Michael, Krishna,
This evening we tried using the tilt-subtracted ground sensor for sensor correction(SC) in the Y-axis on Stage 1 at EndY. Wind speeds were between 0-10 mph. We tried four configurations:
1. 90 mHz blends (Quite_90) only
2. 90 mHz blends and broadband sensor correction using BRS-2
3. 45 mHz blends only and
4. 45 mHz blends with broadband sensor correction using BRS-2.
The results are shown in the attached pdf. The first plot shows the ASD of Stage 1 T240 seismometers. As expected 90 mHz blends do not provide any isolation at the microseism, and 45 mHz blends do well there. But 90 mHz blends and SC performs nearly as well as the 45 mHz blends. The second plot shows the rms of the ST1 capacitive sensors, which is a measure of the net DC motion of the platform. 90 mHz blends + SC does a factor of ~5 better than other configurations. The third plot shows the ground motion during the four configurations and the tilt-subtracted sensor signals in each case, showing the factor of 3-5 improvement.
Bottomline: 90 mHz blends and sensor correction using BRS-2 does nearly as well as 45 mHz blends above 0.1 Hz and improves the low frequency rms by a factor of ~5, even under mildly windy conditions. We will repeat this test under higher wind conditions, where performance is expected to be even better.
(Bubba is the primary contact for this alarm) but I think that this can be dealt with in the morning as long as water use is kept at a minimum tonight.
I just checked - there's lot of water in the tank. Near full.
RO = "reverse osmosis"
(For future reference) The site should be able to supply more than one day's worth of potable water past the initial RO alarm time. As such, an RO alarm initially occurring after hours won't need to be addressed until the next day.
J. Kissel, J. Warner After getting all of the mechanical labor out of the way (see LHO aLOG 26452), Jim and I took to the B&K system for our first attempt at characterizing the resonant frequencies of ISI HAM6 blade springs and GS13 cans. No excitingly awesome results here; mostly just me writing stuff down for future reference. Stay tuned for more substantive results after tomorrow's work is complete. ------------- We hit in two locations with accelerometer as close to the blade spring and GS13 can hit locations as possible. The blade hit location was roughly in the middle of the blade, stroking the hammer upward into the blade. Today's data is from Corner 3's blade alone. The accelerometer location was very similar to what's shown in pg 1 of LHO aLOG 25472's attachment. The GS13 Can hit location -- the H1 GS13 -- again suffered an awkward hammer stroke from beneath the can (because we didn't take the outer wall off). The accelerometer for this hit was mounted on the outer side wall just above the can. We hit in two locations simply because it was the first data we'd seen and it wasn't processed well enough to confirm that we can see both systems of resonance in the blade-spring hit location. Attached are the data, processed. Though the three axes indicate different mode shapes, we weren't particularly careful about aligning the DOFs with any other reference set of coordinates. We'll try harder tomorrow. The processed data is attached. We expect the blade spring's 1st bending resonance to be at ~153 [Hz], and indeed that's exactly what we see. No surprise there. We see some more resonant features in the blade-hit location data which I have not specifically called out with data tips. Those resonances are a function of the blade modes vs. other body modes of the table will be more clearly identified once we damp them, so I'll wait for that. As far as the GS13 can location, we expect its resonant modes to be around ~1.2 and 1.6 kHz. However, with the accelerometer close to the GS13 can there don't really appear to be any features in the data other than a 1.1 kHz mode. Looking back at the blade-hit location, there're feature more akin to those expected. Again, we'll see once we apply the damping material what modes are truly of these structure, but the message is that we can stick with the one blade-spring location for the accelerometer in each corner. Much more to come tomorrow! ------------ Details: The data has been exported from the B&K laptop and ported over to Sei SVN, the data lives here: /ligo/svncommon/SeiSVN/seismic/HAM-ISI/H1/HAM6/Data/BandK/ 