HAM's2 and 3 ISI's were locked this morning. Originally went out just for HAM2, but I saw that Apollo had just finished pulling the last door at HAM3, so I took advantage. I think immediate activities are planned only at HAM2, so if HAM3 needs to be released for some reason, let me know.
J. Kissel Because the DAQ was needed this morning for measurements, we didn't resurrect the dead h1sus56 computer until the afternoon. That meant I didn't get nearly as far as I'd hoped before leaving. My apologies. The current status is reported below, and will most likely remain as such until I get back from vacation on Tuesday. Status: - Mods to the HSTS_MASTER and HLTS_MASTER library parts (and all the library parts they use) are complete -- the HSTS required a little extra thinking, unfortunately - The MC_MASTER remains unchanged, didn't get to it - The top-level HLTS models' (H1SUSPR3, H1SUSSR3) mods are complete, and they compile successfully - The newly compiled front end code for H1SUSSR3 has been installed, and is now running; PR3's has not yet been installed - Of the HSTS', only H1SUSMC1's top level model has been completed and successfully compiled. The remaining HSTS models must be reconnected as h1susmc1's has in order to compile. They are guaranteed to fail compiliation for the time being. - None of the necessary MEDM screen modifications have been made for the HLTS or HSTS. Because I do not commit to the repository until I have a complete package for Stuart, the outstanding svn status I reported yesterday remains.
Transfer functions running on ETMX ISI, starting now. Should end by 8 AM. Running on opsws6.
The LVEA test stands (h1susquadtst, h1susbstst, h1seitst) have been powered off, removed from the DAQ, and removed from the dolphin configuration. All equipment has been left in place for future use.
Kyle Pressure response immediate on PT114, PT124 etc...
We got all the Actuators connected. The last one, V3, took a while as we swapped out the suspect parker valve noted by Jim earlier. The first one we swapped in L170 also appeared to do the same pull down Jim noted. But this may have been a ringer. We had not bled the actuator yet as we wanted to compare the performance and then we learned this valve had been removed from WHAM6. However, we thought we were in trouble as the swap at WHAM6 was possibly unneeded because the problem there was that the Supply & Return hose were miss plumbed. Still, we pulled L170 and swapped in L206 and bled the Actuator for 1.5+hours. After we put the Actuator back in run mode, the Actuator drove back to a near center position so we deemed the valve problem addressed and attached this final Actuator. Dial Indicators suggested we had tilted the table a few tenths, C2/4, so we followed with some tweeks of the big HEPI Springs to bring that back to start position. Have alerted IAS/Jason that we are ready for a post Actuator look--he may be looking now.
On caveat here: since the ISI was floating we couldn't attach the corner cube to get an accurate longitudinal measurement. The total station can use a "Sheet Refelctor" mode to measure the distance to the optic from the HR face directly, but this introduces more error in the measurement. For the purpose of calculating the lateral and vertical errors a few mm of error in the longitudinal measurement means almost nothing at this distance (>33 m; a 5 mm shift in the longitudinal direction only imparts a change of <1 µm in the lateral/vertical error calculation), but I would not use this longitudinal measurement as a basis for the actual optic position (and adjust HEPI when it might not be necessary). When the ISI is locked down we will attach the corner cube and get an accurate measurement of the longitudinal position error of the ITMx (and make any decisions about HEPI adjustments at that point).
These are still running the test code. I changed the sample time to 20 seconds and the hold time to 280 seconds for the dust monitors in the optics lab, vacuum prep lab and bake oven room. These were set through the 'expert' medm screen for each.
I made PSL model changes this week to add channels to the science frame. Pre-empting next week's upgrade to matlab2012b, I modified all the mdl models used to build the code using 2012b (in some cases just adding an SVN $Id$ tag in order to make a change).
Today I tried to open the psl models using matlab2010a. I had no problems with h1psliss and h1pslfss, but h1pslpmc crashed 2010a. There are no problems opening this with 2012b. Presumably PMC uses some matlab parts with are not backwardly compatible with 2010a.
