HAM2 and HAM3 HEPI were released today.
Raw IPS readouts were recorded prior to release. The following offests were set in the actuation path to reestablish alignment, based on the IPS readouts:
OFFSETS (cnt) | ||
HAM2 | HAM3 | |
H1 | 370 | 900 |
H2 | 800 | 200 |
H3 | -600 | 350 |
H4 | -400 | -650 |
V1 | 450 | 300 |
V2 | 460 | -650 |
V3 | 1050 | 200 |
V4 | 900 | 1250 |
IPS readouts were recorded after reestablishing alignment:
NEW READOUTS (cnt) | ||
HAM2 | HAM3 | |
H1 | 1438.7 | 6664 |
H2 | 957.61 | -4089.2 |
H3 | 2089.2 | 9159.3 |
H4 | -1238.1 | -2252.5 |
V1 | -2796.8 | 2709.1 |
V2 | -608.92 | -6394.7 |
V3 | 1029.5 | 2257 |
V4 | -2862.8 | -9434.1 |
The Before/After difference in IPS readout is shown below:
BEFORE/AFTER DIFFERENCE (cnt) | ||
HAM2 | HAM3 | |
H1 | 107.6 | 146.6 |
H2 | 0.11 | 588.7 |
H3 | 68.2 | 169.1 |
H4 | 65.5 | 104.3 |
V1 | 54.1 | 155.1 |
V2 | 97.09 | 32.3 |
V3 | 4.5 | 144.5 |
V4 | 20.1 | 38.3 |
The remaining shift in alignment induced by unlocking HAM2 and HAM3 HEPI is less than 0.25mils (147 counts on the IPS), on each pier, after reestablishing the alignent with offsets on the actuation path.
Vented spool (in place of iLIGO BSC5) annuli -> No inner O-ring leakage -> removed temporary O-ring valves from bolted joint pump ports and connected permanent ion pump piping -> pumping annulus system with aux. cart overnight
For future reference, The calibrated channel representing the frequency of the green laser (in [Hz]) is H1:ALS-Y_REFL_CTRL_OUT_DQ This can be converted to the length of the arm cavity using the free spectral range, Cavity Length = lambda * (L / c) * H1:ALS-Y_REFL_CTRL_OUT_DQ
Careful: Since we implemented a bypass path from the PDH board to the laser, there is 22dB less signal on H1:ALSY_REFL_CTRL_OUT_DQ (the fast slider on the Sigg board is at -22dB, since this signal is now goint to the laser). This means that the calibration has to be corrected by these 22dB. See elog 6900.
GregG unlocked these suspensions and they are now floating again after a long hiatus of sequestration. Please regard the "Unlocked" signs and be mindful of any mechanical shorts against the floating elements. If you drop a screw or other smaller item, don't just forget about it as it may short the suspension too. Please avoid stepping on the Crossbeams if you don't wish to hear about it from BetsyW!
Bubba craning small loads over tube (consulted with Worden) Lots of CDS work for Maintenance… Paradise Water on site. Arnaud started running Transfer Functions on HAM2/HAM3 Work on HAM 2 - Hugh
today we restarted the DAQ about 16 times. I'll abbreviate this as DR.
First due to the restart of h1iscey the new h1pemey model was loaded. A DR was needed to resync the DAQ.
I created a new h1peml0 model as per Roberts latest changes. I restarted h1peml0 and did a DR.
Bad DUST channels in the DAQ. We found that some of the DUST EPICS channels went bad in the frame starting 7/9 when I added EX to the DAQ. To verify this I first took out all the ECAT channels from the EDCU (DR), and then I removed all the EX systems from the DAQ (DR).
The number of EDCU channels changed from 195,003 to 191,506 to 173,744. At the lowest number the DUST channels became good again.
To figure out what number of channels would break it again I put the ECAT channels back in (DR) and then h1susex (DR). The number of channels went to from 177,241 to 185,360 at which point some DUST channels went back to not as many. Later I put the system back to the full 195,007 (DR) and the full set of bad DUST channels came back.
For more info we moved DUST channels around in the H0EDCU_DUST.ini (DR) and found that the position within the file determined which ones were bad (for e.g. 3rd and 5th slots were bad). We moved the order of the H0EDCU_DUST.ini in the master file (DR)(did not change anything) and removed the H0EDCU_WEATHER.ini from the DAQ (DR)(did not change anything).
Hugo found ADC noise on the ISI HAM6 system. We power cycled the h1seih16 front end computer and the IO Chassis with no success. I removed the ADC cable and the noise went away. At this point I handed the problem over to Richard and Filiburto.