2016-04-05_1315PDT_H1ISIHAM6_BladeSpringHit_Corner3_X_Reference.txt 2016-04-05_1315PDT_H1ISIHAM6_BladeSpringHit_Corner3_Y_Reference.txt 2016-04-05_1315PDT_H1ISIHAM6_BladeSpringHit_Corner3_Z_Reference.txt 2016-04-05_1322PDT_H1ISIHAM6_GS13CanHit_H1_X_Reference.txt 2016-04-05_1322PDT_H1ISIHAM6_GS13CanHit_H1_Y_Reference.txt 2016-04-05_1322PDT_H1ISIHAM6_GS13CanHit_H1_Z_Reference.txt and the script to process the data lives here: /ligo/svncommon/SeiSVN/seismic/HAM-ISI/H1/HAM6/Scripts/BandK process_bandkdata_20160405.m ----------------- For those who can imagine the B&K software window and have gotten as far as finishing the measurement: In order to export the data after the measurement is taken, one must - go the "Validation" step on the left bar - go to the upper left corner and click the "display" button - choose the measurement number of your most recent measurement, selecting one of the accelerometer's DOFs at time - right click on the resulting plot, select "properties" from the bottom of the menu, go to the "options" tab of the pop-up window - make sure that the options are set to export as a PULSE ASCII format 4.2 (it doesn't really matter which one of the two of those options; I chose the upper one. I think the lower just adds more info to the header, which we don't use anyways). - hit the copy to clipboard button. - open up a text editor (the windows default is "notepad"), paste in the ASCII, and save as text. - You *can* then wrestle with Tortoise SVN, but I didn't bother. I just pulled out the external disk and brought it over to sensible machine, and loaded in the data there.
The high power oscillator (HPO) is currently on and injection locked. The watch dogs for both the front end laser and the HPO have been engaged. SHOULD THE LASER TRIP OUT, FOR ANY REASON, PLEASE DO NOT ATTEMPT TO RESUSCITATE IT. No real progress towards locking the pre-modecleaner to the HPO.
Completes FRS ticket 5254 NOTE - Vented BSC4 annulus with no response to PT140B, i.e. no inner O-ring leaks for BSC4 annulus volume.
J. Warner, N. Kijbunchoo, T. Shaffer, B. Weaver, T. Sadecki, J. Kissel Good progress today on the HAM6 Damper Install, a la E1600092. Starting on the Tuesday steps: 10) Remove all three chamber doors DONE 11) Pick up floor CC wafers. Take particle counter measurements and record DONE (see LHO aLOG 26450) 13) Pick up table top CC wafers. DONE (see LHO aLOG 26450) 12) Lock HAM ISI DONE 14) Install Septum Window Cover DONE 15) Evaluate, mark and Move Beam Diverter ONLY IF ABSOLUTELY NEEDED for ISI work. IF CABLES of Beam Diverter get removed, a test of the Beam Diverter function will need to be made before closeout. Beam Diverter did not need to be moved. Nice! 16) Start ISI damper install work. a) Remove outer walls and internal "window" access panels in order to access the blade-springs. DONE b) Remove Tuned Mass Dampers (TMDs) DONE We Stopped Here for the day.* c) Install new spring damper assemblies d) Measure new spring modes using the B&K System e) Retune TMDs with new info f) Reinstall TMDs * After removing the TMDs from the blades, we took reference B&K hammer measurements to be sure we'd gotten all the data we need and we understood how to use the B&K system (all of us are relative newbies to the software interface). We have pre-assembled all spring-damper assemblies, but we were at a good stopping point for the day with the reference measurements and will pick up with the installation of the spring damper assemblies tomorrow morning. Stay tuned for processed reference data. The attached image is of Jim removing the inner wall from the -X/+Y corner (to access Corner 3's blade tip), with Nutsinee's knees visible as she's doing the same to the +Y corner (to access Corner 1's blade tip).
TITLE: 04/05 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Planned Engineering
INCOMING OPERATOR: None
SHIFT SUMMARY:
LOG:
Peter was already in the PSL and Kyle was already at HAM6 area when I arrived.
15:12 (8:12) Krihna and Michael to EY checking BRS pump
15:25 (8:25) Gerado joining Kyle at HAM6 area.
15:41 (8:41) portable potty here -- they couldn't get through the tumbleweed so I think they turned around and left.
15:47 (8:47) Jason and Ed joining Peter in the PSL
15:49 (8:49) Betsy to Optics lab
16:10 (9:10) Ken the electronics guy to both ends checking air compressor cabinets (if he could get through tumbleweeds).
