was a huge relief.

08:00 Richard and Fil continue moving GPS Antennae on roof, no further impact on the timing system
     Hugh and Jim W begin ramping down HEPI for pump maintenance
     Jim B heads down to EY to begin timing fanout replacement and SUS fast front-end BIOS upgrade

09:15 Begin corner station SEI / SUS and IMC recovery
     A little bit of trouble bringing the IMC up because the WFS output had not been offloaded in a while, so with the initial alignment the WFS trigger PD had a little bit too low of light. We've since offloaded the IMC WFS DC values to the alignment sliders to alleviate this, LHO aLOG 19814 
     HEPI HAM2 trips during recovery, HAM5 has he usual problems with Rogue Excitation vs. Watchdog RESET, but otherwise a smooth restart of the remaining SEI and SUS.
     Vinny starts PEM sensor calibrations. 

09:30 IMC recovery complete
      Jim's finished with EY Timing fanout and SUS fast FE, find that the BIOS upgrade makes no difference to ADC Timing Errors, but moves on to EX.
      Cable runs for PEM cosmic ray detector begin. LHO aLOG 19804

10:00 Betsy and Jeff begin EY SUS ETMY's RCG upgrade to get all improvements to TrueRMS and SDF sort-on-substring, discover that branch 2.9 (which has all improvements to TrueRMS) has some terrible BURT/SDF restore bug. End result is to compile against RCG 2.9.5 LHO aLOG 19793
      Hugh and Jim finish repair of HEPI pump station, and it's brought back in loop. LHO aLOG 19794

10:15 TCSX Laser trips from PEM cable pulling, recovered shortly after by Nutisnee. Ha! We knew to expect this one! LHO aLOG 19786

10:30 Huh discovers EY HEPI fluid level is significantly lower than last week, suspects accumulator bladder has let in fluid. Executive decision is to wait until next week to assess and repair. LHO aLOG 19796.
      EY HEPI Trips
      Fil and Andres head to EX to being LVLN ESD driver install and cable pulling. The new ETMX ESD LVLN Driver was literally, merely installed in the rack, but has not been hooked up. LHO aLOG 19803

11:00 Discover that EY SUS has IOP DAC output issues after finished of recompilation against RCG 2.9.5, requires full front-end shut down and restart (including TMSX and new EXPI model) LHO aLOG 19793
      Jim B. is done with BIOS upgrade for EX SUS fast-front end. Also no improvement in timing errors.

11:15 EY front-end model recovery complete
      EY SEI / SUS restoration begins
      SEI system up and running, SUS restared to damping and aligned, charge measurements begun to assess health.

11:20 EX front end model restoration begins, encountering the same problems with no IOP DAC outputs. Restart entire front-end, like EY.
      Begin EX SEI / SUS restoration begins 
      Charge measurements on ETMY ESD complete, confirm charge is OK and little no change in charge since yesterday LHO aLOG 19764
      EY recovery complete.

11:40 ISC EX and ISC EY front-end epics settings restored to 05:00
      SUSAUX model updates installed, having been compiled against RCG 2.9.5 LHO aLOG 19780      
      Preventative maintenance reboots begin
            Conlog computer restarted, no problems LHO aLOG 19789

12:00 MX / MY PEM model changes installed LHO aLOG 19809

12:10 DAQ / Frame Builder restart, doesn't come back gracefully, needs a lot of prodding from Jim and Dave. 

12:20 Frontend network switch restarted, does not fix EPICs slowdown issue
      EPICs Gateway restarted, does not fix EPICs slowdown issue
      Hardware injection machines turned OFF, does not fix EPICs slowdown issue

12:45 Guardian machine restarted. Doesn't come back up on first try, does better the second time LHO aLOG 19812
      Attempt to begin ETMX charge measurements to confirm health, but EX Beckhoff PLCs crash, not necessarily unexpectedly. After restoration, settings are restored to 05:00
      Beckhoff crash trips ETMX ESD HV Driver power supplies, and renders Beckhoff remote restart useless. 

