We could not retreive data for the transfer functions that ran overnight on ITMX.
Run_Get_Batch.m, the generic script that the seismic team uses to get data from the frames, would systematically crash when collecting data for Stage1-V1 excitation, in the 500Hz-1000Hz frequency range (see attachement #1). It would also crash when trying to collect data for the other frequency bins.
We looked at the data stored in the frames (DaveB, and ArnaudP helped), and compared it with what was in the slow channels.
- The actuation signals were sent to the correct channels and witnessed by the related slow channels (see attahchement #2)
- However, the daq channel for the excitation along Stage1-V1 does not recall any excitation (see attachement #3), while one was recorded by its slow channel twin (see preious comment).
New channels were added to the master model, and h1isitst.mdl was not restarted since. When JimW and I restarted the TST model yesterday night, those channels were added to the .ini file. JimW and I did not dare restarting the DAQ without CDS people around. This is likely to be the reason why the data in the frame was corrupted.
A couple DAQ restarts were performed as part of the maintenance day today. Last one was around 3pm. We then started to run a quick set of high frequency transfer functions, and made sure that the data collection process went well. It did.
Transfer functions will be running ovenight on ITMX (currently on the test stand, using TST model)
** Work permits signed by ops today: 4157 (EY one-stop cables), 4158 (FW1 reconfigure SSD disks), 4159 (online detchar channel) ** LN2 delivery to CP3 (MY) ** Roofers moved a lot of material and equipment onto the Corner Station roof; also worked on the roof at MY ** Two DAQ re-starts, morning and afternoon ** A number of H1lsc CDS overflow alarms; I checked with Dave and disabled this alarm ** Lots of work at/near HAM1 today ** Usual Tuesday morning vendors on site ** EX spool clean room was cleaned this morning ** Cheryl worked in the EX TMS lab ** Dust monitor 9 continues to alarm in the white condition ** Lots of dust alarms at EX today as personnel were working in the VEA. 12-hour trend attached. ** LVEA has been in laser hazard all day and remains in laser hazard at 17:20 local.
HughR, GregG, JimW
Today, we managed to get most of SEI's cables sorted and connected. Cable brackets were installed, unfortunately I forgot that the GS-13's need to be shorted to the ISI, until Hugh reminded me. Still need to fix corners 2 and 3. Otherwise, it went well, seismometer cables (unplugged at the pods) and CPS's are plugged into the chamber. I watched MEDM as Greg connected CPS's, and we ran an emulator on the seismometer cables, so we are pretty sure the right sensors are plugged into the right spots. Now, we are ready for TMS and SUS to go in, uncover, add the last of their hardware and plug in before SEI can float and balance.
Doug and Jason We are setup and ready to start the in situ position alignment of the ETMx and relay the results to Hugh for ISI tweaks. We will need 30min in the morning to make sure none of out equipment has moved over night. Scott and Randy (Apollo) rechecked and verified the elevation monuments based on BSC9 chamber centering.
The CPS on board target ground lugs are tied to chamber ground. All sensor cables are strain relieved at the feedthru protection shrouds. The EM actuator cables still need to be strain relieved.
(Rich, Alexa, Daniel)
We connected the outside cables on HAM1, labeled them and mounted the feedthrough protectors. The RF cables were routed in the cable tray, but still need to be terminated on the rack side. The only missing cables are the LSC PD DC supplies which we were unable to locate.
We also mounted the RF cables on HAM6, but they are not routed yet.
WP 4157 Moved One-Stop cables at computers and I/O Chassis for h1susey, h1susauxey, h1seiey, and h1iscey computers. One-stop cables are now in numerical order according to the location in the equipment rack of h1iscey, h1susey, h1susauxey, and h1seiey.
Completed reconfiguring the external RAID array attached to FW0 under WP 4158, to match the configuration on FW1. (Note that I had the two FW designations reversed in the original work permit text - it was in fact FW0 that was reconfigured, FW1 was untouched.) It took approximately 3.5 hours to backup the minute trends to the SATABoy, with the SSD filesystem at 40% utilization. For a nearly full SSD filesystem, expect the backup to take the better part of a day.
