At GPS = 1043567822;
Common Mode Board: common gain = 20 db, compensation and 1st common boost on, fast gain = -2 db.
- MC2_M3_LOCK_L filters: gain -1000, FM5 (150:4) and FM8 (CLP100) enabled, limt to 400k.
- MC2_M2_LOCK_L filters: gain = 0.03, FM1 (0.01:0.1), FM2 (0.03:1), FM3 (Stab8:2), FM4 (300:1) and FM10 (ELF80) enabled, limit to 300k.
Attached is a plot of the calibrated IMC control signals for the following relevant control loop settings:
- Common Mode Board: common gain = 20 db, compensation and 1st common boost on, fast gain = -2 db.
- MC2_M3_LOCK_L filters: gain -1000, FM5 (150:4) and FM8 (CLP100) enabled, limt to 400k.
- MC2_M2_LOCK_L filters: gain = 0.03, FM1 (0.01:0.1), FM2 (0.03:1), FM3 (Stab8:2), FM4 (300:1) and FM10 (ELF80) enabled, limit to 300k.
The signals have been calibrated following L1 example. Details on the calibration follow (both for clarity and to check that I correctly interpreted what they has been done at L1).
---- IMC_F ----
- The fast path control signal is directly acquired from the Common Mode Board and becomes the input to the H1IMC-F filter bank (H1:IMC_F_INMON)
- the filter bank contains the following (enabled) filters:
- "cts2V" that converts ADC counts in volts (0.000610016 V/ct)
- "InvGenFilt" that compensates for the withening filter at the output of the Common Mode Board (double pole at 10 Hz, double zero at 100 Hz, DC gain = 0.5)
- "VCO" that accounts for the VCO DC gain (365714 Hz/V, obtained from the test report of unit D0900605-S1200558 and multiplied by 2 to account for the AOM double pass) and filter (pole at 1.6 Hz, zero at 40 Hz)
- "FtoL" (I added this at H1) that converts the frequency variations dF into equivalent modecleaner round-trip variations dL using dL = dF * L * lambda / c, where L is the length of the IMC (16.47 m nominal), lambda = 1064e-9 m, c=3*10^8 m/s.
---- IMC_X ----
- the output of the MC2_[M3/M2]_LOCK_L filters are input in the IMC-X_[M2/M3] filters bank, respectively
- the MC3 filter bank contains the following enabled filters:
- "dc_cal" that converts the length actuation into mass displacement at DC
- "sus_d" that accounts for the TF from force on M3 to displacement of M3 (rescaled for 0 dB DC gain because of the existence of "dc_cal"). Note that this TF has been calculated for L1 in the case of velocity damping; although we are still using velocity damping here at H1, it need to be verified that we are using the same setting that L1 was using when the TF was calculated.
- "white", a whitening filter (two 0.2 Hz poles and two 1000 Hz zeros) used to get rid of digitization noise. It need to be tken into account when analyzing data.
- the MC2 filter bank contains similar filters, except that "sus_d" is replaced by the combination of "sus_d1" an "sus_d2", that account for the TF from force on M2 to displacement of M3.
- the outputs of both filter banks are summed to estimate the actual M3 motion.
There has been a bit of discussion about the value of the VCO gain used here, as it seemd to be different enough from the one used a LLO (as much as peple suspected the double pass had been ignored, thus missing a factor 2).
After a bit of back and forth by e-mail, Anamaria spotted the problem: in obtaining this facor from the testing documentation I had very naively assumed a linear response over the tested range, thus calculating the gain as [range of output frequency] / [range of input voltage].
Plotting the data points reported in the abovementioned documentation for the units in use at LHO (S1200558) and LLO (S1200580) reveals instead that the two units behave very similarly, but not linearly. The gain in the linear region around 0 V is ~ 0.27 MHz/V, versus the ~0.18 MHz/V calculated over the entire range (see attached plot). The gain can be as low as 100 kHz/V in the region around 5V (or -5V).
Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot requested from 5 PM Jan. 28 to 5 PM Jan. 29. Also attached are plots of the modes to show when they were running/acquiring data. Dust monitor 10 in the H1 PSL enclosure is still indicating a calibration failure.