13:00 Fil drives to end station to turn ESD power supplies back ON.

13:15 Charge measurements on ETMX ESD begin, only one data point, but confirms EX ESD is functional
      EY recovery complete
      Begin Full IFO Recovery

13:45 Initial alignment complete for green and PRM. recovery up to this point was MUCH more smooth this week

14:00 Discover that certain visiting commissioners have removed vital MICH and SRCL lock acquisition filters last night. Filters restored from archive, and we move on.

14:20 Initial alignment complete for IFO
      Begin lock acquisition sequence

14:50 Reach DC Readout, increase to high power, but lose lock instantly upon transition from ETMX to ETMY as DARM actuator

      An ETMY ISI trip, later associated with IPC communication problems because of ADC timing issues from SUS front-end. LHO aLOG 19808, LHO aLOG 19799
      several more lock-losses, a few on transition to TR CARM, a few more on transition to ETMY

16:45 Discover that certain visiting commissioners have left the ETMY hierarchical control filters in a half-commissioned state last night. Filters restored from archive.

17:00 Frame builder crashes again. LHO aLOG 19811.
      Evan notices that Trans QPD dark noise for X arm is huge. While the IFO continues to lose lock on TR CARM transition, Evan and Shiela investigate what possible QPD offset settings might have been lost.

18:00 Shiela just gives up, assumes that the dark noise of the QPDs has actually changed (perhaps because of ETMX electronics work earlier in the day)

19:40 Full IFO recovered to a stable 60 [Mpc]!

J. Kissel, K. Izumi, J. Driggers

This morning, we had a little trouble relocking the IMC because the initial alignment of the IMC SUS causes the light on the IMC WFS trigger PD to be just a hair too low. To solve the problem this morning, Jenne temporary decreased the trigger threshold (H1:IMC-IMC_TRIGGER_THRESH_ON) from 40 to 15 [ct], such that the IMC WFS engaged, which steered the IMC to our known good alignment (and then restored the trigger threshold once the WFS had steered the optics back to a good enough alignment that the normal trigger threshold was easily surpassed). To avoid this in the future, we said "we should offload the DC value of IMC WFS control signal to the MC SUS alignment offsets," but at the time didn't know that was the right script / button / thing to call in order to do so since there have been many in the past which had suffered from bit rot. 

Just now, in between lock attempts, I've asked Kiwamu what the right script is, he should me, and I ran it. This changed the alignment offsets as follows:
                                     Before             After
H1:SUS-MC1_M1_OPTICALIGN_P_OFFSET     1207.50       1208.3750756649936
H1:SUS-MC1_M1_OPTICALIGN_Y_OFFSET    -2025.30      -2024.9455029701996 	
H1:SUS-MC2_M1_OPTICALIGN_P_OFFSET      527.00        525.7250127471385
H1:SUS-MC2_M1_OPTICALIGN_Y_OFFSET     -534.03       -535.7085973743652
H1:SUS-MC3_M1_OPTICALIGN_P_OFFSET     -722.14       -721.3522559898023
H1:SUS-MC3_M1_OPTICALIGN_Y_OFFSET    -2034.10      -2034.4394052459747
I've spoken with Kiwamu about the seeminly-now-prolific issue of using double-precision python scripts to produce calculated EPICs settings to ridiculous precision, and he agreed to fix it in due time.

The script that does the offloading is launched from the IMC WFS MASTER screen, from the bright blue button marked "! OFFLOAD WFS" just below the output filters in the middle-bottom right. See attached screenshot.

At 5pm PDT both H1 frame writers restarted. This is a very rare event. At the moment it looks like a chance occurance and not related to ongoing DAQ activities.

Greg, Dan, Dave, Jim:

The MSR single mode fiber optics cable which connects the Q-Logic switch to the patch panel for the MSR-LDAS link which was showing errors was replaced this afternoon. Dan is reporting that so far it looks like the errors have stopped, suggesting the old cable was faulty. The MSR Q-Logic switch was reconfigured to enable the single mode port and disable the multi mode port which was connected the two MSR switches.