A very simple function was made yesterday to help with the HEPI alignment work. It takes an 1x8 vector as input, and puts its content into the DC biases of a given HEPI. It may, or may not, be useful later on.
It is saved under the svn as:
/ligo/svncommon/SeiSVN/seismic/HEPI/Common/Testing_Functions_HEPI/Set_HEPI_Cart_Offsets.m -r7597
Macro substitution text files were intalled for all H1 HEPI and BSC-ISI.
The IOP_NAME macro substitution was added. It corrects the issue seen with the display of the IOP Watchdog status, on the BSC screen. Befrore/After comparison is attahced.
The SITEMAP.adl was updated, and chamber names (bscX) were added to the labels. A screenshot of the new sitemap is attached.
The macro substitution text files, and the new SITEMAP.adl were commited to the SVN. Details below:
[Kiwamu, Stefan, Joe, Paul]
The first measurements for the IMC g-factor have been made, using the sideband sweep method described in e.g. LLO alog 4849 (https://alog.ligo-la.caltech.edu/aLOG/index.php?callRep=4849).
The broadband REFL AIR photodiode was used to observe light transmitted through the IMC while the 45MHz EOM input was driven from the HP network analyzer in the LVEA. We used the minicircuits RF amplifier which is currently clamped to the ISC rack to get enough drive voltage on the EOM: roughly 10Vpp at 30MHz. An iris was used to partially block the beam on ISCT1, breaking the symmetry on the PD such that the higher-order mode signals were not cancelled out due to their spatial orthogonality.
The attached plot shows the transfer function from network analyser drive signal to BB REFL AIR measured signal, over a range of about 1 FSR. The central peak can be attributed to the HG00 mode, but also displays the resonance of the EOM driver circuit at ~45.5MHz. The HG10/01 and HG20/02 mode peaks are clearly visible, and match up nicely with a Finesse model setup of the measurement.
To measure the resonance frequencies of the higher-order modes more accurately, and hence to get an accurate value for the IMC g-factor, we will take zoomed in sweeps around these higher-order mode peaks soon.
(Kiwamu, Paul, Stefan) This morning we found the back-reflection from the PRM, leaking through the Faraday, back on the PSL table. We measured the following powers: - direct beam, sampled by ~30% BS: 339mWatt - return beam, sampled by the same ~30% BS: 37.0uWatt - as reference, we also measured the power on the REFL beam on ISCT1, just after the periscope: 6.94mWatt Faraday isolation ratio: 37.0uW/339mW, minus other losses (IMC visibility, PRM reflectivity)
Some more small information regarding to this measurement:
Doug was tied up with alignment work at Xend today so I did the weekly trends - see attached images. Note that the PSL shut down yesterday due to another power glitch.
Patrick T., Thomas V. Last night Thomas and I swapped the temporary cables being used to test the rotation stage in the CSR with the ones made for the final installation. These are not pulled yet, but are coiled in front of the rack holding the Beckhoff PSL environmental chassis and attached to a rotation stage sitting on a rolling table nearby. For some reason two cables had been unplugged from the controller box (goes between rotation stage and Beckhoff module). One was power and the other was a control cable. I'm not sure why they were unplugged, but we reattached them. I was hoping this would address problems we were having getting the rotation stage to move, but they have persisted.
Patrick T., Thomas V., Vern S. It would appear after testing that the rotation stage itself is burned out.
The "failure" of the Newport URS50BCC rotation stage was of concern to me as we have three of them in each IFO controlling significant amounts of laser power. These were new units and we had been reasonably careful with them. Why should this one die all of a sudden? This is what I found on investigating our broken rotation stage, S/N B12 6940 (Newport's serial number, not the ICS or DCC number).
