Jeff K. recompiling h1susitmy, restarting h1susb123, restarting DAQ Oscillation on TMSX OSEMs observed, see Keita's alog Robert S. running magnetic injections on ITMY 09:15 Mitchel R. going into the LVEA to work on the cyropump baffle assembly between the X arm and H2 PSL enclosure 09:21 Gerardo M. joining Mitchel R. to work on the cyropump baffle assembly 09:30 Jim W. and Greg G. dropping down ISI cabling at ITMX 09:36 Jeff K. restarting h1ascimc 09:51 Richard M. and Peter K. going into the LVEA to look at HAM1, HAM2 Betsy W. running transfer functions on ETMX 10:08 Corey G. looking for alpha wipes in the LVEA 10:46 Driver for Praxair delivery to CP1 called notifying he was 15 minutes away 13:07 Mitchel R. heading back to the LVEA to work on the cyropump baffle assembly 13:20 Gerardo M. heading back to the LVEA to work on the cyropump baffle assembly 13:28 Ken working on conduit at end X, was at end Y 13:47 Hugh R. and Jim W. floating BSC3 on HEPI ~14:21 Lost and regained the midX FMCS channels, except for the pump status. Error reported on the FMCS server as well. 14:44 Stefan B. restarting h1odcmaster, h1odcy and h1odcx models 15:47 Hugh R. and Jim W. done at BSC3, Mitchel R. done at cyropump baffle assembly 16:01 DAQ restart
JimW GregG & HughR
We got all the actuators disconnected and reset for reinstallation. The HEPI is now unlocked and ready for IAS to again tell us where to go. Given that we pulled the system against the Actuators and have now disconnected them, it isn't surprising that we have rotated CW further. Based on the Dial Indicators our level is now out of spec by a few 0.1 mm and I predict we will need to move ~350urads CCW.
JimW GregG & HughR
We unlocked HEPI and suspended, well, you know, like everything on the HEPI DSCW Springs. While the horizontal position shifted we did put the verticals back close to where they started. Then we set up to look at the ISI Optical Table for level. We were level to +- 0.1mm so we did nothing further there. The vertical position is low by 1.8mm:
Nominal - GztoLz = position in local level - observed average = position error
D0901148 - T1100187
1661.7mm-2.8 = 1658.9 - 1657.1 = 1.8mm
Patrick told me that the TMS BOSEMs looked crazy in the morning, and I passed the information to Richard, who instructed Aaron to do something (power cycle the coil driver? I don't remember the details).
I came back to the control room at 4:30 PM and it's still crazy, showing 1727Hz peak and its harmonics. Looks quite similar to Arnaud's alog about BS BOSEMs (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=7237).
I'd like to know if Aaron actually did what was suggested.
Cable-pulling at Y End was completed. Screens were applied to the X Arm Spool cleanroom in the LVEA. Viewport materials were moved to Y Mid.
Don't know why or for how long this valve has been open
For future reference, usually our own. These are D1000225 extension cables.
I wrote a python script to summarize the IPC channels defined by the H1.ipc file. The script is userapps/release/cds/h1/scripts/ipc_channel_report.py
(Arn, Kiss, Weavy, Trav)
Unfortunately, the ETMx TFs from last night returned all zero value data. SO, that needs fixing.
However, this morning we manually took all of the 6 DOF TFs on DTT and then hand compared all frequency peaks to the model. All peaks match the model frequencies to well within a few percent. SO, the ETMx is currently free and we should proceed with final ETMx HEPI and alignment work.
PDF below for all 6 main chain DOFs - cursors show a sample of peaks matching model frequencies.
Attached and for reference, are the trend of the calibrated osem signals for a week starting from wednesday oct 16th until thursday oct 24th
M0_DAMP and R0_DAMP are signals calibrated in urad and are representing relative motion between the structure (ISI) and the top mass suspensions. The largest shift over the week is 28 urad for M0 and 23 urad for R0
The other plots are individual osem signals, calibrated in um, representing the relative motion between the reaction and the main chain
The largest shift seen among those plots is at the lower left osem of the uim (L1_OSEMINF_LL) = 115um, which is clearly due to the reaction chain moving relatively to the ISI at the same time on Friday morning (see R0_DAMP).