Work discussed in the morning meeting: BSC1: Hugh, Apollo: attach HEPI actuators Doug: Check alignment after HEPI actuators attached MC: Recheck IMC cavity pole measurement Hugo: Install blend filters for HAM2,3. BS test stand: Jim: Replace 2 broken Trillium sensors Betsy: Move suspension HAM6: Greg, Apollo: HEPI install LVEA: Richard: Rebalance DC power load on ISC racks HAM2,3: Filiberto: Illuminator install EY: Kyle: Vent chambers No Praxair delivery ----- Replacement of floor mats Sprague Delivery for Ken Dust monitor at location 4 moved into BSC2 cartridge test stand with BS 1:01 Filiberto attached power to HEPI 3:04 Cheryl to close the PSL laser shutter to realign camera on top of HAM 2 Daniel working on h1ecatc1 and h1ecatex Beckhoff computers Kyle began a slow vent at end Y. link Filiberto ran power cables for and installed illuminators in HAM2 and 3. link All the front end computers and associated IO chassis were shutdown except for the test stands and mid X. Richard redistributed the DC power at the ISC rack. The fiber optics ports of the timing system were reconfigured to match LLO. link 2:18 - 4:11 Dan upgraded the firmware on the SATABOY controllers in the MSR link Dave modified the h1ascimc model link The glass BS optic was put in the suspension on the test stand link The last of the HEPI actuators were attached to BSC1 link The PSL laser was accidentally tripped and shutdown. Michael R. swapped out water filters for the diode chillers. link ----- Signed work permits: 3687-3690
Found the Y-end MTP shut down during the night -> Problem turned out to be the "over head" chilled water supply valve to the mechanical room was closed for some reason -> Opened valve and restarted MTP -OK Upon measuring the vent/purge air dew point just prior to venting I discovered a problem -> Found that the left drying tower wasn't regenerating correctly -> problem ended up being a wore out swing-style check valve -> Replaced internal (wear) components -> OK -> These check valves (4 per vent/purge skid) have been trouble-free for the first fifteen years of operation but are now failing regularly due to excessive wear -> I will attempt to replace these site wide time permitting. ~1530 hrs. local Began "slow" vent of Y-end ~1600 hrs. local Made small vent rate adjustment ~1630 hrs. local Made a second small vent rate adjustment
WP3690
Dan from LDAS upgraded the firmware on the SATABOY controllers in the MSR. This is the file system for h1fw1 and h1nds1. The data gap is shown below
-rw-r--r-- 1 controls controls 257745931 Jan 29 14:13 H-H1_R-1043532768-32.gwf
-rw-r--r-- 1 controls controls 265951676 Jan 29 16:13 H-H1_R-1043539968-32.gwf
Gerardo, Betsy, Travis
Using the ergo arm, we stuffed the glass into the suspension. The metal face shileds are in place. Work to suspend it will resume another day.
Mark B. and Arnaud P. Starting TFs with Matlab on ITMy.
The TFs ran largely without incident - only one case of failing to read from NDS, in the 0.01-0.4 band for L. M0 undamped: ^/trunk/QUAD/H1/ITMY/SAGM0/Data/2013-01-29-1043539447_H1SUSITMY_M0_0p01to50Hz_tf.mat R0 undamped: ^/trunk/QUAD/H1/ITMY/SAGR0/Data/2013-01-29-1043558145_H1SUSITMY_R0_0p01to50Hz_tf.mat M0 L, P, Y and T, and all R0 plots are good. M0 V and R show anomalies that don't look like anything in the Coloring Book and may be a measurement artifact. There's a discontinuity at 0.4 Hz, as if the gain of some component had mysteriously increased between bands.
I reran the M0 V TF with DTT and it's fine, with none of the strange steps in the Matlab TF. So it's presumably a measurement artifact of some sort.
Per documents T1200220 and D0901084: Illuminator for HAM2 was installed in viewport VPA1F1. Illuminator for HAM3 was installed in viewport VPA1F3. Filiberto Clara
The ISC group discovered another conflict in EPICS names between the h1ascimc mode (a copy of l1ascimc) and the new Beckhoff ISC slow controls system.
Similarly to the WFS_A and WFS_B Gain channels, the MC2_TRANS_WHITEN and the IM4_TRANS_WHITEN were removed. The change was applied when h1asc0 was restarted after this morning's shutdown.
A big maintenance day today. While the DC power to the front end IO Chassis was being worked on, we shutdown all the CDS frontends with the exception of h1pemmx.
We then reconfigured the fibrer optics ports of the timing system. The MSR master, the two H1CER fanouts and the EY fanout were recofigured to the standard as agreed between LHO and LLO. The temporary fanout in the LVEA for the test stands and the MX fanout were not changed.