The rotation stage's angular encoder was working. The TwinCAT2 control software for the rotation stage was showing angle and the "Mechanicsl Zero" signal was present (a high to low transition on moving from 350 deg to 5 deg). This indicated that the electonics was working and not blown. However, the resistance across the motor terminals (at the DB15 connector, pins 2 and 10) was infinite. Opening the housing that held the DB15 connector and printed circuit while monitoring the motor resistance, I found that it would intermitantly read 57.4 ohms, the proper motor resistance. The problem was traced to a bad connection of the upper ribbon cable which went from the printed circuit board to the motor and shaft angle encoder. The connection was a leaded connector designed for through-hole mounting and soldered onto surface-mounting pads. This is not a method of assembly I would recommend. I patched it back together and confirmed that the stage works. I'll use this for debugging and testing and assign it to our spares.
1st of month, I cycled the idle cooling unit in the MSR. Middle unit is now resting.
I updated the safe.snaps for h1pslfss, h1psliss, h1pslpmc with the current configuration. I am also setting the current on the kepco HV unit to 400mA.
Actually, that was me, signed into the log as jeff.
J. Kissel, H. Paris, A. Pele, S. Ballmer, C. Wipf Yet again today we battled with the HEPI and ISI systems in the ITMY (BSC1) chamber trying to get them back after the weekend's power outage. HEPI is up and running again, but I don't think at the same alignment. The ISI cannot make the final transitions to awesome without its watchdog tripping. Details below. I'm leaving the entire ITMY (BSC1) and BS (BSC2) chambers OFF for the evening so DetCharians can follow up on a study of the 10 [Hz] HEPI pier amplification *AHEM* Laura Nuttall *AHEM*. Ground STS2 Channels ITMY HEPI L4C Channels BS HEPI L4C Channels H1:ISI-GND_STS_HAM2_X_DQ H1:HPI-ITMY_BLND_L4C_X_IN1_DQ H1:HPI-ITMY_BLND_L4C_X_IN1_DQ H1:ISI-GND_STS_HAM2_Y_DQ H1:HPI-ITMY_BLND_L4C_Y_IN1_DQ H1:HPI-ITMY_BLND_L4C_Y_IN1_DQ H1:ISI-GND_STS_HAM2_Z_DQ H1:HPI-ITMY_BLND_L4C_Z_IN1_DQ H1:HPI-ITMY_BLND_L4C_Z_IN1_DQ H1:ISI-GND_STS_HAM5_X_DQ H1:HPI-ITMY_BLND_L4C_RX_IN1_DQ H1:HPI-ITMY_BLND_L4C_RX_IN1_DQ H1:ISI-GND_STS_HAM5_Y_DQ H1:HPI-ITMY_BLND_L4C_RY_IN1_DQ H1:HPI-ITMY_BLND_L4C_RY_IN1_DQ H1:ISI-GND_STS_HAM5_Z_DQ H1:HPI-ITMY_BLND_L4C_RZ_IN1_DQ H1:HPI-ITMY_BLND_L4C_RZ_IN1_DQ H1:ISI-GND_STS_ITMY_X_DQ H1:HPI-ITMY_BLND_L4C_HP_IN1_DQ H1:HPI-ITMY_BLND_L4C_HP_IN1_DQ H1:ISI-GND_STS_ITMY_Y_DQ H1:HPI-ITMY_BLND_L4C_VP_IN1_DQ H1:HPI-ITMY_BLND_L4C_VP_IN1_DQ H1:ISI-GND_STS_ITMY_Z_DQ which I've confirmed are in the frames. Chris is going to do a few more things with the IFO, but things should be off by 00:00 PDT, 07:00 UTC Oct 1 2013, 1064646016. -------- HPI-ITMY The goal here was to get HPI to the same alignment we had during HIFO-Y (see LHO aLOG 6566). However, the biases had been reported in the Colocated basis and we want to progress forward, sticking these biases in the Cartesian basis in calibrated units. After a few unsuccessful attempts, we finally - Turned the position isolation loops OFF - Turned the input to the IPSINF banks OFF - Inserted the desired input readings as offsets in the IPSINF bank - Read off the corresponding values at the input to the blend filter bank (letting the front end do the appropriate basis conversion and calibration for us [thanks Stefan!]) - Turned the IPSINF offsets OFF - Turned the input to the IPSINF banks ON - Turned the 100 [mHz] UGF "position" lock isolation loops ON (watching as the inputs meander off to some weird location) - Installed the desired Cartesian values and a 10 sec ramp into the DCBIAS bank, turned them ON. For the first few minutes, the plan worked like a charm; inputs to the IPSINF moved to exactly where we