The latest Beckhoff ini files were ingested into the DAQ (DR)
A reminder that all our fast DAQ channels show up in RED due to the incorrect dataValid flag. This will be fixed next week when we upgrade to RCG2.7
J. Kissel, A. Pele For some reason the CART2EUL Matrices had been lost over some reboot / burt restore cycle. We've re-installed them, and captured + committed a new safe.snap. Install instructions: ]$ matlab -nosplash -nojvm >> cd /opt/rtcds/userapps/release/isc/common/projections/ >> load ISI2SUS_projection_file >> fill_matrix_values('H1:SUS-BS_M1_CART2EUL',ISI2SUSprojections.h1.bs.CART2EUL) >> fill_matrix_values('H1:SUS-BS_M1_EUL2CART',ISI2SUSprojections.h1.bs.EUL2CART) >> cd /ligo/svncommon/SusSVN/sus/trunk/Common/MatlabTools/ >> save_safe_snap('H1','BS') >> exit ]$ svn commit -m "New safe.snap file for H1:BS" /opt/rtcds/userapps/release/sus/h1/burtfiles/h1susbs_safe.snap
The demodulation phase of the IMC rotated by approximately 90 degrees. This is the third time to see this kind of phase rotation in the IMC. Very strange.
Temporary fix:
Temporarily I switched the signal from the I to Q signal by changing the actual cabling at the corner field rack and flipped the control sign at H1:IMC-REFL_SERVO_IN1POL by toggling the switch in the upper left of the IMC board screen. The IMC is now locking and should serve as an interferometric monitor for the sus and seismic people. The WFS loops are currently disabled by setting H1:IMC-WFS_GAIN to be zero to avoid a meaningless control. It seems the IMC started dropping its lock after Roberts and his company entered the PSL enclosure. The temperature transient somehow caused it ? It is unclear.
By the way the attached is a picture of the current I-mon and Q-mon signals in a X-Y projection. The I-mon is x-axis and Q-mon is y-axis. You can see a line almost in parallel to the y-axis indicating that the IMC length signal shows up mostly in the Q-phase.
To make it clearer, I flipped H1:IMC-REFL_SERVO_IN1POL to be minus (a.k.a. 1).
Prior to releasing HAM2 and HAM3 HEPIs, we recorded the readouts on the postion sensors of the ISIs, and HEPIs, of the input mode cleaner.
HAM2 ISI, raw CPS readouts, in counts
H1 -3388.2
H2 -2936.6
H3 -504.78
V1 -138.78
V2 -966.74
V3 -737.11
HAM3 ISI, raw CPS readouts, in counts
H1 -526.93
H2 266.35
H3 -175.73
V1 -1405.1
V2 -1102.9
V3 -560.2
HAM1 HEPI, raw IPS readouts, in counts
H1 8075.8
H2 -2940.6
H3 -8136.2
H4 -1306.2
V1 182.27
V2 4545.2
V3 -1657.4
V4 -1572.6
HAM2 HEPI, raw IPS readouts, in counts
H1 1331.1
H2 957.72
H3 2157.4
H4 -1303.6
V1 -2742.7
V2 -511.83
V3 1034
V4 -2882.9
HAM3 HEPI, raw IPS readouts, in counts
H1 6517.4
H2 -4677.9
H3 9328.4
H4 -2148.2
V1 2864.2
V2 -6362.4
V3 2112.5
V4 -9395.8
I made, and used, a script to retrevieve all this information at once:
/ligo/svncommon/SeiSVN/seismic/Common/MatlabTools/IMC_ISI_HEPI_position.m
It is commited under the SVN (r7458)
In preparation for unlocking HAM2 and HAM3 HEPIs, here are the alignments and static positions (as measured by the OSEMs) before unlocking, for reference. HAM2 (* indicates alignment offsets are OFF currently, because Arnaud is measuring some lower stage transfer functions) MC1 MC3 PR3 PRM P [urad] +312 +261 -148* -770* Y [urad] -968 -1081 124* -1200* M1T1 [ct] 12247 15407 11976 12430 M1T2 [ct] 13061 15166 16039 14890 M1T3 [ct] 13600 14370 15890 13874 M1LF [ct] 15975 15170 10659 11306 M1RT [ct] 9539 7682 10424 11068 M1SD [ct] 11831 11578 13022 12813 M2UL [ct] 16070 18355 7212 17435 M2LL [ct] 20649 14340 10466 2510 M2UR [ct] 7266 12539 10768 17270 M2LR [ct] 14008 9908 8883 2492 M3UL [ct] 14414 15096 7667 16049 M3LL [ct] 21812 15740 10337 3730 M3UR [ct] 10138 11881 10654 14385 M3LR [ct] 18289 7221 7899 4566 HAM3 MC2 PR2 P [urad] +434 +716* Y [urad] +244 +120* M1T1 [ct] 13708 14293 M1T2 [ct] 12670 12692 M1T3 [ct] 13585 12476 M1LF [ct] 12330 13635 M1RT [ct] 13636 14635 M1SD [ct] 14254 13127 M2UL [ct] 11830 14296 M2LL [ct] 9944 15369 M2UR [ct] 12493 12516 M2LR [ct] 13180 12042 M3UL [ct] 11333 12223 M3LL [ct] 9213 14342 M3UR [ct] 11356 8504 M3LR [ct] 8200 14608
The BS optical lever has been realigned with HEPI unlocked (and untripped), ISI damped and SUS damped. This is also with a verbal OK from ISC saying that the current alignment of the optic is good.
