Joe and I commenced the cleaning party on the two SM2 this morning. It took a couple of hours to pre-clean the tapped holes on each mirror. Very little debris came out with the Liquinox swabs and, as usual, rinsing the Liquinox was good, clean fun. Once we were satisfied with the pre-cleaning of the holes, we stopped for lunch. After lunch, we prepared the isopropanol bath in the designated pan under the fume hood. We wanded the tapped holes and then each of the honey-comb pockets on the back of the first mirror. Next, we transferred the mirror to a flow bench and removed the Teflon disk so we could perform the isopropanol rinse/N2 blow-off of the optic face. After a brief inspection, we replaced the Teflon disk and set that mirror aside. Then, we ran through the wanding etc. on the second mirror, inspected it and replaced the disk. The mirrors are setting on the flow bench until they are needed. We came out and talked to John about what he wanted us to do WRT FTIR. John asked to see the procedure that we had followed and asked a few questions. After hearing our responses, he is willing to accept the SM2s as clean without an FTIR.
There were some network outages today starting at 10am local time. This was do to a reconfiguration of the network to provide more flexible routing of traffic between PNNL (local transit provider) and ESnet (main ISP). The testing of changes was completed around 11:15am local time. The main work is done there are a few minor changes which may interrupt the network again. These interruptions will be on the order of seconds. I will post a follow up comment when the work is complete.
Looks like the PSL went offline yesterday (10/28) at around 11:09 local. Sheila will try to get in and restart it later today. This is Justin not Gerardo.
At about 9:03 PDT, there was a timing error detected by the DACs on h1sush2b, which resulted in all outputs being set to 0. The only remedy is to restart the IOP model, so I killed h1susim, restarted h1iopsush2b, then started h1susim again. This has cleared the DAC error. I burt restored to 8:00 PDT.
A simple 'sudo /etc/startWorld.sh' will do all this
Checking the Dial Indicators this morning-- Largest horizontal shift is 3 mils (<0.1mm). So HEPI has not moved.
Stopped test code running on opsws8 (alog 8207) at 09:11 (controlling dust monitors in the optics labs) Started code in 'screen' on h0epics: cd /ligo/home/patrick.thomas/epics/iocs/dust/dust_met_one_227b_comp_ctrl/iocBoot/ioch0_dust_lab/ screen -S 'h0dustlab' -d -m ../../bin/linux-x86_64/dust st.cmd Ran burtwb on /ligo/cds/lho/h0/burt/2013/10/28/00:00/h0dustlab.snap This is the only IOC currently running on h0epics. The other dust monitor IOCs are running on h0epics2. Will let this test run for a few days before migrating the other dust monitor IOCs to it.
The 8 coefficients for MIMO tilt decoupling have been installed. Since we don't have an alignment matrix like in the ISI, I installed the product of the Local to Cartesian Position Sensor matrix and the tilt correction matrix. Data has bee taken to check the decoupling results, both for the matrix resulting from SISO calculations and from the MIMO calculations (see yesterday's log for details) The figure attached shows that the tilt horizontal couplings has been reduced to the order of 1e-3 for all 4 transfer functions (on the left of the vertical gray line, and under the horizontal one). X to RX results show why MIMO calculation is slightly better. The effect could be larger if the RX to RY couplings were higher on other units. We'll review these results at tomorrow's testing meeting, and will likely start working on a procedure to automate this process.
Modified /ligo/cdscfg and /opt/cdscfg setup scripts to allow a user to use an alternate userapps checkout. To use, check out a private copy of the userapps repository into your home directory and then: source $HOME/userapps/etc/userapps-env.sh or set the PRIVATE_USERAPPS environment variable and source the top-level CDS configuration script: export PRIVATE_USESRAPPS=$HOME/userapps source /ligo/cdscfg/stdsetup.sh
- HEPI actuator attachment at ETMX: Greg G, Jim W - MC Baffle assembly in West Bay: Mitch R, Gerardo M. - Work in H2 PSL Enclosure: Peter K. - Transfer functions on ITMX SUS: Arnaud P. - TCS Steering mirror cleaning: Jodi F, Justin B - Pump bagged vacuum joints on GV6: Kyle R - Restarting ECATC1, working on PLC2: Sheila D - Removing beam tube insulation: Contractors - Work in TMS lab at EX: Cheryl - Roof top camera was blown or bumped resulting in loss of range of motion, Richard M notified
After serious searching (OSB Optics Lab twice, VPW) this morning, Greg and I were able to locate the cleaning fixtures shortly after lunch: they were still attached to some obsolete steering mirrors that were not stored with the bulk of the TCS parts. I carefully removed the tooling (Teflon disk, gussets, handle) from the old mirrors and put it on the new mirrors. This task was completed at about 3:15 PM. We will try cleaning the mirrors tomorrow.
Commencing Matlab TFs on IM1 at 3:00 pm to debug new HAUX support in the script.