All the front end machines were powered down for this work. The MSR front ends were powered up in the order:
h1psl0
h1seib1,h16,h23
h1susauxh2
h1susb123,h2a,h2b,h34
h1lsc0,h1asc0, h1oaf0
EY front ends
LVEA Test Stand Front Ends
Some problem with h1pemey IO Chassis timing, moving the tcs SFP over to the pem IO Chassis fixed this.
All system should now be burt restored and switch on.
Completed the last three attachments. No real problems encountered. Post attachment Dial Indicator readings suggests only a little bit of shift: Level/Tilt--The West and South sides dropped down ~0.1mm; how this actually translates into the suspended ISI tilting is difficult to deduce but I'd say the effect is less. If it does fully tilt the Optical Table, using my level observations of the ISI table Thursday 24 January, we are still in spec; runout of 0.2mm. It may possibly cause some grief to SUS though, SEI's condolences. Horizontal position--I see about 0.1mm of maximum translation with most readings 1/2 or even less than that. I would forecast a maximum rotation of 50urad CCW. So all & all pretty good; thanks to Randy Thompson. There is one issue with this system though. The South Crossbeam is only about 1mm away from the HEPI Housing on the SE Corner. This is enough room for nominal operation but over time if the system walks North for what ever reason, we may have to address that. Attached is a photo.
The laser shut down earlier today due to a communication failure between the control boxes and the Beckhoff computer. What I think happened is the power to the ethernet to fiber converter for the LVEA control box was lost - this is plugged into an AC outlet and the prongs were barely in, so it probably got bumped. I'll wait until the CDS work is done to bring the laser back up.
Restarted the laser. I had noticed that the diode chiller flow level was slowly decreasing over the past two months, and thought it might be due to clogged filters. I swapped both filters out from the diode chiller but the flow was less than it was before. I increased the flow as high as I could, to 22 lpm, but it has continued to decrease, although it looks like it is slowly settling down.
I included a 60 day trend of the flow rate, which shows that something happened at the end of December which decreased the flow and hasn't stopped. I don't believe it is due to anything we have done, as this occured over holiday.
Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot requested from 5 PM Jan. 27 to 5 PM Jan. 28. Also attached are plots of the modes to show when they were running/acquiring data. Dust monitor 10 in the H1 PSL enclosure is still indicating a calibration failure. The data was taken from h1nds0. 1260 seconds worth of data was unavailable on this server 1440.0 minutes of trend displayed
Transfer functions were measured during the week-end. Stage 1 and stage 2 look good (cf TF_L4C & TF_GS13). On the other hand, the TFs from stage 1 actuators to T240s in corner 2 and corner 3 don't look right (gain and resonance). These 2 instruments don't seem to work properly.
Matt, Keita
As noted in entry 5279 we had some trouble with changing MC2 acuation after the MC2 model change/restart. It is not clear if the state before the change was the "right" one, and no explanation for the change was found (all MC2 coil driver bits were flipped, with no apparent change). This should be looked at again with a model of the system in hand so that the measurements can be compared with expectations. Since the "new" MC2_M2 TFs make the L1 filters stable, I am inclined to guess that we have returned to the "right" state.
In this state I changed the M2 stability filter to be very simple (zero at 2Hz, pole at 8Hz). This gives the M2 cross-over resonable phase margin for all UGFs below 10Hz. The attached images show the MC2 TFs with the M2 gain set to zero (no M2 actuation), M2 gain set to 0.03 (generally enough, and good for locking), and M2 gain of 0.1 (good for ISI work or high seismic times). The maximum "good" value of the M2 gain is about 0.2, more than 0.3 is unstable.
Note that in the TFs involving MC2_M3, this system is working with heirachical control, so the measurement includes M3 and M2 (e.g., the whole MC2 actuation, or combine MC2 M2 and M3 actuator). These TFs show that the M3 gain has about 8dB of gain margin, so gains more than about -2000 are likely to cause problems (-1000 is the nominal value).
Mark suggested to modify the "plot****_dtttfs.m" scripts for QUAD HLTS HSTS HAUX and BSFM. From now, it will automatically create a Results folder in the right path (if it is not already existing) to save the pdf figures. For example .../SusSVN/sus/trunk/HAUX/Common/MatlabTools/plotHAUX_dtttfs.m line 108 to 113 : DIR=exist(resultsDir,'dir'); if DIR==7 else system(['mkdir -p ' resultsDir]) end