The connection was lost after Cyrus's network work today (work permit 4127).
ETMx Silica-tipped Earthquake Stops Positioned on Quad (Andres, Corey)
The face EQ stops for the ETMx needed to be positioned, but we don't want "glass on glass" contact. We didn't have thin teflon strips to insert between the EQ stops and the glass, so per Betsy's suggestion, we used Alpha Wipes as a buffer between the Stops & Glass.
SUS TMS Cable Fixed
Installed screws on an in-vac cable which had these screws missing from its connector. All TMS cables are good to go for dressing on the ISI.
ITMX transfer functions ran overnight are showing a better agreement with the model after the recent modifications (see alog 7755). Results are described below
(1) In black ITMX last TF, after aligning two top mass osems and adjusting lower mass face EQ stops , compared with the measurement took 2 days ago, after having retracted top EQ stop touching the top of the test mass (in orange), compared with the model
(2) In red, ITMX last TF compared with wire rehang model
Data Scripts and PDF files have been commited to the svn
In preparation of RGA scanning later today
Just checked the ITMx alignment after the latest round of rubbing fixes. Bottom line is the ITMx is still within spec and aligned. Position alignment still remains unchanged from log 7638. Pitch/yaw numbers are as follows:
J. Bartlett, M. Barton, J. Kissel, A. Pele Catching up on a lot of measurements with overdue aLOGs. Executive summary -- we think the QUAD is free, but we're taking overnight TFs to confirm. The main chain is the problem, reaction chain looks great. ----- After initial "release" of H1 SUS ITMX as a wire(re)hang glass test mass, Arnaud had taken a set of transfer functions (2013−09−09) that revealed rubbing. Bartlett then investigated, and found some barrel stops on the main chain (M0) test mass were in contact. Arnaud then took another set of overnight transfer functions (2013−09−11). However, those transfer functions revealed poor results as well, still indicating rubbing somewhere on the main chain. Bartlett went in to investigate, and immediate found "face" EQ stops rubbing the front of both the PUM and TST. Once these were alleviated, in order to get a quick turn around Kissel took as set of DTT transfer functions. Though the results showed significant improvement, poor coherence still remain in the 0.5 [Hz] region for most DOFs, and a vertical mode (at 0.51 [Hz]) stuck up in the ROLL degree of freedom. Both raised suspicion that we might not yet be out of the woods. The problem regarding the V feature in the R TF is in this region where the TFs had very little coherence. So Kissel cooked up a (very plausible) story (IHHO), about a locked ISI and lots of air currents leading to lots of vertical motion (especially with the QUAD undamped) which, for a poorly aligned sensors, would result incoherent, common mode (vertical) noise in the differential (roll) transfer function. We've seen this before on many suspensions (and suspension types), and seen it get better and/or disappear both with an unlocked ISI, and when in-chamber. After some further discussion, Bartlett went back in and re-centered the M0 RT and M0 F1 OSEMs about their flags (which he said were visible mis-centered, if not even rubbing themselves, see LHO aLOG 7752). The feature was certainly reduced after this, but still present (and with even less coherence). To further support the theory of too much ambient motion, we (a) took a (DTT) swept sine TF in this frequency band, which also revealed low coherence (not attached), and (b) took a (DTT) white noise transfer function with the V damping loop ON (ONLY). (a) revealed that even with all the power focused in this frequency region (as Arnaud says it "the best possible drive"), we still were not able to obtain good coherence. (b) Further reduced the feature becomes ill-defined, to a point that it begins to just look like "normal" incoherent noise. Note that we *only* looked at ROLL after Bartlett re-centered the OSEMs. The first attachment shows the white noise, M0, ROLL (only) investigations, the second attachment documents all of the complete set of M0 TFs we have from 2013-09-09, 2013-09-11, and today (2013-09-12). The third attachment shows that the reaction chain is great. Because it's "free," we're taking over night Matlab transfer functions, which hopefully will have good SNR on Roll, and it will also allow us to check the rest of the DOFs, to make sure those are still fine.
A new round of overnight TF on ITMX has been started at ~6pm.
The connection was lost after Cyrus's network work (alog 7753).
