HAM3 is now locked. Went smoothly, but I noticed that the soft covers on both sides are very tight, like they are undersized, and hit the corners of the ISI. I'm not sure exactly when the west door was removed in relation to MC alignment activities, so it would probably be worth if for those teams to take another look at their alignments. With the ISI unlocked, a dragging cover is more than enough to pull the ISI out of position. Covers were pulled completely clear of the ISI before locking, so it should be back to its nominal position, now.
Chamber cleanroom moved from BSC9 to BSC5 so that leak hunt at the dome could begin as soon as a soft roof is available. (Clean garb shipment due tomorrow.)Special care was taken with GV20 (hard closed) and GV19 (soft closed by Kyle).
East door removed.
X2 accumulation from Friday.
I had to go home in the afternoon, so we couldn't do any alignment work, which should be done on Monday. EY is in laser hazard.
We did check the beam diverter. (This one should be replaced anyway, but not before moving the entire EY to H1 position. https://services.ligo-wa.caltech.edu/integrationissues/show_bug.cgi?id=83)
Beam diverter is connected to F2-2C2 feedthrough on BSC6. From outside, pin1-14 is the coil, pin2-15 is one of the reed switches (SW1), and pin3-16 is the other (SW2).
When pin1 is positive and pin14 negative, the beam diverter moves from "open" position where the mirror is tipped toward EY to "closed" position where it is tipped away from EY.
However, we've found that pin1 is short-circuited to the chamber and the building ground. This means that you need a floating supply to drive it (which is not a problem for a proper driver but is a problem when you're using a power supply with one of the pins grounded). If for example the negative is grounded and if you try to "close" it by supplying positive to pin 1, all current goes to the ground via short circuit, not the coil.
Anyway, the problem is somewhere in the chamber. This should be checked when EY is moved to the H1 position, most likely an offending cable should be replaced.
One of the reed switches is still stuck, despite a previous report that it was somehow magically unstuck: https://services.ligo-wa.caltech.edu/integrationissues/show_bug.cgi?id=2
When in "open" position both of the switches are closed. When "closed" one of the switch opens.
After building five Beam Diverters, I noticed issues with the Reed Switches. I was able to get them to work after orienting the Sensor Magnet in a position which was related to polarity (Chris Guido at LLO noted that the position of this little magnet determined Switch functionality). So, it will be good to check this Beam Diverter out when we pull it out.
M. Barton, D. Bridges, J. Kissel These are the beginnings of the aLIGO Production Matlab Dynamical Model of the Output Mode Cleaner Suspension (OMCS). The last (known) attempt at this was during the Final Design Review: T080138, but the OMC breadboard suspension design has changed dramatically since (see the aWiki for details). Mark has brushed off the dust and greased the wheels of the core model components, ${SusSVN}/sus/trunk/Common/MatlabTools/DoubleModel_Production/ssmake2MBf.m ${SusSVN}/sus/trunk/Common/MatlabTools/DoubleModel_Production/symbexport2full.m, and Derek and I have gathered enough information from his SolidWorks Assembly (D0900295) for me to create new parameter files: ${SusSVN}/sus/trunk/Common/MatlabTools/DoubleModel_Production/omcsopt_metal.m ${SusSVN}/sus/trunk/Common/MatlabTools/DoubleModel_Production/omcsopt_glass.m