16:13 (9:13) Keita to HAM6.
16:33 (9:33) HVE-EY:INSTAIR_PT499 alarm handler went off
17:01 (10:01) Travis to LVEA
17:10 (10:10) TJ to HAM6
17:12 (10:12) Travis out
17:19 (10:19) Richard to LVEA
17:20 (10:20) Dave to beer garden checking on ITMY hardware watchdog
17:32 (10:32) Ken back from end stations
17:40 (10:40) Dave out, going to CER to replace WD.
17:53 (10:53) Calos working on the net box in the CER. Will affect the internet.
18:26 (11:26) Calos done
19:05 (12:05) Dave swapping out ITMY WD chassis.
19:08 (12:08) Jason, Peter, Ed out of PSL for lunch.
20:13 (13:13) TJ and Nutsinee to LVEA
22:02 (15:02) TJ out
22:27 (15:27) Nutsinee out
22:40 (15:40) Ed out of the PSL for the day.
22:43 (15:44) Jeff and Jim out of HAM6 for the day
16:02 (23:02) Jason and Peter done for the day. HPO box is left on but the shutter is closed.
This morning all three doors on HAM6 were removed.The particle counts during the removal were, with one exception, very good. There was one spike of 0.3µ particles to 8480 counts when opening the east door. The door was open but still on 4 bolts. It looked as if we stirred up a cluster of dust when the door popped open. The air flow was positive, so there should have been little contamination into the chamber.
Time | 0.3µ | 0.5µ | |
---|---|---|---|
Door On | 08:52 | 0 | 0 |
Door Open | 08:55 | 90 | 70 |
Cover On | 09:08 | 0 | 0 |
Inside Chamber - Cover on | 09:18 | 0 | 0 |
Time | 0.3µ | 0.5µ | |
---|---|---|---|
Door On | 09:27 | 60 | 40 |
Door Open | 09:31 | 8480 | 40 |
Cover On | 10:10 | 60 | 60 |
Inside Chamber - Cover on | 10:12 | 20 | 0 |
The North door was removed while I was working on setting up the east side cleanroom so no readings were taken. Checks later and a 48 hours trend (posted below) did not show any serious contamination control issues. The 48 hour trend showed only one spike of 0.5µ particles over 100 counts, recorded while the crew were doing the door removal. The two large spikes on 04/05 were caused by the in first cleaning of the main HAM6 cleanroom. These spikes were expected. In general, the contamination control around HAM6, in the rest of the LVEA and in the VEAs has been excellent. This is mostly due to the efforts of Christina and Karen.
The 4" wafer was removed and will; be sent to CIT for analysis.
WP 5814 Short version: Everything except the MEDM screens should start on boot. There is a shortcut on the Desktop (sitemap.xlaunch) to start the MEDM screens. The CP LLCV should now only change position in increments larger than 0.5%. Long version: I have set the system to automatically run on boot. I did this by selecting 'Auto Boot' and 'Auto Login' and setting the Windows user name and password under Solution -> H0VACEX -> SYSTEM -> Settings tab (see boot_settings.png). I have also set the PLC to run on boot. I did this by right clicking on PLC1 and clicking 'Activate Boot Project...' and then 'Autostart Boot Project' (see plc_settings.png). I have also set the EPICS IOC to start on boot. I did this by creating a shortcut to C:/SlowControls/TwinCAT3/Vacuum/Target/H0VACEX/h0vacex_start_ioc.bat in C:/TwinCAT/3.1/Target/StartUp. I also installed XLaunch (a component installed with Xming). I uninstalled and reinstalled Xming to do this. I received errors trying to reinstall Xming and it only worked after I restarted the computer. I put a link on the Desktop, sitemap.xlaunch, which opens a small MEDM with a drop down menu containing links to the new MEDM screens. I tried to get the MEDM screens to start on boot, but when they did all that showed up were white rectangular screens. I'm not sure why, this worked on the test computer in the EE lab. I have updated the Vacuum library and PLC code to put a dead band on the control of the CP LLCV. An additional channel was made, H0:VAC-EX_CP8_LIC500_LLCV_POS_CTRL_PCT_DEAD_BAND and is currently set to 0.5. This means that H0:VAC-EX_CP8_LIC500_LLCV_POS_CTRL_PCT will only change if its current calculated value differs by its last output value by more than 0.5. That is, the CP LLCV should only open and close in increments larger than 0.5%.