When I made the transition, the original 2010a versioned file was save by matlab with a 2010a prefix. We should always do this.
The Apollo crew and I started clean-up around HAM6 this morning. Randy and Eddy moved the squeezer table legs under the mezzanine stairs, which is an excellent long-term storage place for them. The pallets containing SEI masses and hardware were moved to less problematic spots. I located a rack for hanging garb, a shelving unit to hold dispensers and bins, and a short bench that we can use for the garbing cleanroom. I will try to round up dispensers and staging tables tomorrow.
Yesterday Keita had Alexa and I mount some optics in tip tilts to prepare for HAM6 work.
For OM1 we mounted an optic E1201104 SN3 in tip tilt serial number 2
For OM2 we mounted E1100056-1 serial number 12 in tip tilt serial number 4.
Both of these were pitched downward after installing the optic. We tried to balance them by turning the bolt that is on the bottom of the mirror holder, but found that for each of them we had to turn the bolt until it was hitting the tip tilt mount. This is similar to the situation Kiwamu described in https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=7897
We replaced the silvered, 1and 1/2" 8-83 bolts with 1 inch stainless bolts, which is what is called for in D1001396. I then thought we should try the same thing for the tip tilt assmebly that had been rejected durring the HAM1 work (serial number 9). With the optic installed and the 1 inch bolt, the mirror was pitched up, we would have needed to add a longer screw to correct it. I didn't realize that Joe had reclamped the wire in a effort to solve the large pitch seen in HAM1.
As I tried to remove the optic I chipped it, Alexa has taken photos, one is attached.
The small pice that is used to put some tension on the mirror (D1001648, the tip tilt nudge) sticks out somewhat into the mirror holder even when the bolt is loose, meaning that we put a lot of pressure on the optics to get them in. This creates small scraches along the barrel of the optic where the other two contact points are as we push the optic in, and this is where the optic chipped. This optic had no serial number and an unpolished barrel.
We have another optic and will clean it this afternoon.
Here is the link for pictures of the chipped optic (this is a class A clean optic BS99% 1064nm, ref E1000457-v1)
The chamber was cleaned this morning but the cleaning crew reported that it was quite clean. Bubba and the Apollo crew waited for a go ahead from Cheryl and Mark Barton (HAM Aux TFs?). One door was off before lunch and the second door was taken off right after lunch.
Patrick, Dave
The DAQ was reconfigured to acquire some CONLOG EPICS channels via the EDCU. A new H1EDCU_CONLOG.ini file was created by Patrick, I added it to the h1dc0 master file. The configuration change was activated by the DAQ restart at noon today.
Jim, Dave
We waited until the HAM2 SUS IM team had completed their measurements before attempting to recover h1sush56 front end. The computer was is a locked state from last night during model changes to h1sussr3. We took the user models out of rtsystab (so they would not auto start), remotely took h1sush56 out of the Dolphin fabric and pressed the RESET front panel button. The computer booted and ran the IOP model with no problems. We restored the systab and started the h1sussr3 (modified) and h1sussrm (unmodified). They both exhibited a 0x2bad DAQ status, but the IOP kept good. We then restarted the DAQ, restarted these models and all is well on h1sush56. Looks like mx_streams between the FE and DAQ was in a strange state, but why was the IOP model good?
After the DAQ restart, the h1susauxey front end went into a bad DAQ state which was cleared by restarting its mx streams.
Cheryl asked if we could get TFs for IM2, IM3 and IM4 to complete the IM1 set from 8287 before the HAM door is removed, so I set them going in parallel in three Matlab sessions on opsws7. They should be done by about 11:30 am.
All data taken OK. Individual plots attached.
There's still a bug in the script that does comparison plots so stay tuned for that.
Note that the model now includes eddy-current damping, but the values are from by-eye fitting to IM1 data, not actual estimates from magnet and copper block properties.
Chamber is released for door work.