wanted them. However, after those few minutes and currently, the location has drifted to some other location, mostly in horizontal DOFs. Other commissioners were not as successful at restoring light this far into the interferometer, so we never got the chance to test whether this alignment is good enough. Out of frustration, we'll leave it as is. Here're the final results: Colocated Channel Value [ct] | Cartesian Channel Value [nm or nrad] CART Bias Channel [nm or nrad] H1:HPI-ITMY_IPSINF_H1_INMON -873.2 | H1:HPI-ITMY_BLND_IPS_X_INMON -7156.64 H1:HPI-ITMY_DCBIAS_X_OFFSET = 7156 H1:HPI-ITMY_IPSINF_H2_INMON -822.083 | H1:HPI-ITMY_BLND_IPS_Y_INMON -33425.1 H1:HPI-ITMY_DCBIAS_Y_OFFSET = 33425 H1:HPI-ITMY_IPSINF_H3_INMON -2786.58 | H1:HPI-ITMY_BLND_IPS_RZ_INMON -4374.85 H1:HPI-ITMY_DCBIAS_RZ_OFFSET = 4374 H1:HPI-ITMY_IPSINF_H4_INMON -3773.34 | H1:HPI-ITMY_BLND_IPS_HP_INMON -1073200 H1:HPI-ITMY_DCBIAS_HP_OFFSET = 1073200 H1:HPI-ITMY_IPSINF_V1_INMON -6121.07 | H1:HPI-ITMY_BLND_IPS_Z_INMON -71553.4 H1:HPI-ITMY_DCBIAS_Z_OFFSET = 71553 H1:HPI-ITMY_IPSINF_V2_INMON -1038.65 | H1:HPI-ITMY_BLND_IPS_RX_INMON -46597.7 H1:HPI-ITMY_DCBIAS_RX_OFFSET = 46597 H1:HPI-ITMY_IPSINF_V3_INMON 2772.24 | H1:HPI-ITMY_BLND_IPS_RY_INMON +72672. H1:HPI-ITMY_DCBIAS_RY_OFFSET = -72672 H1:HPI-ITMY_IPSINF_V4_INMON -2990.78 | H1:HPI-ITMY_BLND_IPS_VP_INMON +75188.8 H1:HPI-ITMY_DCBIAS_VP_OFFSET = -75188 ------- ISI-ITMY The goal was to get the ISI in as best-performing state in order to use the GS13s (our best sensors of the isolation system at this point) to measure whether the 6.18 [Hz] feature had been reduced as much as had been shown by Arnaud in LHO aLOG 7919. As Hugo had stated earlier (see LHO aLOG 7915), he and Arnaud got things most of the way up, but not nearly as good as Vincent's work during the original discovery (see 7747) because the configuration Vincent outlines has just enough typos to be confusing. So, I tried my hand at it, having had a few more conversations with Vincent about the configuration but to no avail. My attempts failed at various points along the way, but most often because the watchdog would trip on the actuators as I'd tried to switch in the T240s to the ST1 blend filters. I'm too tired to dig in deeper, but we'll try again tomorrow. Here's what I found broken / frustrating / interesting along the way: - The ISI-ITMY model must not have been restored to a good time (or not at all). The damping loops were OFF entirely (inputs, outputs, gains, and filters). I restored them by hitting the "damp ITMY" command script. - The damping loop status lights are red when functioning nominally. The "correct state" is set to 0x6000034, when the current state is 0x1401 and the loops are running as designed. This should be an easy fix. - ctrlDown of the isolation loops is now a part of the front end, so one no longer needs to remember hit the command button before restarting the isolation process. - The ST0 to ST1 feed forward is not a part of the ctrlDown or isolate command scripts. One has to manually remember to SLOWLY ramp them off (or on). - The Sensor correction is not part of the ctrlDown or isolate command scripts. One has to manually remember to SLOWLY ramp them off (or on). - The Sensor Correction overview screen is grouped by degree of freedom instead of by instrument source -- and otherwise it looks identical. These should be at least *colored* differently. It's tough to tell whether I'm turning on the ST0 L4C sensor correction (which I shouldn't) or the GND STS2 sensor correction (which I should). - The TRAMP for the MATCH banks on the Sensor Correction overview screen is missing. I had to dive into the MATCH banks internal screen to get to the TRAMP. - The full filter bank sub-screens for anything on the Sensor Correction overview screen are not obviously accessible. You can do it by