Timing Upgrades [WP5370]

Richard, Filiberto, Andres, Jim, Dave:

The timing source for the MSR timing master was transitioned from internal GPS receiver to external 1PPS. The transition went smoothly. The broken timing fanout at EY was replaced with a spare. The timing MEDM screen is now GREEN with no errors.

BIOS changes of end station SUS front end computers [WP5374]

Jim:

The BIOS settings for h1susex and h1susey were modified to be identical to the LLO computers. This was an attempt to clear the IOP glitching seen since the faster computers were installed last Tuesday, it did not clear the glitching.

RCG upgrade of end station SUS and OAF [WP5372]

Jeff, Betsy, Jim, Dave:

the h1susetmy and h1susetmypi models were built against RCG branch-2.9 to install the TrueRMS fix. We found an EPICS initialization problem and reverted the models back to RCG-2.9.5. Rolf has found the error so we can try this upgrade again.

New h1susetmxpi model [WP5365]

Dave, Jim:

the first roll-out of the h1susetmxpi model was made. To remove the "DAQ too small" error and the writing of channels to the commissioning frame which were not needed, I removed all DAQ Channels definitions in PI_MASTER.mdl. At this point the RCG chose two fast channels and set them to "acquire=0" in the ini file. This solves the problem without writing data at 64kHz to the frame. I applied this change to both h1susetmypi and h1susetmxpi

Add Beam Tube Accelerometers to PEM mid stations [WP5375]

Robert, Vinny, Dave:

the models h1pemmx and h1pemmy were modified to change the 11th ADC channel from a generic input to the BEAMTUBE_CRYO acceleromter channel. They were added to the science frame at 2kHz.

MSR Front end Rack Network Switch Reboot [WP5371]

Jim:

As part of the FE channel access problem investigation, the Netgear switch which supplies the FE-VLAN to the front end computers, boot, build and guardian was rebooted. The downtime was about a minute and all systems reconnected seamlessly afterwards. It does not appear to have fixed the problem.

MX PEM power supply failure

Jim, Dave, Richard, Filiberto:

While restarting h1pemmx models I noticed the ADC data was zero, starting around 9pm Friday 17th July PDT. We found the -18V power supply to be off. We replaced it with a spare and the AA chassis is now operating correctly.

Server reboots

Patrick, Dave, Jim

The following servers were rebooted as part of preventative maintenance: all conlog machines, h1guardian0, cdsegwe0, h1hwinj0, h1hwinj1.

after the first reboot of h1guardian0 the guardian nodes did not auto start. After a second reboot guardian started correctly.

Travis, Nutsinee, Jeff K., Betsy, Ed, Cheryl, Hugh, Jim, Stefan, Sheila, Elli, pretty much everyone in the control room!

This log is meant for fellow operators, but commissioners might find it useful as well.

- We started relocking in the afternoon. Locking green went well. There was an argument about whether to request LOCKED NO SLOW or LOCKED WITH SLOW FEEDBACK (as stated in the Initial Alignment Wiki). I went for LOCKED NO SLOW because I knew it worked two Saturdays ago but I didn't know when was the last time wiki got updated (and I also don't know what LOCKED WITH SLOW FEEDBACK was). It is a good idea for the person who's maintaining the Ops wiki to put in the date and the content that was being updated (this was already brought up by TJ, but I just wanted to emphasize how important it is if we are going to use the wiki as the reference so we know we can trust it). 

- Everything went well up until MICH DARK LOCKED. The counts on the BS output were awfully high and the AS port wouldn't go dark. The solution to this in a normal situation would be to request DOWN and wait until the beam splitter becomes settle. However, Stefan found that there was a change in the LSC MICH filter and the Guardian tried to turn on the filter that no longer existed (Guradian turning on a blank filter wasn't a problem, the problem was a filter was supposed to be there). He reverted the filter to the way it was the night before. Then everything worked again. I wasn't sure if this difference showed up in the SDF or not.

