corey.gray@LIGO.ORG - posted 08:44, Tuesday 09 December 2014 (15505)
Diode Room Crystal Chiller Re-filled
Topped off the Chiller this morning (~190mL).
Topped off the Chiller this morning (~190mL).
no restarts reported. Conlog frequently changing channels list attached. Sheila reports the channel which is changing continuously ( H1:ALS-Y_REFL_SERVO_IN1EN) is a Beckhoff PLC controlled channel, attempting to lock the Y-arm.
Interestingly the channel H1:FEC-92_MSG shows greater than 600 changes in one hour. This is the h1isietmx model. There were three load-all-filters-coefficients performed in one hour (between 13:00 and 14:00 PST yesterday) and the large number of filters in the model happens to give a total number of filters loaded > 600, so no problems there.
The contractors for the new DCS addition will be at it again today, all day outside with a larger piece of equipment. Some ground shaking. The inside portion of the construction will include hanging the indoor cooling units and taping of the sheet rock for the new wall.
We are in the process of preparing a cleaning of ETMY. Even so, it is not conclusive, ETMY has a clear problem as indicated by the green ring type absorption pattern which is thought to be due to first contact material left behind. At least 2 of the bright spots are extended and do not look like single point defects. A vent in EY is planned to start Monday next week. The opening of HAM1 has been delayed to next week as well to allow us to keep working with the Y-arm this week. Since we do not want to "bake" contaminants to the TMs, we decided to postpone any locking attempts and instead concentrate on a couple of task which previously had second priority.
Here is the list of commissioning task for the next 7-14 days:
Y-arm loss investigation:
Seismic:
Length team:
Alignment team:
Miscellaneous:
RF:
J. Kissel, H. Radkins, K. Venkateswara, J. Warner
To expand on Daniel's Seismic point, we *know* why LHO's seismic performance is lagging behind LLO. Here're our action items to fix it:
(1) Get sensor correction running on all DOFs of all chambers.
(a) Assess functionality of STS-2s in the corner station, now that they've been iglooed.
(b) Figure out why X&Y HAM2 / HAM3 sensor correction is only intermittently successful, get it implemented on all other HAMs
(c) Use Rich's tilt-free IIR filter for Z on all HAMs instead of Hua filter.
(d) Import LLO's X&Y BSC-ISI "0.43 [Hz]- only" sensor correction filter, see if it works for our seismic environment
(e) Make sure CPS / IPS, low-frequency tilt-decoupling matrix elements (for HEPI and ISIs) are doing us good, if not fix them
(2) Figure out what's limiting each platform's motion at all frequencies using a NoiseBudget
(a) Re-update Rich's BSC-ISI noisebudget model to be better organized, and to include getting live data and live filters like the full IFO's noise budget
(b) Make it produce plots that tell us what's limiting the BSC platform motion after sensor correction is turned on. We know that residual RY motion is limiting the performance, but we're hoping that the noise budget will help us reveal *why*.
(c) Create / Update a HAM-ISI noise budget model in the same vein
(d) Make it produce plots that tell us what's limiting HAM-ISI noise performance.
Dan and I have left an MC measurement going overnight. It should only last about 6 hrs, so should be complete by the morning. In case I am not in early enough to turn everything off and it is interfering with people, do the following:
1. Turn off the input of MC common mode board Exc A (I will clear the awggui's that are running; as long as this excitation input is off they won't be doing anything)
2. Set the three LSC lockin oscillators Amp to 0
3. caput H1:ISC-EXTRA_C_REFL_AO_1 = 0
4. Disable MOD on IFR; this step is not really necessary and I can do it once I am in.
I have turned off the measurement. Everything is restored.
Evan, Krishna, Alexa, Dan, Kiwamu,
At some point in this evening, we noticed that the IMC kept dropping its lock for some unknown reason. It turned out that the HAM2 ISI sensor correction was amplifying seismic at 0.3-ish Hz which resulted in a large drive in MC2 to keep it locked. So we turned the sensor correction off for now. This fixed the issue.
Even though the MC2 coil DACs were not saturating, somehow the motion was big enough to unlock the IMC. It is still unclear why it could unlock. Anyway, after turning off the sensor correction, the amont of drive in MC2 suspension reduced by a factor of 5 or so and IMC became able to stay locked. With the sensor correction on, the MC length was moving approximately +/- 5 um according to IMC_X displacement monitors. We don't know what changed in the HAM2 ISI sensor correction as it has been running fine for a while recently. Only thing we immidiately noticed was that the seismic freq-limited RMSs were pretty high in 0.1-0.3 Hz and 0.03-0.1 Hz bands in this evening. The attached is a time series of the IMC_X signals when we did a simple test of turning on and off the sensor correction in the HAM2 ISI. Since we were a bit rough for turning on and off the correction, it gave a transient when switcing it, but one can still see that when the correction was running, the IMC_X signal increased by roughly a factor of 5.
According to Valera, we should have sensor correction on for both HAM2 and HAM3, or for neither.
If both of them have sensor correction on, the MC benefits from the reduced seismic noise. If both of them have sensor correction off, the MC benefits from some amount of common-mode rejection in the ground motion between HAM2 and HAM3. If only one or the other has sensor correction on, then the MC is fully exposed to the ground motion, since one platform moves with the ground and the other is isolated.