just clicking around blindly and landing on a hidden link, but there's no obvious button like there are for most other sub-screens. - The FULL command script for "isolate ITMY lvl3" froze several times in the middle of ramping up ST2. I didn't diagnose why, but it fails in the middle of ramping up the ST2 RX and RY loops. I had to run ctrlDown and run each of the Stages (and at one point each of the degrees of freedom) separately. - The "isolate ITMY lvl3" script turns on the "Boost_2" filters. Here're some good philosophical discussions that came up: [Kissel with himself] - "There's a 'switch all' screen on the blend filter screen for each stage. Vincent at one point [a month ago now] had set it up that -- even though not all of the blends were the same for each DOF -- the Y DOF's filter characteristics defined the name of all the DOF's filter names for that level of button. This way, the user knew 'I get the best performance if I hit the "T40mHz_NO.44' button on the switch all' screen." This is opposed to having to know the best blend for each of the 6 DOFs and in which order to switch them in. I prefer the "one-button switches in all the right blends." Much better for configuration control. [Ballmer with Kissel] - Instead of writing in Perl, Python, or folding it into the Guardian structure, we should now have the ability (thanks to the cdsCtrlFilt2) to move the entire isolation loop turn on dance into the front end as a c-code state machine. This, if coded correctly (1) shouldn't take more than a month off offsite development, (2) removes all reliance on the EPICs network and interaction with scripts, (3) makes the user/guardian interface a more straightforward state-machine, and (4) removes flexibility to mess things up. I think we should do it.
This is slightly silly, but it appears that the easiest way to park a HEPI at a chosen set of IPSINF values is to run eight instances of ezcaservo in parallel (servoing H1-H4 and V1-V4 sensors/actuators). I set the OUTF offsets of the HAM2, HAM3, and ITMY HEPIs using this technique. HAM2 and HAM3 were set according to the IPSINF values given in alog 7180. For ITMY, the values Jeff had left in the IPSINF filter bank offsets were used. Here's an example of how ezcaservo was run:
ezcaservo -r H1:HPI-ITMY_IPSINF_H1_INMON -s -5730 -f 0.1 -g -1 H1:HPI-ITMY_OUTF_H1_OFFSET
Note: ITMY and BS HEPIs were left switched off and ready for overnight DetChar-ing.
Edited to add: the ITMY IPSINF numbers in the filter bank offsets were bogus. (They come from alog 6566, but that entry is talking about the ETMY HEPI.) We'll have to do some further rummaging to uncover the old ITMY alignment.
Here's some tips that Sebastian gave me to isolate ITMY BSC-ISI
- isolate the two stages one after the other
- start with stage 1 and make sure the gain filter in the GS13 is engaged (otherwise, GS13 will saturate when isolating stage 1)
- when stage 1 is stable, disable the gain filter and isolate level 2
HugoP, JimW
After finishing up floating and balancing the ISI this morning (Stage 2 was floated on Friday, Stage 1 required minimal adjusting this morning), we started testing the ISI in the afternoon. First we built and installed a recently updated BSC model, then populated the MEDM, then used a somewhat shady epics variable short cut (exclusive to the ISI test stand) to de-activate the SUS watchdog (the quad is on stops, so the watchdog is not doing anything right now, anyway). After all that, we were able to start driving, first the actuator static local basis test to verify the actuators behaving properly, and now an overnight transfer function. Should finish by the morning meeting.
Jim, Hugo,
The L4C WD threshold came back to its nominal vaule and tripped the ISI at 4:04am PCT.
We are now collecting TF data, which should still be valid in the 10Hz-1000Hz range.