patrick.thomas@LIGO.ORG - posted 14:28, Tuesday 11 February 2014 (9989)
Arnaud restarting sus models
h1susauxb123 h1susauxh2 h1susauxh56 h1susauxh34 h1susauxex h1susauxey
h1susauxb123 h1susauxh2 h1susauxh56 h1susauxh34 h1susauxex h1susauxey
The in chamber power spectra (aLOG #9961) for H1-SR2 showed noise at around 10Hz and up which does not appear in the LLO spectra for the same type suspension and phase of testing. Per the testing group’s request, the spectra data for H1-SR2 was reprocessed to show the individual channel data for each DOF and with additional comparisons to other LHO HSTS suspensions. The plots are posted below. All scripts and data files have been committed to the SVN repository.
Done.
Dave, Jim, Stefan Stefan restarted the h1odcmaster model which has been running since Friday evening and it did two things: locked up the h1oaf0 machine and caused DAQ errors on both end station front end computers, plus LSC, ASC and SUSAUXB123. Running start-streamers on the affected front-ends did not fix the problem. We took h1oaf0 out of the dolphin fabric, its console was locked up so we power cycled the CPU. All the models started OK and the MX stream error resolved itself. A few minutes later the IOP IRIGB drifted upwards to 66 and then back down to 15. I stopped the h1oafmaster model and then started it again. Once again it froze up the CPU but this time no MXSTREAM errors on other front ends. We took h1oaf0 out of the fabric and power cycled, once again h1odcmaster started (with 0x2000 DAQ error due to INI file mismatch). We are now testing on the DTS.
The login machine has been updated today, including an updated authentication module. This work required a reboot of the system.
The first power amplifier after the IFR function generator was saturation at an input of +13dBm. - I reduced IFR from +13dBm to +10dBm to avoid saturation. - I backed out all 3dB attenuators in the various feed lines. - I added a 3-way power combiner in the the 9MHz line to the EOM. This increased the drive to the 9MHz EOM by 10*log10(3)=4.77dB (compared to the original setting), or 1.77dB compared to yesterday's setting. We now should have a 9MHz modulation index Gamma of about 0.17.
A new version of gds tools has been installed for the DAQ test stand computers running Ubuntu 12.04 or Mac OS X. This will be effective with new logins or tools started from the command line of a new terminal window. The changes include a fix for diaggui to allow it to display reference traces with the proper x-axis values (Bugzilla 254), and a fix for foton to increase the number of significant digits used in design strings.
A new version of gds tools has been installed for the control room computers running Ubuntu 12.04 or Mac OS X. This will be effective with new logins or tools started from the command line of a new terminal window. The changes include a fix for diaggui to allow it to display reference traces with the proper x-axis values (Bugzilla 254), and a fix for foton to increase the number of significant digits used in design strings.
Translating TMS on test stand. Doug or Jason watching alignment.
Gerardo, Mitchell, Yesterday the ACB was successfully balanced. Today I will work on getting the tooling and hardware needed for install kitted up.
Yesterday we made some measurements for the noise model (end sation PDH spectra and open loop), and ALS COMM slow path, fast path and open loop).
Then we moved on to trying to make some noise measurements. We tuned the demod phase of RF A 9 to put the signal in the Q phase (we don't have enough range with the phase shifter to put it in the I phase). The demod is now on local control (remote controls not working) with al switches down except for 8, 2, 1, and 1/2 (11.5 ns delay).
We also tried to check the IR alingment into the arm cavity by locking at the transmitted PD. We found the 00 mode with the COMM PLL offset at 1.23, and saw up to 90 counts in transmission. We saw a misalingment mode (between pitch and yaw) at 6.48, with 17 counts in transmission. We strugged to improve this, mostly because the transmitted power was fluctuating so much from the frequency noise.
We engaged the boost on the PLL readback path, (adding more DC gain to ALS CARM). This reduced the noise significantly, although we were still moving by mre than a linewidth, so we can't really make a linear measurement when locked with ALS COMM.
Evan, Daniel
We found that the controls cable is missing from the second delay line, The cable we need is ISC-332 (DB25). We did not find any cable with this number, but we found two ISC-323. One is a DB25, the other a DB15. ISC-323 is for a shutter on ISCT6 (DB15). We think that ISC-332 (DB25) has been mislabeled. Unfortunately, it was pulled to the AS port rather than REFL.