FRS ticket 5240
The LED error condition has been triggering on the ITMY HWWD recently. This measures the LED drive current as a voltage monitor, and trips if any of the 6 LEDs voltages drop below a threshold. Jim and I checked the long db37 cables which run from the HWWD in the CER to the M0 top satellite amps in the LVEA biergarten. All connectors are seated and tightened down.
We replaced the HWWD chassis SN1301708 with SN1301704. Both have been recently tested. The LED problem is still present with the new unit. We'll keep the FRS open.
Krishna,
I noticed that the tilt-subtraction with BRS-2 was being affected by the lack of a response inversion filter for the STS-2 (pair of poles at ~8.2 mHz). Therefore I tried a filter with a pair of zeros at 8.2 mHz and to prevent it running away to infinity, I put in two poles at 0.1 mHz. Unfortunately I turned it on without thinking much about it and it seems to have produced huge numbers saturating some filter banks. I have since produced a much more sensible filter with a pair of zeros at 8.2 mHz and a pair of poles at 3.5 mHz which should only increase the low-frequency numbers by a factor of ~7 instead of ~70,000!
Apologies for this issue.
It's not taking any useful data right now. Rather save the SLEDs life time.
J. Oberling, P. King
LHO WP #5810.
Short Version
IT'S ALIVE!!!
Long Version
Today we began the HPO turn on procedure. First things first, we put a water cooled 200W power meter head in the main beam path after the PMC (see attached picture 200W_power_meter_20160404.jpg). This way no light is getting out of the PSL enclosure until we are ready for it, and once we have a recovered PMC we simply need to plug in a power meter so we can set the available power with IO_MB_HWP1.
In an abundance of caution we began by repeating the HPO Laser Head turn on test we did back on March 22nd (see LHO alog 26193). Everything came up without issues, just as in the previous test. We then removed the HR mirror that sat in front of the HPO Output Coupler (OC); it is currently left in the HPO box, set to the side and out of the beam path (see attached picture OC_removed_20160404.jpg). This taken care of, we set all the laser heads to 30 A of pump diode current and turned on the HPO. As the currents were coming up we saw the HPO go through the first, low power stability range, but nothing at the high power stability range (which is what we expected, not enough pump power yet). We repeated this at 35 A, 40 A and 45 A with the same results. Getting a little discouraged we input the previous operating currents (from Peter's alog here) and turned on the HPO. This time the laser came on! It started at ~120 W of total power (~70 W in the forward direction and ~50 W in the backward direction). We then increased the diode currents individually until they all read ~100% power and then let it sit for ~10 minutes so everything could come to thermal equilibrium; the output power settled at ~145 W. As a final check we moved the pump currents up and then down slightly (~0.4A in both directions) to check that we were on the correct side of the stability range; we are.
Now with a running HPO we decided to try injection locking the 35 W FE. Following the instructions in T1200259 we powered up the FE laser and pressed the "Lock" button on the PSL Beckhoff computer control screen. It locked, without issue. Total output power now read 208 W. This seemed really high, and looking at the HPO diode currents it seemed the diodes for head 4 were suddenly delivering more power without changing current; power changed from 100% to 105% at the same diode current. This is very odd and we currently cannot explain it. We lowered the current back to 100% power and the total power out of the full system was reading 183 W.
It was at this point that we realized we had forgot to align mirror M5, which controls the beam out of the HPO. This is important as this mirror needed to be tweaked when the HR mirror previously in front of the OC was installed back in 2014. In the process of trying to realign this mirror we lost the alignment entirely and spent the afternoon trying to recover it. We eventually turned the HPO off, reinstalled the HR mirror and used it and M5 to mostly recover the alignment. We are currently at a state where spending more time recovering this alignment is a waste as we are going to have to do alignment work with the HPO on anyway.
We left the system with the HPO OFF and the lid on; as a safety measure we wrote down the last used HPO diode currents and set the current input fields to 0 A. We the left the FE laser ON but shuttered, so there is no laser light available to the IFO. We left the environmental controls (HEPA fans, AC, and Make-Up air) ON.