Here's the comparison plot, including one old trace for IM1 taken by Szymon with DTT on 1/23, and the four new traces for IM1-4 taken with Matlab recently.
The consistency from measurement to measurement is very good, even between DTT and Matlab.
There appears to be a different gain error for each DOF, which is probably an issue with the model:
L: model is about 0.6 of measurement
P: model is about 1.4 of measurement
Y: model is about 2.4 of measurement
Arnaud spotted that the script error mentioned above turned out to be that plothaux_matlabtfs.m and plotHAUX_dtttfs.m were writing intermediate data files of slightly different formats. We tweaked plotHAUX_dtttfs.m and regenerated ^/trunk/HAUX/H1/IM1/SAGM1/Results/2013-01-23_1600_H1SUSIM1_M1.mat.
I will start transfer functions of top mass of main and reaction chain of ITMX phase 3a (in chamber, under vacuum) on computer opsws2 and top mass phase 2b SRM (test stand) on computer opsws0 from 8pm tonight.
Measurements will run until saturday morning. Please do not close those matlab sessions if you come over the week end.
The channel list used to get the data from the framebuilder (in generate_channel_list.m living in sus/trunk/Common/Matlabtools/SchroederPhaseTools) has been modified to match with the recent model upgrade of the QUADS. Since channels H1:SUS-${QUAD}_${L1/L2}_DAMP_${DOF}_IN1_DQ do not exist any more. They are now called H1:SUS-${QUAD}_${L1/L2}_WIT_${DOF}.
This is related to why we couldn't get data out of the frame for the ETMX on thursday
This is also related to why we couldn't see any good calibrated rotation signal with dataviewer on ETMX for the UIM and the PUM
I ment ITMX in the title
[Kiwamu, Paul]
Just before the vent today we measured new noise spectra for the frequency and length feedback paths.
The 'whitening filters' in the IMC_X path (see Giacomo's LHO alog entry 5311) were not engaged. They were put in the model at LLO in an attempt to combat the digital noise in IMC_X above ~50Hz (see Anamaria's LLO alog entry 5821) but need to be taken account of when plotting the data.
We corrected the IMC_F frequency to length calibration to give the equivalent motion of MC2, instead of the equivalent round trip length change. This has now been edited in the filter bank for the IMC_X path; the FtoL filter has been changed to a DC gain of 5.8465e-14 (dX=df x lambda/(2FSR)).
Relevant paramters during the measurement:
Common mode gain 20dB
Common mode filters: Compensation and 1st Boost
Fast gain -2dB
Slow path bypassed offsets and filters
MC2 M3 lock gain -1000, engaged filters: 150:4 and CLP100
MC2 M2 lock gain 0.06, engaged filters: 0.01:0.1, 0.03:1, Stab8:2, 300:1 and ELF80
Damping filter: resg and Ellip50 (the filter in IMC_X path was switched to the wresg filter to reflect this change)
Jeff quite rightly pointed out that there are actually 2 more filter banks in the length path for the IMC which I didn't mention in the above post.
One of these is the H1:IMC_L bank, with the following 3 filters engaged and with Gain=1: antiWhite, roll3notch and bouncenotch.
The other bank is the H1:LSC-IMC bank, with 1 blank filter engaged with Gain=1. I'm not sure yet what the purpose of this bank is but would happy to know if someone else does! I'll try to get a signal flow chart up here soon, though it may be a little trickier from off-site.
I was revisiting this measurement and I thought it might be useful to add the PSL frequency noise requirement line for comparison (see attached).
You have to use the 'Export Model to > Previous Version' if you want to save a model in the earlier format. Incompatible blocks that postdate the export file format version are converted to empty masked subsystem blocks in the exported file:
http://www.mathworks.com/help/simulink/ug/saving-a-model.html#f2-86382
The default file format is set in the preferences, or via command line:
http://www.mathworks.com/help/simulink/gui/simulink-preferences-window-main-pane.html#bs_hdzw