Attached are damped and undamped top mass transfer function results taken last friday night for SRM phase 2b (test bench, surrogate glass mounted)
Undamped TF are matching the model, cf the first attachement. Damped TF are slightly different for longitudinal and roll (cf second attachement), which might be due to a discrepancy between the design gain and the medm one. Since the filters will be redesigned soon, there's nothing to worry about for now.
The last attachement shows a comparison between SRM phase 1b (staging building), the last phase 2b measurement, and PRM phase 2b back in february.
The only 2 differences seen are for the pitch DOF (5th page)
1) SRM gain at DC is slightly higher by a factor of ~1.3 compared to PRM
2) Phase 2b shows a slight increase of the longitudinal to pitch cross-coupling at 2.8Hz compared to phase 1b
Other than this results are good.
Travis will be centering the lower stages osems this afternoon in order to check the actuation chain, before closing phase 2b.
Note : SRM has the "side" osem, which drives the "transverse" DOF, on the opposite side as PRM. The gain for the side drive is then different, (+1), cf the picture attached as reference.
The new settings have been saved under the "safe" snapshot.
I made changes to plc2 on h1ecatc1, adding variables and function blocks for the PLL boards and the FindResonance library, which can be used to automate the search for the VCO frequency offset needed to bring the arm cavity on resonance using ALS. I did not make any changes to the system manager, this will need to be done for the PLLs once the corner 6 chasis arrives.
Everything is in SVN.
When I try to use the installation scripts, activate and run always fails for PLC3. (TCS stuff) However, I can go the PLC control, loggin and run and it seems to run fine.
Continuing the work on HEPI to ISI tilt decoupling for HAM2, I found that MIMO calculations were necessary to properly account for the non negligible cross couplings and minimize all four tilt couplings mechanisms (X to RX, X to RY, Y to RX and Y to RY). Also useful to highlights how unsymmetrical the tilt coupling behavior is (almost two orders of magnitudes??) The document attached is a draft of a procedure accounting for the cross couplings. The last pages show predicted/calculated results. The correction matrix will be implemented tomorrow and we'll take data to see whether this works as predicted.
Summary: no coupling hot spots (e.g. coil drivers, satellite amps, or feedthroughs) were found in the OSEM systems; L4Cs were the coupling hot spots in HPI and ISI systems. Magnetic coupling to sensors was not high enough to be a problem in normal operation but is a problem for our large magnetic injections. Cable coupling is worse for AOSEMs than BOSEMs.
Our magnetic coupling experiments suffer from our inability to rely on L4C, T240, BOSEM, and AOSEM channels to measure motion because of direct magnetic coupling to these sensor channels. I studied coupling to these channels, particularly whether there were local sites (e.g. to magnets in the sensors, or to feed-throughs) or diffuse (e.g. to cables) in order to help us interpret coupling results.
The technique I used was to generate a global field (roughly the same over the entire system) with a distant coil and a local field at a slightly different frequency with a local coil. A magnetometer was set up at the tested location (e.g. a feedthrough) and the fields from both coils were adjusted to give equal amplitudes at the magnetometer (Figure 1). If the feedthrough was the dominant coupling site, then the peak from the local field would be comparable to the peak from the global field. If the local field made a much smaller peak in the OSEM channel than the global field did (as for the Figure 1 location) then the feedthrough could not be the dominant coupling site.
In addition to balanced magnetic fields, I also sometimes increased the local field to see how much leeway there was at the local coupling site. Figure 2 shows that coupling at the coil driver is lower than elsewhere even when the local field is much larger (good design!).
Results (good between 10 and 40 Hz)
SUS: I tested for local coupling at the chamber, the feedthroughs, the satellite amp, and the coil drivers. Coupling at each of these locations was small compared to global coupling, suggesting that the cable runs were the most important coupling sites. Unfortunately, the worst coupling was to the AOSEMs on the PUM, the closest OSEMs to the test mass. To see if there was a specific problem with the AOSEM cabling etc., I switched PUM AOSEM and UIM BOSEM cables at the BSC feedthrough, and the coupling was still high, so it is not because of a bad cable, etc. Tests of an isolated AOSEM and its in-chamber cabling indicated very low coupling, so it seems that the EMF generated in cables influences the AOSEM signal more than the BOSEM signal. For ITMY we are getting a coupling to the AOSEMs of as high as 0.05 m/T for a15 Hz field that is roughly uniform in the LVEA and ebay. The BOSEMs are lower by at least an order of magnitude.
ISI: Coupling to the L4Cs is greater than to the T240s or the GS-13s. It is good that they are only on the first stage. Coupling to the L4Cs dominates over cable coupling and ebay coupling. I am not sure what dominates in the T240 and GS-13 sensor channels because the coupling was so low.
HPI: the dominant coupling is at the L4Cs.
Robert S., Richard McCarthy.