M0 is the main chain R0 is the reaction chain L1 is the UIM L2 is the PUM, UL LL LR UR are upper left, lower left, lower right, and upper right osems
Further notes on Arnauds statement:
"...the lower left osem of the uim (L1_OSEMINF_LL) = 115um, which is clearly due to the reaction chain moving relatively to the ISI..."
could be clarified to
"...the lower left osem of the uim (L1_OSEMINF_LL) = 115um, which is a function of relative motion between the main and reaction chains, and the relative motion between the reaction chain and the ISI, and the possibility of a bumped flag in the process*. "
Specifically noteable is that none of the other 3 OSEMs on this same mass see this size of a motion which means that this signal is not representative of what the mass is actually doing in pitch/yaw. You cannot have only 1 OSEM signal change by a lot and assume that the mass pitches or yaws. 2-3 OSEMs must show motion that agrees.
*I am surmising that over the course of Thur/Fri HEPI adjustments (which are done 1 corner at a time), the ISI started out level, went unlevel corner by corner, and was restored to level by the time the 4th corner was finished. Possibly, during this unlevel time between corner 1 and 4 work, the ETMx suspension UIM LL magnet flag was close enough to get knocked around on the BOSEM a bit (likely because it started off not perfectly centered in range). When the level was restored, the flag snapped back into place, but a slightly different place relative to the LL BOSEM sensor - hence the observed larger 115um shift on just this OSEM.
Remember that this BOSEM senses relative motion between the main and reaction chain since the flag hangs on the main chain and the BOSEM hangs on the reaction chain. So, as Arnaud has discovered over the last few days of this investigation, quite difficult to use in diagnostics.
J. Kissel I found a few bugs in my modifications to the QUAD models yesterday, once I began to make MEDM fields for all the new variables (which, took so long because of Wednesday's meeting hell) -- and also, after speaking with Kiwamu, Jamie, Jenne, Rana, and Joe Betz, I revamped how the lock-in amplifier part was installed, so that he LSC, ASC, and SUS models are all using identical Simulink parts, if not at least a common MEDM screen. This required me to perform the 4 R's: Re-compile Re-install Re-start and Re-store (full functionality) [an oft-forgotten step] all four QUAD models, and restart the Frame Builder (FB) / Data Concentrator (DC) / Data Acquisition System (DAQ) / h1dc0. Note -- because Robert's been working with ITMY, I 4R'ed h1susetmy, h1susetmx, and h1susitmx and rebooted the DAQ last night before I left for the evening, and then did h1susitmy when Robert gave the OK. Note -- I'll post pretty pictures and an explanation later, but the revamp of the lock-in part was to go from sending one lock-in oscillator excitation signal to each coil, I'm sending one for pitch and one for yaw.
Last Thursday, Jason, Greg & Jim logged the results 8150 of the Initial Alignment of the ETMx using the HEPI load Springs. Having been assessed as aligned, SEI proceeded with HEPI Actuator attachment. That concluded Tuesday with Greg logging the dial indicator readings in 8225. While these readings indicate some CCW motion of the platform, they are of order <100urad: taking the single largest reading as the amount of rotation, .008" over the 79" from center of table to DI location (end of Support Tube) gives 101urads. The actual rotation is less as other readings indicate translations rather than pure rotation.
