kyle.ryan@LIGO.ORG - posted 21:34, Wednesday 06 February 2013 (5400)
X-mid instrument air alarm -> May result in CP5/CP6 level swings -> Will fix tomorrow -> Ok for overnight
[Hugo, Rodica, Paul]
Today we were finding it impossible to lock the IMC again. When we looked at the SUS-MC2-M3_LOCK_L_OUT_DQ channel, the signal appeared to be always railing (on both sides). We have seen this before at times, due to very noisy seismic states (see e.g. entry 5145). However, even with what we expected to be a quiet siesmic and ISI state, we still found this signal railing at ±40,000 counts, making it impossible to lock using the length path. In the end, Hugo realised that when the model was recently updated, the limit on the MC2 M3 lock filters had reverted to its initial old value of 40k counts. Apparently at some point this was increased to 400k counts, but not saved in a safe.snap file. Hugo increased the limit back to this value and the mode cleaner began locking again.
We should update the safe.snap file so that any further reboots revert to the 400k counts limit.
Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot requested from 5 PM Feb. 5 to 5 PM Feb 6. Also attached are plots of the modes to show when they were running/acquiring data. Data was taken from h1nds1. 1440.0 minutes of trend displayed
Transfer function measurements are running overnight on HAM3-ISI.
Active control will be turned back on once it is done.
Same comments apply to HAM2-ISI
Measurements are done. Level 2 Isolation loops are back ON, on both HAM2 and HAM3 ISIs.
Starting Monday the HAM3 HEPI was unlocked for the L4C leveling. Each corner is relocked during that corner's leveling work. As the work proceeds there is invariably some position change but the process is done intended not to move things. It works pretty well with most horizontal changes counteracting one another in the end. At first unlocking, the HAM HEPI (& presumably the ISI and Optical Table Payload) moved a small amount: <1mil X or Y and < 1urad Rz. There was a drop of the platform of 1.1mil(0.0011") the four corners dropping close to the same but different enough to yield a 1.7urad Ry but only 0.5urad Rx. After the last Corner was completed the total motion from the Monday (unlocked) position was: <1mil X or Y and -3.5uradRz. The payload came back up 0.1mil and rotated about the X & Y axes each -1.1urads. This will presumably be the final floating (no DC HEPI drive) when next released. Who know what actually having fluid flowing though HEPI will do even though it should be unaffected. The system was relocked and the motion after this (where it will be for some time) is as follows: Again, less than a mil X or Y and -1.6urad Rz. The table is a bit higher, .5mils with -2.2urad Ry & <.1urad Rx. How the heck does he do it? you ask. Well, the Inductive Position Sensors are 655cts/mil sensitive, by watching the readings while being locked the position can be held to a few hundred counts at worst, and by careful reconnection of the Actuator so it is not stressed.
Turns out we left cables attached to the TFIN and TFA/B inputs on the PMC field box in the racks, with the MEDM switch enabling the TF path. Removing the cables got rid of the noise, and I disabled the path. There is still some 60 Hz noise but it is nothing like it was before.
This is something we've forgotten since the science run ended...always check the cables when noise hunting!!
Checked the PSL PDA spectrum and ODC lock status again this morning and all looks fine. Thanks for following this up Michael.
Excess power diagram of the past day or so. It's very clear that the high frequency noise diminishes and disappears in the recolored channel right around the time that the problem was corrected.
MitchR, JimW BSC2 was unlocked and balanced this morning, mostly problem-free (except for a CPS power cable that had come unplugged) requiring only small changes to the BSC's trim mass on both stages. Locked/unlocked shifts are <500 cts for St1 and <1000 for St2. SUS's C3 cover and optic shields were carefully removed and are sitting on the nearby tables if needed (please notify SEI people before re-installing these). The ISI was left floating for testing, while we have the window.
For some reasons the big offset that had been observed at ASC_ADC_1_7 [1][2] disappeared at around 11:24 AM this morning and never came back. I think this corresponds to the time when I was unplugging the SCSI cable of the AA board. I was trying to look at the raw voltage coming out from the AA board and so for the reason I was attaching a SCSI break out board. I didn't see any raw offset at the time, put the cable back to place and realized that the offset had been magically gone. This is strange because yesterday it came back after I unplugged and plugged the SCSI. This doesn't mean the issue is solved, but anyway we will keep our eyes on this particular ADC channel and AA board.
After the offset was gone, I checked the signal chain from the inputs of the QPD interface box through the whitening filter and ADC to the digitized numbers by injecting analog excitation signal (0.01 Vpp at 3 Hz with DS345). This was a part of the QPD check. The digitized value agreed with an expected count of +/-550 counts with the whitening off. So I conclude that the ADC and AA board are healthy at this moment.
Note :
one of the QPD interface box [3] in the ISC R4 rack doesn't have a frequency response. It is documented in [3]. The rest of two guys do have a frequency response as designed.