- Mode Hopping! Soon as you approach DRMI LOCKED, get the IFO_ALIGN window ready. You will have to touch the beam splitter if the RF18 (pink) is low. You have to act quick! Or the interferometer will give up on you. Use AS port camera as a reference for pitch/yaw and watch "pink" and "purple" to see if you go the right way.

- We lost lock at LOW NOISE ESD ETMY. After that we had difficulty relocking ETMY green. Power dropped after WFS was engaged. Shortly after Travis found that the TMSY slide bar was off from Kissel's screenshot in the morning (or last night?). After we put TMSY yaw back to where it was and the arm was able to lock in green. Not only the screenshot never hurt, but it has also proven to be helpful! 

- We also switched ISI Windy Blend filters on for ALL Quads because the wind was reaching 35 mph (and we thought might have lost lock because of that). However, TURNING ON WINDY BLENDS TRIPPED BOTH ETMX AND ETMY ISI. Jeff demands all ISI filters are set the same while Sheila thinks it's okay to leave the Windy blend at the corner station on. Personally I would turn on the corner station windy blends iff I'm desperate. Since the wind wasn't too bad we switched the filters back to "Quite" (I'm sure it should spelled "Quiet"). This configuration worked fine the night before.

- Then we lost lock at LOW NOISE ESD agian. At least we got all the way there right after maintenance period!

- Ed was updating the Wiki during the relocking. It should be as up-to-date as it possibly can right now.

From the time the end station sus machines were upgraded to the faster model, we have been seeing gliches on the SUS IOP model, which translates to IPC error over to SEI and ISC front ends. Sometimes these trip the ISI watchdogs. This has happened 11 times in the past week. Attached is a two week minute trend which shows the following:

top right plot, TIMING or ADC error rate on h1iopsusey whch starts when the new computer was started 9am Tues 14th July PDT

bottom left plot, receive errors on the ISI ETMY model, these latch on until DIAG_RESET

top left plot: ISI ETMY trip flag, this initiates the watchdog trip

bottom right plot: ISI ETMY watchdog trips, latched until user untrips. 4 = full trip, happens each time trip flag is set to one.

The high voltage cables for the cosmic ray detector were pulled from the CER (PEM-C1) to BSC3. Cables need to be terminated with MHV connectors at both ends. Four BNC cables were pulled last week. Electronics are currently being modified for installation in the CER PEM rack.

Started installation of Low Voltage Noise ESD Driver chassis at EX. Field cables were pulled from electronics bay (SUS-C1) to field rack (SUS-R1). Items for next Tuesday:

1. Modify H1:SUS_ESD_06 cable and pull from SUS-C1 to SUS-R1.
2. Install driver chassis in same slot (SUS-R1) as unit at EY.
3. Build ±24V power cable for driver chassis. 

Ever since the 3ifo long-term storage containers have been utilizing CP2's dewar boil-off for purge gas we have been getting alarms due to poor PID control following a Praxair LN2 delivery.  Today I valved-out the 3ifo line (0830 - 1600 hrs local) for the Praxair LN2 delivery and the control of CP2's level was significantly improved.