Summary: scattering evaluation photos of ETMX, the ACB, the cryobaffle and the p-cal periscope show no major problems.
Figure 1 shows approximately what the beam spot on ETMY “sees”. I use these photos to look for potential scattering problems; the camera is placed very close to the center of the optic where the beam spot will be. The flash mimics the wide-angle scattering of interferometer light from the beam spot and lights up regions that will retroreflect interferometer light back to the beam spot, where it can recombine with the main beam to produce scattering noise. Of course the technique is limited by different angular distributions of scattered IFO light and light from the flash, as well as by color differences, but it gives a rough idea. The camera is about 10 cm in front of ETMX, so it sees a little more through the hole in the baffle than the beam spot on the optic will. This is why you can see the baffle walls on either side of the baffle aperture. I think the actual view of the beam spot is just narrow enough that it doesnt see these baffle walls. The "H" at the middle is the sheet metal cover at the spool with an "H"-shaped aperture in it for IAS. The flash is reflecting off of this sheet metal cover.
I dont see any terrible glints coming from the pcal periscope, though it reflects more than the cryobaffle behind it. I think the lower reflectivity of the cryobaffle is not needed here, but is needed for its twin baffles near the ITMs where the angle is smaller. The white spot at the right of center is one of the periscope mirrors, which is directing the flash to a reflective surface at the port. This is a reminder that we need to be careful about backscattering from the pcal boxes and other equipment at the ports.
Figure 2 is an approximation of what ITMX will see of ETMX. The camera is in the beam path and so the flash is reflecting off of the ETM. Of course the camera is a lot closer than the ITM, so the camera sees more through the baffle aperture of the region around the ETM than the ITM will see. The ETM is covered with first contact, which produces the irregular reflections. The view of everything, except the ACB, is blocked by the cryopump baffles. The ACB looks good, the light spots are the photodiodes.
We removed the large eddy current damper magnets from the UIM blade spring in the quads, and there remains a question about whether we can use a small magnet to attach a tuned mechanical damper to the blade spring. I have measured the relative permeability of a blade to be greater than 6 (here), so it is possible that the blade spring will “amplify” the moment of the magnet. For this reason, I attached a magnet to the blade spring at the proposed location and measured the magnetic moment of the blade spring in my moment balance (Figure 1).
Adding the 0.013 J/T 2x6 mm magnet to the blade spring increased the moment of a single blade spring by 0.12 J/T in the beamline direction. The resulting displacement noise of the test mass from ambient fields acting on a single such blade spring would be about 3.1 e-20 m/sqrt(Hz) at 10 Hz (assuming 2e-11 T/ sqrt(Hz) ambient magnetic field). For this reason, I think that we should not use 2x6 or 2x3 magnets on the blade springs without a cancelation scheme. As a comparison, the large magnets used by the BOSEMs have a moment of 0.72 J/T, but with all of the cancelations from orientation, I have estimated that the remaining moment (with no orientation errors) would be about 0.2 J/T (here). If there are no other ways of attachment, I can try cancellation schemes, but they may not be very successful because one of the magnets is in contact with the ferromagnetic material.
We have often had the discussion about "what blend filtering the ISI should be in while working on HEPI tilt decoupling". The first plot compares the response from "HEPI IPS" motion to "ISI GS13 motion" for two different configurations: "ISI damped" and "ISI isolated". The low frequency part of the transfer functions don't change, indicating that the tilt coupling ratio is unchanged. The second plot compares tilt decoupling results for several blend configurations. The black line is the reference (no tilt correction). The exact same tilt-correction factor was used for all other measurements (for the records, the decoupling factor used was 7.0%, calculated from the measurement shown in the first plot). This second plot shows that the tilt correction implemented works just as well for all blend configuration, which is really good news.
We looked at Arnaud's TFs taken last night on SRM and determined that there was something unhealthy in the vertical/roll, pitch directions of motion as seen by the top BOSEMs (T1, T2, T3). Travis investigated and tweeked a few seemingly non-issues:
Slightly offcenter T2, possible loose cable connections (ha!), possibl droopy C3 cover at the top. A quick look at DTT roll and vertical and these don't seem to have helped... Arnaud is looking at verifying direction of excitation motion now.
Richard found that the binary cabling was pointing incorrectly. He recabled according to spec and actuation has restored to the BOSEMs we couldn't see actuation on previously. A look at a new roll DTT TF and things look healthier.
HOWEVER - Arnaud also pointed out that the side actuation has a sign flip somehwere. Travis dug into documentation and found that the side BOSEM is mounted on the incorrect side of the suspension. We'll switch it and retake the full TF set this weekend.
Side osem actuates in the right direction, so top mass actuation is now fully functionning. Phase 2b transfer functions will be running tonight.
In fact, the SRM is supposed to have the side BOSEM mounted opposite to most othere HSTS, with an appropriate sign flip in the side coil gain, functionality is as expected.