The h1iscey front-end computer lost contact with the I/O chassis Monday evening, it's unknown why. Tuesday morning I disconnected the computer from the dolphin network, then powered off computer. On power up, the I/O chassis connected OK, and the models started. Note that the H1PEMEY model has a 0x2000 DAQ status, indicating that most likely the H1PEMEY.ini file has changed. The 0x2000 status will require a DAQ restart to clear.
Emily, Tristan, Terra, Thomas ITMY optical lever has been showing lots of drift in yaw for the past few weeks (~ 1 urad/1 min), today we were able to close the gate valves and shutter the PSL so that we can open the transmitter and receiver units. At first we thought a few screws were loose and they were tightened, but that didn't help. Then we strain-relieved the fiber optic feeding into the transmitter but that also did not fix this drift problem. I found that the alignment was sensitive to me pushing on a nearby cable tray that was connected to a conduit that was leaning on a copper pipe that the optical fiber was laying on (picture attached). The optical fiber was being tugged on by the copper pipe as well as the conduit and cable tray so we isolated the fiber to decouple the connection. After the fix, the drift was much better, about +/- 1 urad for the course of a few hours, this will hopefully be a good set up to take spectra and long term testing. Also, we re-centered the PR3 and BS optical levers and will continue to recenter as needed for testing by SEI and SUS.
The dust monitors in the LVEA are NOT currently being recorded. It appears swapping the dust monitor in the H1 PSL enclosure has broken the communications.
Upon startup the IOC communicates correctly with each dust monitor until it gets to location 16 (the one that was swapped yesterday). After this it starts reporting back errors of the form: Error: ../commands.c: 49: Sent command � not echoed Received ?
I powercycled the Comtrol this morning. It worked after location 16 for a little while, but the error has returned.
Robert says he swapped the dust monitor in the H1 PSL laser enclosure. First one dust monitor was disconnected from the breakout box outside the entire H1 PSL enclosure. If I recall correctly, the dust monitor at location 16 was then still found by the IOC. The communication errors persisted. The first dust monitor was plugged back in and the other one disconnected. The IOC still found the dust monitor at location 16, but the communication errors went away. The dust monitor at location 16 reported calibration errors. It may be that the wrong dust monitor was swapped, leading to two set at the same location number, but this would not explain why the communication errors persisted after the first one was disconnected. As it stands, one of the dust monitors in the H1 PSL enclosure is disconnected. The dust monitor at location 16 is reporting calibration errors. I am not sure where the dust monitor at location 16 is. The dust monitor at location 10 is not found by the IOC. The remainder of the dust monitors in the LVEA are running again.
Sheila swapped the dust monitor in the anteroom with one programmed at location 10. The one she removed from the anteroom is labeled 'H'. It had no charge left in the battery when I got it. There was no change in the status. The dust monitor at location 10 is still unseen, and the dust monitor at location 16 is still giving calibration errors. This leads me to believe that: The dust monitor at location 16 is in the laser room and has calibration errors. The dust monitor at location 10 is in the anteroom and is unplugged at the breakout box outside the enclosure.
Daniel Halbe, Jess McIver
Daniel has shown a strong, persistent line at 6.8 Hz in all DOFs of the top stage BOSEMs of the ITMY since at least June 12.
As a follow up to his study, I looked for this line in the ITMY ISI and found it in stage 2: very sharply in RX, strongly in RY, somewhat fainter in Z, and much quieter in X, Y, and RZ.
The line is not seen in any DOF in stage 1, looking at the T240s.
Normalized spectrograms of representative DOFs are attached.
I have discovered a line very similar to this at Livingston and it occurs at 7 Hz. This line is found in Roll (very strongly) and not as strong in pitch. It appears to only show up in those two degrees of freedom and only in the top SUS mass. It does not show up in the ISIWIT channels or the top stage of the ISI. Also it does not show up in any of the other suspension stages.