The old iLIGO Foundry ethernet switches were deinstalled from the outbuildings (MX,MY,EX,EY) today (see WP4127). The remaining FMCS and serial server connections were moved to the replacement switches in the process. Work will continue tomorrow in the AM to remove the last remaining switch from the LVEA, and the supporting switch in the MSR. Then the last vestiges of the iLIGO network will be expunged. (Yay!)
The transfer functions showed the rubbing issue from this morning has been resolved, however they also revealed cross coupling. Found the F1 and Rt top BOSEM flags close too if not touching the side of the BOSEM. I have adjusted the centering alignment of these BOSEMs (and checked the other four). Arnaud will rerun the transfer functions tonight.
Put ETM-X in stops so Seismic can adjust weight on ISI.
Overnight TFs showed rubbing on ITM-X. Found all four Rt side face stops on the main chain penultimate and test masses touching. Retracted the stops and checked the rest of the stops on the suspension. Arnaud is rerunning the transfer functions.
- "Roofers": material on the OSB roof
- Arnaud: measurements of ETMX structures resonances
- Richard: from ~10:00 to ~10:30 working on GPS antennas
- Bubba: Mid Y crane
- Jeff B.: End X work (ask him for more deatils)
- Cyrus: End Y, Mid Y works (check work permit 4127 or ask him)
Firefox crashed again, so, I just open a terminal and:
$ rm -r /ligo/home/ops/.mozilla/firefox
a bookmark can be found in Documents/pablo/
Have a great evening!
pablo
J. Kissel [[[ EDIT -- The feature excited in the attached measurements is at 5.1 [Hz], NOT 6.1 [Hz]. I was reading the plot too quickly, had 6.1 [Hz] on the brain, and assumed it was 6.1 that I was ringing up. This 5.1 [Hz] feature is most certainly the last "transverse" mode (that has roughly equal components in L, R, and P as well), modeled to be at 5.08 [Hz]. Only configuration (3) shows anything at 6.18 [Hz]. In the other configurations, it is probably just that I'm only using one of two vertical sensors that excites roll motion -- which is NOT damped, because the R loop is open during these measurements. Note, that this doesn't rule out that M0 LF has a problem. But it certainly makes concluding anything more complicated. Sorry for the confusion. Thanks to Dr. Lantz and J. McIver for reminding me how to read a logarithmic plot.]]] Inspired by a suggestion of the good Dr. Lantz, I tried to further narrow down the source of the 6.18 [Hz] feature that has been plaguing H1 SUS ITMY. With the ISI still in it's best performing state, L,T,P, and Y SUS damping loops running, and the R loop off, I tried running the M0 (main chain) V SUS damping loops (which use the average of M0 LF and M0 RT OSEMs) with only M0 LF and only M0 RT OSEMs. I did this testing 5 different configurations: (1) V loop open (2) V loop closed with V to M0 LF OSEM2EUL element as zero (V to RT element at nominal -0.5, EUL2OSEM elements using both as nominal) (3) V loop closed with V to M0 RT OSEM2EUL element as zero (V to LF element at nominal -0.5, EUL2OSEM elements using both as nominal) (4) V loop closed with both V to M0 LF OSEM2EUL and EUL2OSEM elements as zero (V to RT elements at nominal -0.5) (5) V loop closed with both V to M0 RT OSEM2EUL and EUL2OSEM elements as zero (V to LF elements at nominal -0.5) Configurations (1), (2), and (3) are shown in the first attachment (*_0435_*.pdf), and configurations (1), (4), and (5) are shown in the second attachment (*_0521_*.pdf). Configurations (1) and (4) did NOT show the 6.18 [Hz] feature, while configurations (2), (3), and (5) all showed the 6.18 [Hz] feature (plus various other mechanical modes and harmonics rung up as well). This directly points the finger at the M0 LF sensor/actuator. Indeed, because the ISI input motion is so small, without the M0 LF OSEM in use, the *ENTIRE* spectra is limited by the expected OSEM sensor noise floor (see, e.g. configuration (1)). It's therefore blatantly obvious when the feature oscillates that it excites the whole suspension in many DOFs. If one squints, one *CAN* see the feature in all configurations, implying that wherever the source, it is visible in BOTH LF and RT sensors. In the configurations where the feature is not triggered into oscillation, it's only a factor a few above the noise floor, but it's still there. I'm not sure that this data set rules out the sensor, actuator, a nasty cross-coupling, or something else inherent to the motion of the suspension, but at least it narrows down the oscillation problem to one rogue OSEM. Please swap out the M0 LFRT, R0 LFRT satellite amplifier tomorrow, and we'll test whether the feature disappears.