Evan, Yuta, Stefan, Kiwamu,
As a preparation of ASC for PRMI, we briefly checked the WFS RF chains and then adjusted all the demodulation phases. We didn't attempt to close an ASC loop yet.
RF signal check:
When we checked the RF output from the WFSs, surprisingly we didn't find a reasonable signal at all. It eventually turned out that the REFL was not aligned on the WFSs due to some accidental servo on RM1 and RM2. Anyway, since we were already at the floor with an network analyzer, we went through all the segment by injecting an excitation at the RF test input and measuring the RF output for a sanity check. All the segment looked good and the measurement we got was consistent with what Rich did a long time ago (see alog 8022).
Also, we did the I-Q balance adjustment on a demodulator for REFL_B_RF9 by doing the same test as we did before (see alog 9100) because we haven't adjusted it due to a electronics malfunction. The resultant phase difference is:
H1:ASC-REFL_B_RF9_SEG4_PHASE_D = 88.27
The amplitude of I and Q were differ by 1% which is noticable, but we didn't correct it because in soem future we might change the whitening gain settings which messes up the amplitude balancing.
Adjustment of demod phases:
To have a pure in-phase signal, we excited the IMC length by injecting a 500 Hz sinusoidal signal into the IMC servo board. At the beginning, we had the PRMI locked and used it as a PM-to-AM convertor. However, for some reason, the PRMI doesn't stay locked for a long time at that time. Instead we simply used unintetionally-existing RFAM reflected off of PRM without any interferometer aligned. Of course, the RFAM was off from the true in-phase. Therefore we had to add a correction offset to all the demod phase at the end by using LSC_RF9 and LSC_RF45 since we knew that they were well adjusted for a in-phase signal.
The results are:
After the adjustment, we started shaking PR2 in pitch at 95 Hz. The signals mostly showed up in the Q signals everywhere. We are not certian what this indicates at this point.
Sheila, Kiwamu, Stefan
For the green team: The green beam needs to be realigned on ISCT1 because we had to move PR3.
We noticed that the maximum green arm transmitted light had dropped significantly (from 830 cts to about 500 cts) over the past day.
Also when we locked the PRMI on sidebands, we noticed that the POP18 buildup had dropped from 24000cts tp 4000cts. Looking at the in-vac POP QPDs we found that the actual build-up didn't drop significantly. Thus we suspected clipping, and indeed the green beam was on the edge of the first iris on the table. We moved PR3 to (P:-237.328, Y:-257.276) to center the beam on the iris. Indeed, when we locked up PRMI again we were back at 24000cts.
The old slider values were (P:-240.328, Y:-252.276), so we moved 3urad in pit, and 5urad in yaw.
We looked at how much of a jump the PR3 move was on the witness sensors and the optical levers old new delta Witness sensor P: -311.8 -307.2 +4.6 Optical lever P: 26.2 28.6 +2.4 Alignment slider P: -240.328 -237.328 +3.0 Witness sensor Y: -214.9 -220.0 -5.1 Optical lever Y: 23.4 19.4 -4.0 Alignment slider Y: -252.276 -257.276 -5.0 All are supposed to be urad... The witness sensors are the least reliable, they seem to drift on the order of 5urad over hours. But it's not clear to me that any of those signals are predictive of what we saw. Attached is also a 30day trend of all those signals.
Brett S Arnaud P
This afternoon we measured the resonnance frequencies of the ETMY in two locking configurations in order for Brett to improve the parameters of the quad model. We used PUM/UIM osem signals as well as the "temporary" optical lever to measure them. We plugged the pd electronics to the three first channels of h1pemey ADC0 in order to read the op lev signal from dtt.
For the first measurement, the top mass was locked and PUM, UIM, test mass were free. At first we didn't have a really clear signal from the PD in Pitch and Yaw, and since the PUM flags seemed to be really low and touching the osems Brett had to adjust the earthquake stops of the reaction chain in order to get the lower masses freely suspended. After the few adjustements, measurement showed all the needed frequencies in the osems as well as in the op lev.
For the second measurement, the top mass and UIM were locked. This time, the reflected beam was moving too much due to excess pitch/yaw motion, and after waiting a long time for the suspension to settle down, we changed the plan, and pointed the beam to the right edge of the test mass (bottom edge of the ear) to measure only the bounce modes. The dtt spectra of the op lev sum looked somehow funny. Fortunately the PUM osem signals were sufficient and showed all the frequencies, even the ones of the bounce modes.