I looked at spectrograms, time series, and ASDs for the CPS sensors on ST2 of the H1 ITMY ISI in all global degrees of freedom, but saw no evidence for a line at 6.8 Hz. An ASD of each CPS DOF during this time is attached in .fig form.
Summary: The HIFO-Y feature at 70 Hz is produced by both ALS periscopes on ISCT1. The features below 1 Hz are coherent with ground motion and OSEM sensors. We did not identify the dominant source of noise between 0.8 and 2 Hz, though this band contributes little to the HIFO-Y RMS. We may be able to reduce the RMS by using cylindrical periscopes that have resonant frequencies closer to 250 Hz, but we will also likely get a free factor of about two reduction in RMS (in the 60-80 Hz band) after noisy installation activities cease.
Vibration coupling to HIFO-Y was reduced by establishing a science mode and a commissioning mode for PSL air handling (Link). There remain two frequency bands that contribute significantly to the HIFO-Y RMS, 60-80 Hz, and 0.3-0.7 Hz.
Figure 1 shows spectra for the PEM sensors that we recently installed whose signals have the highest coherence with the HIFO-Y signal above 1 Hz. The PSL was operating in science mode. The accelerometer that we mounted temporarily on one of the two similar ALS telescopes on ISCT1 (red trace), the one that carries infrared light, has high coherence with and similar shape to the 50-100 Hz band noise in HIFO-Y (red and black traces). Further excitation tests on the ISCT1 table confirmed that the 70 Hz feature is associated with the lowest resonant frequencies of both ALS periscopes. Neither the green nor the infrared periscope appears to dominate, and I could find no particular source of, e.g., backscattering noise. It may be that the noise is associated with the differential motion of the two periscopes in the locked path.
The other coherence features in the spectrum are, at about 180 Hz, one of the higher-Q resonances of the PSL periscope – perhaps driven by vibrations at the 60 Hz harmonic. These 180 Hz vibrations may be smaller after installation, and if they are not suffieciently smaller, we may be able to damp at this frequency. The 12 and 13 Hz coherence regions are produced by table sway resonances of ISCT1. I am less certain about the feature between 5 and 6 Hz, but there is a building resonance at that frequency excited by wind. The seismometers, some distance from the PSL and ISCT1, also show strong coherence with HIFO-Y at 5.5 Hz, supporting the building resonance interpretation.
Coherences with features in the low frequency region are shown in Figure 2. Seismometer and OSEM signals have high coherence with the HIFO-Y signal, except in the band between 0.8 and 2 Hz. We did not find the source of noise in this region.
It may be useful for the DetChar group to look for coherence in this 0.8-2 Hz band (I only guessed the most important channels) and, if none is found, search for upconversion between the 0.1-0.7 Hz band and, say, the 1.2-1.8 Hz band.
It may be possible to reduce the ALS RMS by replacing the periscopes with periscopes that have higher resonant frequencies. We can probably get from 70 Hz to about 250 Hz by using the cylindrical periscopes from iLIGO. Stefan also suggested that we use a single periscope for both the red and green beams, reducing differential motion. I placed an accelerometer on the double periscope, also on ISCT1, and found the lowest resonance to be about 60 Hz. It is not clear to me whether putting both beams on a 60 Hz periscope or keeping them separate on 250 Hz periscopes, would result in the better RMS.
We may get a free factor of about 2 reduction in RMS contribution from the periscope motion after installation activities cease, because Figure 3 shows that ground motion in this band was about a factor of 2 lower during S6. After installation we would expect the ground motion to drop back down to the S6 level, reducing the need to change periscopes.
Figure 4 is a photo of the cylindrical periscopes from iLIGO that we gathered together near the squeezer bay with Corey’s help. None of them are complete, some parts would probably have to be remade.
Robert Schofield, Stefan Ballmer, Emily Maaske, Terra Hardwick, Vincent Roma, Brian Dawes, Tristan Shoemaker
I have computed the coherence between H1_ASC-Y_TR_A_NSUM_OUT_DQ and many seismic and suspension channels in the 0.1-2 Hz band. As Robert noted, there is significant coherence between many of the seismic channels and the output channel. There are also a few suspension channels that also indicate high coherence in the 0.6-0.8 Hz band. I can also broaden my scope if there is interest. This is a slightly random channel list, and if there are channels that people would think could be interesting, I am happy to include them.
Note: Isolation loops tripped only on HAM2-ISI. They are back ON now.
The DC part of those loops (CPS Biases) remain unchanged from prior to HEPI release, hence ensuring consistent alignment of the platforms.