With the usual-style wrapper function, ${SusSVN}/sus/trunk/Common/MatlabTools/DoubleModel_Production/generate_Double_Model_Production.m, I've compared all three parameter sets (where omcsopt_doublep.m is the FDR version from T080138), and attached the TOP to TOP transfer functions. Details of the differences in parameters are listed below, but the major differences are (1) Bug fix with moments of inertia (a mix up between SolidWorks and Euler Suspension coordinate systems that's now resolved) (2) M1 to M2 wire length is shorter by 3 [cm] (3) The suspension break-off point (the "d2") of the breadboard increased from 3 [mm] to 40 [mm] (4) We're now correctly accounting for the imperfect flexing of the wires (i.e. turned on the flexure correction) With these differences, the rotational dynamics are totally different (modes frequencies are all shuffled around), but not nearly as bad as I thought it would be because of the shorter wires and larger d2 (with which I would expect *higher* mode frequencies). I'll be interested to see these parameters plugged back into the full Mathematica model to check out the mode shapes. ANYWAYS, Step 2 is to get some data once we've built up one of these suckers, and compare. Then we'll do the usual compare and contrast to see if the model's in any way accurate*. Step 3 is to develop the whole suite of software that we have for the other suspensions, i.e. transfer function comparisons, actuation range calculations, residual ground estimates, etc. Step 4 is to design damping loops (though not much thought is needed, just a little gain tuning on the basic loops should be fine for now). ... but at least we're done with Step 1 finally! ----------------- * Note that there's very little difference in dynamics between the glass and metal parameter set, but I didn't know that before getting started, so I created both. Once we get a few measurements under our belt, if we see that the assembly variance outweighs the dynamical variance, we can toss one of the sets in the trash and move on. Detailed breakdown of parameter differences: 'Fields' 'omcsopt_doublep' 'omcsopt_metal' 'Abs. Difference' 'Rel. Difference [%]' 'm2' [ 6.895] [ 6.971] [ 0.076] '1.1%' 'I2x' [ 0.139] [ 0.0152] [ -0.1238] '-89.1%' 'I2y' [ 0.0164] [ 0.148] [ 0.1316] '802%' 'I2z' [ 0.15] [ 0.136] [ -0.014] '-9.33%' 'l1' [ 0.25] [ 0.2496] [ -0.0004] '-0.16%' 'l2' [ 0.25] [ 0.22] [ -0.03] '-12%' 'r2' [ 0.000102] [ 0.0001005] [ -1.5e-06] '-1.47%' 'Y1' [ 2.12e+11] [ 2.119e+11] [ -100000000] '-0.0472%' 'Y2' [ 2.12e+11] [ 2.119e+11] [ -100000000] '-0.0472%' 'ufc1' [ 2.38] [ 2.34] [ -0.04] '-1.68%' 'stage2' [ 0] [ 1] [ 1] 'Inf%' 'd0' [ 0.0018] [ 0.001] [ -0.0008] '-44.4%' 'd1' [ 0.0015] [ 0.001] [ -0.0005] '-33.3%' 'd2' [ 0.003] [ 0.0402] [ 0.0372] '1.24e+03%' 'Parameter' 'omcsopt_metal' 'omcsopt_glass' 'Abs. Difference' 'Rel. Difference [%]' 'm2' [ 6.971] [ 6.916] [ -0.055] '-0.789%' 'I2x' [ 0.0152] [ 0.0142] [ -0.001] '-6.58%' 'I2y' [ 0.148] [ 0.132] [ -0.016] '-10.8%' 'I2z' [ 0.136] [ 0.122] [ -0.014] '-10.3%' 'd2' [ 0.0402] [ 0.0391] [ -0.0011] '-2.74%' 'flex1' [ 0.0018715] [ 0.0018768] [ 5.3147e-06] '0.284%' 'flex2' [ 0.00099653] [ 0.0010005] [ 3.9546e-06] '0.397%' 'kc2' [ 624.29] [ 619.36] [ -4.9255] '-0.789%' 'tl2' [ 0.2612] [ 0.2601] [ -0.0011] '-0.421%' 'l_suspoint_to_centreofoptic' [ 0.5118] [ 0.5107] [ -0.0011] '-0.215%' Also scripts and functions are committed to ${SusSVN}/sus/trunk/Common/MatlabTools/DoubleModel_Production/ as of this entry. Svn up and away!
Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot requested from 5 PM March 7 to 5 PM March 8. Also attached are plots of the modes to show when they were running/acquiring data. Data was taken from h1nds1. 13-03-08-06-01-47T0=13-03-08-01-00-00; Length=86400 (s) 1800 seconds worth of data was unavailable on this server 1439.0 minutes of trend displayed For .3 and .5 micron plots: 1800 seconds worth of data was unavailable on this server 1439.0 minutes of trend displayed read(); errno=9 read(); errno=9 T0=13-03-08-01-00-00; Length=86400 (s) No data output.
Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot requested from 5 PM March 6 to 5 PM March 7. Also attached are plots of the modes to show when they were running/acquiring data. Data was taken from h1nds1. T0=13-03-07-01-00-00; Length=86400 (s) 480 seconds worth of data was unavailable on this server 1397.0 minutes of trend displayed
Some of today's activity:
After Jeff K created the front end model for OMC (see alog entry 5717) the blank epics values and filters have been filled in for :
-ISI sensor input filters (copied from MC1)
-Offload ouput filters (copied from MC1)
-Osem input filters (copied from MC1)
-OSEM2EUL and EUL2OSEM matrices (calculated with the newly created make_susomcs_projections.m living in /ligo/svncommon/SusSVN/sus/trunk/OMCS/Common/MatlabTools). The lever arms/distances between OSEMS to calculate the matrices have been taken from the drawings attached.
-Coil output filters (copied from MC1)
-Watchdog DC/AC/actuator Band limiters (copied from MC1)
-ISC input filters (copied from MC1)
A new safe.snap (/opt/rtcds/lho/h1/target/h1susomc/h1susomcepics/burt) has been created pointing to the saved snapshot h1susomc_safe.snap (/opt/rtcds/userapps/trunk/sus/h1/burtfiles)
The filter file, the burtsafe file, and the make_susomcsprojection script have been commited on the svn
Still to do : CART2EUL, EUL2CART matrices, and damping filters
Finally got enough fluid pumped in and enough air out of the lines to hold a steady level at the Mezzinine Reservoir. Currently all four pumps are running in servo mode at 19psi; this drives the output to the motor at a little less than 3/4 of the max. This has been the state for a little over an hour. I just checked the Mezzinine and the level remains constant and a thorough check of the LVEA reveals no leaks. The system will throttle back if a block occurs and should not see any pressure spikes; it will shut down if fluid levels drop too low.
The polarization into HAM 2 is changed inside the PSL table using an additional wave plate. After aligning the IMC yesterday back to flashes, today we checked the beam on the periscope mirrors in the PSL and on IMC2. It is reasonably well centered on the periscope mirrors (within 2-3mm) and as far as we can tell within 2mm centered on IMC2 in Yaw and within 1mm in Pitch. We removed the wave plate after the IMC from its Siskiyou mount and bolted it on the table next to the mount. This turns the polarization back to the polarization we want in the Faraday and the beam clears the Faraday w/o a problem. One issue: IM3 is misaligned by a lot in yaw and some on pitch, barely clearing the elliptical baffle. This is consistent with the changes we saw in OSEM readings of 18mrad in yaw and 4 mrad in pitch between December and now. It is not clear where this is coming from. Other mirrors on HAM2 don't show these large changes but we also had to tweak IMC1 and 3 by a few mrad to close the beam path. Something to keep an eye on. We left the polarization in this state; Chris and Rodica will continue alignment and power measurements next week in this stage. Joe Gleason, Deepak Kumar, GM
Joe Gleason, Guido Mueller, Deepak Kumar We've installed the ALS light pipe. Richard has installed the lock and tag #2097, key# 3123. Much like the Main beam pipe the ALS one shows a horizontal offset from the upper periscope mirror and is fully offset in the PVC pipe in order to pass through. This may require re-positioning the upper ALS periscope mirror if it is found the beam cannot pass to HAM1 safely.