In order to continue to get good-performance measurement time with the ISI, and so there's not some terrible version control snafu, I've restored the OSEM2EUL and EUL2OSEM matrices to their nominal state of using both LF and RT OSEMs at -0.5, and left the V and R loops OFF.
I removed the temperature/relative humidity probes from the dust monitors in the OSB optics lab, vacuum prep lab and bake oven room. I also power cycled each dust monitor after I removed the probe. Ever since these were added there were constant intermittent communication errors with these dust monitors. This closes work permit 4130.
After figuring out that the 6.18Hz was coming from the vertical loops of M0 (The left OSEM doesn't behave properly), spectra and transmissibility transfer functions were measured with the following configuration:
- HEPI with position control (100mHz UGF)
- Feedforward HEPI L4C to stage 1
- Stage 1 blend
o 750mHz in Rx, Ry, Rz
o 100mHz with notch at 440mHz in X
o 40mHz with notch at 440mHz in X
- Stage 2 blend
o 750mHz in Rx, Ry, Rz|
o 100mHz with notch at 440mHz in X and Y
- Level 3 controllers
o Stage 1 – 40Hz
o Stage 2 32-35Hz
- SUS damped except on M0 in Roll and vertical
- No sensor correction
Spectra presented were measured at 1062915770 (start time) during 45 minutes. Spectra in all DOFs are presented in attachment. Performance is great.
Transmissibility transfer functions are measured while driving white noise on the HEPI actuators. Two transmissibility plots are presented. In the first one, the transmissibility is computed using the HEPI L4C as input.
In the second one, the transmissibility is computed using the HEPI excitation and the supposed input motion is calculated (alog https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=7692)
It is important to note that:
- The isolation filters of the BSC-ISI were designed with the HEPI locked
- These loops are stable and robust in time. There were closed few days.
One of requirements of the ALIGO optical levers are to perform as well or better than their ILIGO predecessors in terms of sensitivity and their respective resonances. I'm not sure of the state of the ILIGO optics during the measurement, but for ALIGO, we had the suspension damping loops running, the ISI damped, and HEPI locked. Attached are 4 graphs comparing the ITM and ETM spectra from the LHO HIFO-Y test to Kate Dooley's spectra from ILIGO. Both sets of data were calibrated to uradians so they should be directly comparable: - First off, we see that the sensitivity at frequencies >2Hz, the ALIGO sensitivity is better or the same in pitch for both ITM and ETM but yaw is not as good; the noise floor for ITMY pit is due to the ADC noise floor, as is true of the rest of the OL signals, but the other DOFS don't quite get down to that level of sensitivity till ~20Hz, this should be improved. It looks like the noise is dominated by electronic that is ~1/f dependence, most investigation is needed. - The good news is that the first resonances of the ILIGO piers occur at ~21Hz whereas the first ALIGO resonances that come from the pier peak at ~31 Hz, and keep in mind that this is done without grout which shown in ALOG 7746, will help push the resonant peaks higher in frequency.
Robert Schofield, Thomas Vo While looking at ITMY spectra from HIFO-Y of the optical levers, we saw that the first pier resonance occurred at ~31Hz, which was caused by the transmitter. The real question was whether or not this peak is due to rigid body motion or if it's from flexing/vibrations in the pier. This is important because the piers are not yet grouted and if the resonances are from rigid body motion, then grouting will help but it wouldn't help the vibrations and flexing of the pier as much. So we set up a little experiment to see if grouting will help this particularly troublesome 31Hz peak. First we used an accelerometer and took spectra as is, with only three bolts holding up the pylon, then we added 4 machine jacks to help support the pylon, much like grout would do. Attached are pictures of the two setups, and spectra showing that the 31Hz peak shifted up 50 Hz after adding the machine jacks, yay. Of course, we'll repeat a similar measurement when grout is in place.