[1] LHO alog 5371 "offset at ASC_ADC_1_7 seem from the AA board"
[2] LHO alog 5373 "Maintenance summary"
[3] S1102815-v3 "ISC Dual QPD Transimpedance Amplifier Chassis"
Commencing Matlab TFs on the BS.
Checked at 6:38 am, TFs had finished, apparently OK. Plots pending. Undamped: ^/trunk/BSFM/H1/BSTST/SAGM1/Data/2013-02-05-1044155727_H1SUSBSTST_M1_0p01to50Hz_tf.mat Damped: ^/trunk/BSFM/H1/BSTST/SAGM1/Data/2013-02-06-1044174784_H1SUSBSTST_M1_0p01to50Hz_tf.mat
There was a dropout in the central section of V in the undamped plots, and a dropout in P in the damped plots, but otherwise everything is pristine.
Completed the Leveling of the final corner's L4Cs on HAM3. I'm still crunching the IPS readings to go from the starting locked position to floating to after leveling work floating to current locked position. I expect these to be a few mils & urads at the worst but I'll go through the motions and report, just not now. HAM3 HEPI is ***Locked*** but try to avoid the Crossbeams anyway.
With the EY envelope vented, both doors removed on BSC10, and the downstream nozzle leading to the new spool and to BSC6 dammed, we deinstalled H1 ETMY (final iLIGO test mass on the project). We made an inspection and photo survey for particulate of the chamber, nozzles, optic and SUS, with not much out of the ordinary observed, save what looks like a tan epoxy-like residue on one chamber wall (sample preserved - the in-chamber cleaning crew has seen a variety of residues before, that have typically FTIR'd clean). We then clamped the optic with EQ stops and teflon, extracted OSEMs, removed DLC mount/black glass etc from the optics table, and un-dogged/lowered the SUS to the flooring. We then removed it from the chamber. It will be set in the EY TMS lab for optic extraction from the SUS. This test mass is bound for KAGRA. Next up is passive stack removal. (Fauver/Landry/Gateley/Layne/Z. Haux/Healis)
We also aligned the HR and AR BS elliptical baffles. The spec for baffle alignment is ±0.5 mm both laterally and vertically. The AR elliptical baffle was aligned with a lateral error of 0.47 mm to the right (facing the baffle from the AR side) and a vertical error of 0.05 mm up. The HR elliptical baffle was aligned with a lateral error of 0.25 mm to the left (facing the baffle from the HR side) and a vertical error of 0.0 mm. This closes out the BS cartidge alignment.
Betsy, Travis
Further notes on the elliptical baffle:
1) We did not encounter the issue that the alignment pins on the target would not fit in the baffle as seen at LLO in this alog - https://alog.ligo-la.caltech.edu/aLOG/index.php?callRep=4425. Our pins fit fine into the baffle.
2) However, we DID run into the problem they mentioned where one of the press-fit pins was too tight of a fit in the target. At LLO they apparently reamed out the hole, again mentioned in their alog cited just above. However, since I read that log too late, we forced the pin into one of the targets and broke it off. (Phlltt!) We weren't given much of a hint to assembling/using these, however we were given a spare! Lesson-learned the hard way and we more-carefully assembled the second AOS baffle target.
3) I don't see how the baffle alignment is captured in the mechanics of the baffle if it gets removed. We've been working under the impression that the baffle cannot come off once aligned, which does make installation a bit trickier. Baffle RODA's assure us the baffle is supposed to be "removeable" so we'll pursue how to maintain the alignment with AOS as we haven't stumbled on that in an LLO log.
4) The method of how we aligned the baffle today seems to be the same as what LLO reported in the cited alog above.
4) Lastly, it seems strange to me that the optic height tolerance is 1mm, while the baffle tolerance is 0.5mm.
I restarted and burtrestored the dust monitor and weather station IOCs on h0epics2 that lost communication after the work on the network.
Somehow the hypothesis has been that the peak in the MC OLTF is the structure in FSS and we should be able to mitigate that guy if we retune the FSS notch.
That was attempted today, but the problem is that the peaks the FSS notch is notching out are at about 780 kHz, and MC "peak" we're talking about is a rather broad thing at around 710 kHz. We could move the FSS notch to there, but if we do that the FSS structures are almost totally out of the notch. See attached. Note that the MC OLTF was measured without Evans Notch (TM).
It's clear that the "peak" we're talking about is not the FSS peak per se, it's either a simple gain peaking or something that is not there in FSS. Keeping Evans Notch (TM) sounds fine to me, but if somebody feels the need to put move it out on the option board in the common mode board, please feel free.
Correction: Both the OLTF of MC without (middle) and with (bottom) the notch in the MC path are in the plot.
Thanks Jeff, I have now done that (I hope succesfullly). I left all the gains, limits, etc. in the useful condition for locking, but turned off all the damping filters, lock filters, alignment offsets and master switch, and then committed the new safe.snap files to the svn.