I went out first thing this morning and took a few particle counts at HAM3 west door (all particle sizes read zero at both 6 ft and 1 ft from floor), near the electronics rack near HAM3 (hits above 150 on all particle sizes at both 6 ft and 1 ft from floor), and in the gap between BSC2 and HAM3 (hits above 25 at all particle sizes 6 ft and 1 ft from floor with cleaning going on at HAM3 east door). I wrapped the electronics rack/cable trays in CPStat and took additional particle counts (all particle sizes read zero at both 6 ft and 1 ft from floor). The cleaning crew was working on second cleaning while I finished up the rack. Mick and Ed closed the gap in the HAM3 cleanroom by running a sheet of CPStat underneath the beam tube. The anticipated order of events for the rest of the morning is as follows: 1. Remove west door from HAM3/replace with soft cover (Randy and crew) 2. Remove hard cover from HAM3/replace with soft cover (Randy and crew) 3. Remove and gently lower BSC2 curtains (Randy and crew) 4. Let dust settle (20-30 minutes) 5. Damp mop floors with special attention at HAM3 west door and at the border between cleanrooms (Karen) 6. Check particle counts (Mark L.) 7. If counts good, proceed. If not, remediate. (Randy and crew) 8. Cleanroom and contamination check and photos in HAM3 (Robert S.) 9. IAS set-up (Jason) 10.Prep for HAM3 east door removal (Randy and crew) Dale is operator: he has this list and will serve as point of contact.
This morning I turned on the CPSes at BSC2 to see if they all survived, and it looks like one didn't make it. The horizontal St1 sensor on corner 1 is reading 32k counts, which probably means a broken in-vacuum cable. I went out to the chamber and checked all of the in air connections, which looked good. I then swapped the horizontal and vertical cables at the box to rule out the board, but the problem followed the probe. At this point it could be the in-air cable (they rarely fail), the feed-thru (seems unlikely), the in-vacuum connection (which can't be accessed right now, there's a hard cover on the chamber) and the probe. I'll try to swap the in-air cable later today, but the rest will have to wait until we can get back into the lower part of BSC2. We can still float the ISI and HEPI, but no post-flight transfer functions until this is resolved. I am a little surprised that this particular sensor broke, because the horizontal sensors are better protected than the verticals, and this corner had a better routing scheme than the other two.
After more investigating today, Hugh, Mitch and I discovered that the allegedly bad CPS seems to be innocent. When we swapped in-vac cables, the channel in MEDM didn't change, as it would have if the probe were broken. This leaves the feed-thru and the in-air cable. IAS and SUS are currently working in HAM3, so we will wait until tomorrow to try swapping more parts on the in-air side.
More investigating this morning, turns out the in-air cable was bad. Replaced it, and all the sensors are now working.
This time without any IOCs running.
Some flashing was seen on REFL, but without the MC servo, no locking.
The IMC alignment did not survive the recent vent, swap of dummy mass from HAM2, and installation of PRM and PR3, and bellows removal from HAM3. Changes to the PSL have changed the input pointing into the vacuum system. HAM3, while in theory seems to be OK, has not been fully vetted at this point, and there are some changes in the way MC2 is hanging in it's cage. Changes to HAM2 include: - a rotation of ISI by 19um - an elevation change of ISI by 30um - a shift of HEPI by 4um East (so small, likely thermal) - an elevation change 1-4um in HEPI (miniscule) MC1 - length -9um - pitch +549urad - yaw +307urad - MC1 also had a side OSEM flag that was touching but is centered now MC3 - length -42um - pitch +199rad - yaw +95urad HAM3 - bellows removed. MC2 - length -11um - pitch +89urad - yaw -12urad PSL - beam from PMC is a bit high on it's alignment iris, and significantly better centered after the room temperature was lowered by MichaelR, and the laser is again thermally stable. - alignment shift on my second alignment iris on the leakage beam from the bottom periscope is now attributed to a mix of changes made to the pickoff beam splitter for that iris made while I was out with the flu, and the beam shift seen coming from the PMC. - the current beam coming from the PSL doesn't hit either set of marks we used for our previous alignments. The PSL beams marked on the wall have two locations marked - one in-air that we used to do the original alignment of the IMC, and one under-vacuum when the IMC was locking well. The PSL beams are now splitting those marks. While some of these change seem small, I have concerns that a stuck or damaged sensor could be relaying false information about the state of a system. I believe it's possible that the tilt and rotation of the HAM2 table, when combined, is having a greater effect than just looking at the one set of sensors for each degree of freedom.