K. Izumi, M. Barton [J. Kissel as moral support and scribe]
Bolstered by my bold claim of stable H1 IMC locks last night, Mark began launching MC2 violin mode measurements. However, the lock stretches were not long enough to get a full ring-down. My unsubstantiated claim yesterday was air-currents in the PSL were the source of lock losses (see LHO aLOG 7736), but Kiwamu identified it to be low frequency excursions that saturated the MC2's M3 stage drive -- by normal control that should be adequately offloaded to the upper stages of MC2.
Why? Two days ago, during Keita's attempt at debugging the IMC (see LHO aLOG 7724), he had modified Stefan's auto-locker (MClockwatch) to use the smaller gains he'd found to be stable. We had forgotten this yesterday, among all the other things wrong. Further, after triple-checking against the known filter configuration (from LLO aLOG 7974), we found FM4 (a [z:p] = [1:100] filter) in the M2 LOCK L bank had been disabled some time during this recent week of guardian commissioning / problem solving (who knows why or by whom, doesn't matter). We've restored the correct gains into MClockwatch, and engaged M2 LOCK L FM4, and the H1 IMC has been locked ever since.
NOW the H1 IMC is back to full awesomeness. MClockwatch is running on opsws5. PSL is still in commissioning mode.
As a reminder, the length loop settings in this configuration are
During Lock Acquisition
H1LSC H1:IMC-L = FM1 (antiWhite) * G = +1.0
H1LSC H1:LSC-MC = FM10 ( :0.3) * G = +1.0
H1SUSMC2 H1:SUS-MC2_M3_ISCINF_L = (no filters, G = +1.0)
H1:SUS-MC2_M3_LOCK_L = FM3 (150:40) * FM9 (CLP300) * G = -300
H1:SUS-MC2_M2_LOCK_L = FM4 (100:1) * FM10 (ELP70) * G = 0
H1:SUS-MC2_M1_LOCK_L = FM2 (0.1.0.01) * G = 0
While Locked:
H1LSC H1:IMC-L = FM1 (antiWhite) * G = +1.0
H1LSC H1:LSC-MC = (no filters, G = +1.0)
H1SUSMC2 H1:SUS-MC2_M3_ISCINF_L = (no filters, G = +1.0)
H1:SUS-MC2_M3_LOCK_L = FM3 (150:40) * FM9 (CLP300) * G = -300
H1:SUS-MC2_M2_LOCK_L = FM3 (0.1:1) * FM4 (100:1) * FM10 (ELP70) * G = +0.2
H1:SUS-MC2_M1_LOCK_L = FM1 (0:0.01) * FM2 (0.1.0.01) * G = +1.0
In the optimal laser configuration (current = 1.503A, see alog7695), the power before ALS-FI2 was measured to be 39.7mW, while the power after the FI was measured to be 37.6mW. Thus, the throughput of the ALS-F12 was about 94.7%. The first jpg (532nmALSfI2FullCurrent.jpg) shows the beam profile after the ALS-FI2.
The first pdf (BeamProfile532nmRefALSFI2.pdf) is a graph displaying the mode measurement of the 532nm beam with ALS-F12 as a reference. Meanwhile, the second pdf (BeamProfile1064nmRefALSM1.pdf) is a graph displaying the mode measurement of the 1064nm beam with ALS-M1 as a reference. Using the 13.5% width (or 1/e^2 diameter) of both the horizontal and vertical profiles, the radii for the two profiles were determined at various distances from the respective reference points. In addition, the graph includes error bars of one standard deviation. A line of best fit was produced, and the difference between the line of best fit and the data points was plotted in the lower portion of both figures.
Sorry, I meant ALS-FI3 in the table layout.
Also, as of today, we have moved the laser 1inch closer to the beam path (relative to when the beam profiles mentioned in the alog were taken).
Note, the above states that the face stops were adjusted to clamp the ETMx - from afar, I have clarified with Andres that indeed all of the appropriate barrel stops are actua;;y doing the clamping as I had prescribed for SEI balancing work.