Tomorrow
Tomorrow we plan on once again removing the HR mirror and then beginning the recovery of the PSL subsystems downstream of the HPO; PMC, ISS, FSS, DBB (likely in that order).
Excellent news!
Jason, Peter -- nice work! This is really excellent news!
Excellent news! Well done.
This really good news!!!
F. Clara, N. Kijbunchoo, R. McCarthy, V. Sandberg, P. Thomas
As per Work Permit #5809, we performed maintenance inspection and repairs as necessary on the EtherCAT DC Motor Controllers and Rotation Stages located on the PSL and the TCS X and Y arms. The EtherCAT chassis "X1-PSL-C1" described in D1101114 and located in the LVEA CER was removed to the Electronics Lab. There the chassis was inspected, the wiring was verified, and all connections were re-dressed and re-seated in their Beckhoff or spring-key connector blocks. A number of build quality issues were found (e.g., too short a stripped wire, wires not properly seated, and ground screws not tight enough) and corrected. The chassis was reinstalled and powered up. The PSL rotation stage was checked by hand entry of commanded angles and found to operate satisfactory. We left it at the minimal power angle, ~ -24 deg. The TSC rotation stages were tested with a script that generated 100 random angles to be requested and recorded the final angle position. Results are shown in the attached plots. In all cases the measured angle was within 0.1 deg of the requested angle. The angles spanned the range of -90 deg to +90 deg. There was no observed problem dealing with angles in the neighborhood of zero degrees.
Excellent!
F. Clara, N. Kijbunchoo, and V. Sandberg
We measured the DC resistance of the cable plus motor coil at the input connector to the EtherCAT chassis for each of the rotation stages. These were then loaded into the corresponding TwinCAT CoE parameters with the values shown below.
Measured cable + motor coil resistance:
Subsystem Measured Resistance [ohms] DC Motor, CoE Index, & Value (old value)
========= ========================== =======================================
PSL 57.6 L1_2 Ch.1 8020:04 5760 (6350)
spare --- L1_2 Ch. 2 8030:04 set default to 5900
TCS X 56.9 L3_4 Ch. 1 8020:04 5690 (6350)
TCS Y 59.0 L3_4 Ch. 2 8030:04 5900 (6350)
These are stored in the Beckhoff CoE table under, e.g., 8020:04 in units of 0.01 ohm.
A sampling of 50 random angles were then sent to each of the rotation stages. Plots of the requested angle vs the actual angle and the residuals are shown in the attached figures for the PSL, TCS-X, and TCS-Y rotation stages.
The Beckhoff EtherCAT + TwinCAT software + interface electronics (cable drivers) + Newport Rotation Stage appears to work under conditions of normal CDS network activity.
Keita, Sheila, Jenne, Evan, Hang
The mystery about AS 90 WFS may be (partially) due to the drift of dark offset.
We checked a time of today when IMC-MC2_TRANS_SUM was 0, and found that the output of a segment of AS 90 WFS was as large as ~100 cts when it was dark. As a comparison, the total sum of 4 segments at DC_readout was ~ 1000 cts. After corrected for it, the AS 90 WFS signal looked more reasonable.
We then checked the dark offset of the same WFS roughly days ago, and found that it drifted by ~ 100 ct (see, e.g., two seg1s in the attached plots). This large drift seemed make it not a very reliable sensor as the drift of dark offset was at least comparable to the signal we looked for. If we wanted to use it, we might have to check the offset frequently.
I looked at the dark offset drift of the AS 90 signals over several hours Monday night, when the PSL was shuttered. The dark offset for AS 90 signals jump around by a few tens of counts, even though the amount of light into the vacuum is not changing. Over the same period, the AS 36 and AS45 signals are stable to within one count. When we are at the Increase Power step of lock acquisition, the signals change from ~100 counts to several hundred counts when the power is changed from 2W to 8W, so these offset jumps are 10% or more of the signal size. Seems no good.
I only looked at the AS_A signals for this time stretch, but I have no reason to suspect that AS_B will be any different. Note that the last ~hour and a half of this time stretch the cleaning crew was working on getting HAM6 ready for the vent, so it's not surprising that we see offset changes, since we know that happens any time anyone is near the HAM6 racks.