[Keita and Kiwamu]
Details are coming soon.
EY stays laser hazard tonight. Alignment of the red beam path will be performed tomorrow.
PZT mirror: PZT1 (1" holder) and its controller were replaced with the proper ones.
Newport mirror mount for 1" mirror was a hair larger than the optics. Corey tried both our green optic as well as Newport 1" optic, and both were loose to the point where the mirror would fall off of the mount.
I shimmed the mirror using about a half-inch length of a plastic bag strip and it was perfect fit. Not loose, not overly tight.
We first restored the offset to PZT2, zero-ed the offset to PZT1, moved PZT1 so the beam is roughly centered on all three irises on the ALS table, and the beam was already hitting one of in-vac green QPDs at that point. I played with the offset of PZT1 so that the beam hits both of the in-vac QPDs, turned on the QPD centering servo, and we were back.
The whitening gain of the QPDs are apparently super large: I had to reduce the servo gain by a factor of 1000 to keep the QPD centering servo stable, and with this reduced gain the servo was reasonably fast. Whitening MEDM screen was all white, and I don't know if the whitening gains were burt restored, but it's not that important for the moment. I'll talk to Daniel tomorrow.
TMS/EY relative alignment: Good now.
After the PZT servo started working, we aligned the TMS to the EY such that the retro reflected beam from EY clears the Faraday and goes to ALS PDH diode. It's kind of tricky as TMS and EY are moving in the air, but it seems like on average it's good.
ISI was locked: Jim and Travis locked ISI, put an aluminum cover on EY and put a sock on EY.
Broken in-vac IR QPD2 was replaced in situ: All quadrants are working now.
The QPD in question is the one closer to EY: https://services.ligo-wa.caltech.edu/integrationissues/show_bug.cgi?id=1
QPD S/N = 3.
In the chamber, the broken QPD was removed by first detaching a rhombus-shaped face plate to expose the QPD, then plying the QPD from the PEEK/PCB/alumina plate assembly using the tip of a small screw driver that was thoroughy wiped clean. (We first tried to use our fingers but they were too big and week for the job, so we called Koji at Caltech to learn the proper technique.)
We inserted a new QPD (serial number 14) into position and put the face plate on.
These were done without removing the QPD cable assembly from the sled (nor the sled from TMS), so we don't have to worry about any gross change in the QPD position on the sled.
After this, we confirmed that all quadrants respond to light by using a head light Kiwamu was wearing while monitoring signal on dataviewer . (We also confirmed that the other QPD still works.)
Tomorrow: Will have to remove EY sock to get the retro-reflection back, and align the IR path using green leakage. Also we'll check the polarity of EY beam diverter.
PZT settings after the beam centering on the green QPDs
H1:ALS-Y_PZT1_PIT_OUT = 3480.973 counts
H1:ALS-Y_PZT1_YAW_OUT = 7083.696 counts
H1:ALS-Y_PZT2_PIT_OUT = -260.878 counts
H1:ALS-Y_PZT2_YAW_OUT = 1702.087 counts
TMS settings after the alignment
H1:SUS-TMSY_M1_OPTICALIGN_P_OUTPUT = -68000 counts
H1:SUS-TMSY_M1_OPTICALIGN_YY_OUTPUT = 53000 counts
Trends from just prior the removal of the door on Friday 8th until lock-up at ~noon today. Final values are hard to see on the plot, but are all within +/- 300cts.