Brett will post the results when they will be processed
At first glance the mode frequencies look pretty good. The pitch modes of the double pendulum hang (UIM locked) were a few percent off and the bounce modes of the same case were few percent high. The bounce modes were similarly high in the single hang case, so that is consistent at least. We left the ETMY with the main chain top mass locked and the reaction chain UIM and PUM locked, in case some measurements need to be repeated after I look at them in more detail. I plan to post the measurements to the log as I look at them in more detail.
This is the commercial software controlling the Lighthouse particle counters in the H1 PSL enclosure. Logging into h0dust (Windows 7 Pro SP1 64 bit virtual machine) through VNC this afternoon I found the error message in the attached pictures. I found two related events under Event Viewer (Local) -> Windows Logs -> Application on 2/7/2014 11:02:01 AM. The text of the errors is attached. Searching for related information I found the following: http://support.microsoft.com/kb/2640103
It looks like the LMS software install includes the .NET 4 runtime, but it was probably out of date at install. 4.5.1 is available through Windows Update, I would assume it includes the fix mentioned in the KB article (since it's over a year old now), but I haven't been able to confirm what's included in the update. It's probably the first thing to try though, but it means downtime for the updates to install.
I was on a mission of figuring out what we need to get done for running the OAF (Online Adaptive Filtering) model. I started the mission from simply copying Livingston's model. It seems that there are many things to be done mainly on some other models e.g. HEPI, ISI and PEM models.
Here are the items we have to do to get the model fully compiled:
All ISIs already are sending out their final stage's GS13s for general consumption. However, we should check what LLO does with these signals before just blindly copying -- if it's just to calibrate and project into the suspension point basis, then it will be redundant with what's already done in the suspension models. Since none of the SUS models are not running close to their cycle computation limit, the move would merely be for aesthetics...
HAM3-ISI tripped recently and needed to be turned back on. Commissioners noticed that the command buttons were not working anymore.
It turns out that someone ran an SVN update on the following file: /opt/rtcds/userapps/release/isi/common/scripts/HAMISItool, which was to be updated along with new models, screens and scripts, after testing at MIT. Having a model/script mismatch, the command script would be looking for inexistent channels, and fail.
I reverted HAMISItool to -r5830. This script should not be updated from the SVN again, until the latest HAM-ISI update gets propagated.
HAM3-ISI was turned back on with Isolation Lv3 and 01_28 blend filters. Target values were not changed to ensure returning to previous alignment.
The WD plotting software, which lives in the same folder, had stoped working too. It turns out that the related scripts were updated too, while they should not have been yet.
I reverted opt/rtcds/userapps/release/isi/common/scripts/wd_plots/ to -r6477, and the WD plotting software works again.
My bad, I updated the directory before modifying BSCISItool !
Jeff B. & Kate G. A horizontal 4" witness plate (wafer) and horizontal 1" optic (S/N 1247) were placed near the center of the SEI table on Feb 5th. A vertical 1" optic (S/N 1245) was mounted on the HR side of SR2. A vertical wafer still needs to be set near SR2. Important notes: Witness samples were not placed immediately after the chamber was opened. Jeff did some cleaning before SR2 was installed and the wafers and optics were placed. Also, the 1" optics were removed from their PET G containers, put in PEEK holders, double bagged, and stuck in a bag with other tools before going in chamber. I didn't see obvious scratches in the First Contact, so hopefully the samples weren't damaged.
On Feb 7th, I used a handheld particle counter in the cleanroom around HAM4 and in chamber before starting work. The counts seemed really high. The particle counter (used for trending data) on the +X side of the chamber was not working.
| Size (um) | Cleanroom Count | Chamber Count |
|---|---|---|
| 0.3 | 10725 | 17177 |
| 0.5 | 2518 | 3895 |
| 0.7 | 1075 | 1562 |
| 1.0 | 586 | 798 |
| 2.0 | 169 | 213 |
| 5.0 | 1 | 0 |
I placed a vertical wafer near the HR side of SR2, and inspected the cleanliness of the chamber with a flashlight array. There were a few particles on the witness samples (which had not seen work since they were placed two days earlier), and a lot of contamination on the TCS HWS optics and SEI table. The dark iPhone photos don't do it justice.There was only time for a quick cleaning using the TigerVac and IPA soaked wipes. It would be helpful to get a PET swipe sample the next time work is done in HAM4.