Times PST

7:40 Jeff B to LVEA for N2 dewar and CC work

8:11 Jim B to EY for timing fanout and BIOS work

8:13 Jeff B done

8:16 Hugh and Jim to HEPI mezzanine for pump station work

8:19 Andre to LVEA for cosmic ray detector cabling

8:20 Praxair on site

8:31 Vinny to EY for PEM work

8:33 Karen and Cristina to both ends

8:36 Kyle to both mid stations

9:00 Richard done w/ GPS work

9:15 Vinny to LVEA for tiltmeter work

9:23 Kyle back from mids, going to LVEA

9:37 Jim B done at EY, going to EX for BIOS work

9:40 Dave doing EY software updates

9:48 Kyle done in LVEA

9:58 restarting EY SUS

10:00 Karen and Cristina  done

10:15 Nutsinee out to CER to reset TCS-X laser

10:20 Fil to EX for ESD LVLN install

10:38 Jim B done

10:45 Praxair on site

10:45 Kyle touring LN2 dewars

10:47 Hugh done

11:04 Nutsinee and Elli to LVEA TCS sensor check

11:38 Kyle done

12:08 DAQ restart

14:45 Dave to MX for PEM power supply replacement

14:52 Richard to CER

15:14 Dave back

J. Kissel

The renewed number of IPC errors caused by timing errors from the new fast SUS front ends are causing a problematic number of ISI trips when there are no suspension payload trips. I remembered questioning this issue back in March, and rediscovered that the BSC-ISIs are mishandling the error signals from the Dolphin IPC parts -- see the original identification of the problem LHO aLOG 17044. The state of the payload watchdog flag is still exactly as quoted in that aLOG.

We need to solve this problem, because it's unclear how/when we'll fix the timing errors on the SUS. I'm half-inclined to just make this change *now,* but after reading the aLOG, I don't remember which is the right way to fix the problem. I've tried calling Stanford, but I didn't get an answer, will send an email.

Thank you Evan for helping clear out the accumulated ASC/LSC SDF Diffs last night.  After the reboots from today, I used SDF to check for settings which were not restored (for some reason) today.  The only offenders I found were that the L2 PIT OLDAMP loop outputs were turned off on ITMX, ITMY, and SR3.  After trending that they have been on for the last week or 2 (commissioners confirm they should be on), I switched them back on.

Rana, Matt, Hang

We did some further investigation of HAM5's coherence with DARM, as suggested by Gabriele (aLog # 19756). A list of what we did:

05:35:10 (UTC): Tapped the lower middle flange south of HAM5. 
05:35:32 (UTC); Tapped the same flange harder.
05:35:54 (UTC): Tapped gull wing. The interferometer lost lock. Before the unlock, the IFO was operating at 24 W with LSC FF on.

A plot of the corresponded time series was attached (TSaccHam5vsDarm.png). We could see ringing corresponding to tapping the flange in HAM5 channel, yet they did not seem to have a significant effect on DARM.  The frequency of the ringing was 207Hz with a decay time of 1.6s. No 90 Hz feature seen. The gull wing tapping did not appear in the ACC_HAM5_SR1 that Gabriele noted.  Summary: no clear connection between this ACC and DARM.

We continue the charge measurements on ETMs.
Results for ETMX are consistent with negative trend, now the charge is from 10 to 20 [V] Effective Bias Voltage for all the quadrants.
Results for ETMY do not not show a significant trend (probably, the data are beginning to be consistent with positive trend). Charge is below the 10 [V] Effective Bias Voltage for all the quadrants.

Note:  We had positive bias on ETMX and negative bias on ETMY after discharging procedure. So it seems possible that charging is caused by the bias voltage.

Hang, Matt, Evan, Stefan, Rana

WE mostly were chasing weird locklosses and instabilities, but also managed to implement a few noise improvements:

1. SRCL FF better shaping via fitting and better measurements.
2. A2L decoupling gave us 10-30 Hz improvement.
3. MICH FF tuned up a bit. New measurments taken; fitting and filter generation yet to be done.
4. ASC tune up: hard/soft modes for CARM.
5. ETMY Butterfly mode identified and notched.

Of the locklosses, some were just due to marginal points in the transitions and loop margins. The main issue over the past few days turned out to be that the TIDAL servo was turned OFF somehow on Friday evening. After switching that back on for ETMY, we have returned to having long locks. The highpassing of SRCLFF has removed the high power pitch instabilities that we were seeing.

We were able to get > 40 Mpc with 10W input. The butterfly mode of ETMY @ 6053.81 Hz is preventing us from turning on the OMC DC whitening right now, so we don't know how good our range really is, but our low frequency

It should not be on during data taking.