Yesterday IAS gave a look and directed SEI to rotate CW ~500urads. We did this and brought the Yaw error down to 25urads. In this position, the dial indicators read:
Pier | 1 | 2 | 3 | 4 |
X | .511 | .625 | ||
Y | .233 | .297 | .669 | .509 |
Z | .414 | .422 | .409 | .480 |
Computing the change from last Thursday's aligned readings gives in mils:
Pier | 1 | 2 | 3 | 4 |
X | 16N | 10S | ||
Y | 37E | 32W | 32W | 37E |
Z | 2U | 2U | 3U | 3U |
Given that the Ends of the Support Tubes are not at 45degrees from table center, the X DIs see a much smaller component of the true rotation. The Y DIs do see a correspondingly much greater (than the Xs) amount of the rotation. Using the Y average 0.345" over 79" gives 443urads of rotation CW relative to the aligned position from last Thursday. In other words, we were aligned last Thursday, yesterday we rotated 443urads from that position and now we are again in alignment.
We are working on the HEPI tilt decoupling on HAM2: - The latest set of filters designed for this unit has been installed - Turned on the HEPI position loops (5 Hz ugf) and ISI isolation (high blend, control level 2). - Set up tilt coupling measurements parameters: Added cases for that in exc channel lists (to drive X and Y only) and resp channel list (to record HEPI and ISI) Set up drive values: 200 um (200000counts excitation at the error point) for drives from quasi DC up to 100 mHz. 50 um (50000counts excitation at the error point) for drives up from 100 mHz to 700 mHz. 2 um (2000counts excitation at the error point) for drives between 700 mHz and 10 Hz. - Started the measurements. They should be completed around 9am tomorrow morning. Things to look at more carefully tomorrow: - I had to add quite a lot of phase at low frequency in the Z loop, so we need to check there is no confusion with what we think is the latest version of filters for this unit. (the non-boosted open loop had only 11 degrees of phase at 0.2 Hz, so it was close from being conditionally stable, and therefore ringing while ramping on the loop). Made it larger (>45 degrees). The loop turns on well now. - Could not turn on the pringle loops - Could not offset using the bias block - The blend offset was off because of the "IPS_Switch". Could not find this switch in the MEDM screens. Turned it on using "caput". HAM 6: We have installed and turned on the position sensor loops controller designed by Jim and Rich last week. Some work need to be done on the HAM-ISI to be ready to drive for tilt decoupling.
In order to re-assess the ETMx pointing status, Arnaud agreed to re-run fresh TFs tonight. Hopefully they are running.
Yes they are ! Started them at around 7pm, and they should be done by 02:00 am tomorrow
Randy and Scott retrieved the garbing rack and bench from X End-Buffer cleanroom for use in the LVEA-Test Stand area. They spent the rest of the day working on feedthroughs at BSC3 and completed those installations by the end of the day. Tyler helped with CPB stuff all day.
Rolf, Jim, Dave
After running at RCG2.8 for several hours we reverted the end station and ISC systems back to RCG2.7.2.
The 2.8 test was a success, we did not find any data errors transferring slow data directly to the data concentrator. CPU usage, network usages and frame sizes changed as expected.
During the make install-modelname process to reinstall 2.7.2, I accidentally recompiled the h1susauxex, h1iscex and h1susex models. There does not appear to be any local changes to these models, so the same code should be running on these systems. Jeff subsequently changed the quad models after we had reverted.
Jason and I rechecked the alignment equipment as it was left setup from this morning and found no changes (YAW was at 25micro radians CCW needing to rotate CW to fix). Next we broke down and reset all of IAS equipment again. We found a YAW rotation of 65 micro radians CCW (a delta of 40 micro radians)needing a CW fix. Considering the complexity of the whole setup, the number of equipment pieces and fairly long shots this difference of 40 micro radians (8 arc sec) is probably not out of line. Ball is now in the ISI vs SUS court
Jeff, Dave.
After Jeff installed the new h1susetmx model I plotted its cpu usage histogram and compared this with the previous version. Plotted is the histogram of the number of cycles of the 16384 cycles in one second against the cpu processing time for that cycle (in microseconds, 60us being the max). Most cycles were taking 45uS and are now taking 36uS. There are alway a few outliers at the beginning of the second when certain housekeeping activities are process. These usually take an extra 6uS, so the CPU max was 52uS and is now 41us.