For PRC/SRC the sensor correction has to be ON on all HAMs (as BSCs are already inertial due to low blend) to avoid the full ground motion impression on these cavities.
S. Dwyer, J. Warner, H. Radkins, K. Venkateswara
We did a quick test of sensor correction (SC) along X direction to Stage 1 ISI, previously described in 15146. This attempts to use the ground seismometers located near the test mass chambers in feed-forward like manner to reduce the cavity motion. Sheila and Jim aligned and locked the X-arm using the green laser and data was recorded with SC OFF and ON.
The first pdf shows the signals with SC OFF and the second shows them with SC ON. The signals shown first are the Stage 1 T240s (inertial sensors), the CPSs (positions sensors) and the X arm control signal. The inertial sensors indicate a significant reduction in motion of Stage 1 on both ETMX and ITMX between 50-500 mHz. The microseismic motion is suppressed by factor of ~4. Yet, surprisingly, the X-arm control signal looks almost similar. The oplevs shown on pages 3, 4 indicate that the angular motion of the optics was unchanged so perhaps it might be limiting the control signal.
I 've added a few more lines in to the above plots. In particular, it shows that Ry motion of Stage 1 on both ETMX and ITMX, has large correlation with Pitch of optics. Improvement in angular control of Stage 1 may be needed before longitudinal sensor correction can improve the cavity control signal.
K. Venkateswara
I revived the BRS system after a crash towards the end of last week. It had successfully worked for a period of 16-17 days since the last restart.
As described in SEI alog 638, I've modified the pendulum response-inversion filter. I've also modified the tilt-subtraction scheme as described in the subsequent comments to the log. A high-pass filter at 5 mHz has been added in the sensor correction path (in the calibration filter bank) to roll off the low frequency increase in noise.
I've attached a pdf showing the output of the new tilt-subtraction scheme. The red 'super-sensor' curve looks similar to the blue ground curve, so we'll have to wait for a windy period to see if the tilt-subtraction has been improved or not.
Dave, Evan
Here is a larger scan of the the Y arm loss as a function of ETMY spot position. I tried to get out to a 6 cm radius from the nominal, but eventually the alignment was so bad that the arm broke lock. Also, I've masked out points for which the TRY dc value is less than 10 ct.
Also attached is Friday's measurement, with the colors rescaled to match the colors of today's measurement. Note that Friday's measurement is a smaller spiral. It's hard to know how to compare these measurements, but I've attached another plot where I suggest where Friday's small spiral may fit into today's big spiral.
Note that the zero points of the displacements are not consistent between different measurements; each time I have simply aligned the arm to get good buildup and then started the spiral. Attempting to reproduce the pointing from day to day would be a much more involved process.
Ran DBB scans of the PSL. The plots are attached below.
Day's Activities:
Laser Status: SysStat is good Output power is 33.1 W (should be around 30 W) FRONTEND WATCHdog is Active HPO WATCH is RED PMC: It has been locked 7 day, 21 hr 39 minutes (should be days/weeks) Reflected power is 2.4 Watts, and PowerSum = 25.8 Watts. (Reflected Power should be <= 10% of PowerSum) FSS: It has been locked for 0 days, 0 h and 17 min (should be days/weeks) Threshold on transmitted photo-detector PD = 1.555V (should be 0.9V) ISS: The diffracted power is around 5.700% (should be 5-15%) Last saturation event was 0 days, 0 h and 15 minutes ago (should be days/weeks)
The HEPI Pumps stopped due to a level trip in the reservoir fluid height. I did a 100% scan of the system in the LVEA and found no evidence of a leak. I should have done similar in the MR but went back to my previous assumption that the fluid level declines were from the known motor drips on PS8 (1.) I had lowered the trip position to its lowest level last Tuesday so I proceeded to add fluid (Best to do while pumps are off.) Added ~2.4 gallons and then got the pump stations back running.
It seems someone else had seen a leak on the mezzinine previously--Just before leaving I noticed something (can't remember exactly what) behind the blue drum behind the reservoir, on the floor was an absorbent pad and other puddles. I first thought TCS but no it was HEPI fluid. I found a 5 second drip coming from the pipe/hose joint at the electrical break. I tightened this clamp a bit and will regularly monitor for a few days. I suspect the leak rate may have increased over time but if it had started at full rate, it would take 105 hours to leak a gallon. Since last Tuesday, I'd say the level had to drop about 3/8" or 0.3 gallons. Given the mess I cleaned up this morning, that seems about right.
So, I don't know when this leak started, it may have been much slower for a long time. Was the absorbent pad just luckily in the right spot or had someone put it there? Did someone tell me about it (a long time ago,) possibly?
How do you know that ITMY is not the problem? What came of the scans of that mirror?
model restarts logged for Sun 07/Dec/2014
2014_12_07 15:28 h1fw1
2014_12_07 22:42 h1fw0
Both restarts unexpected. Conlog frequently changing channel report attached. FYI: H1:ALS-Y_REFL_SERVO_IN1EN is being toggled every 2 seconds by a non-guardian script.
SudarshanK, TravisS, EvanH, AlexaS, RickS Using the Pcal beam localization cameras at both end stations, we took images of the ETM surfaces under three conditions: IR and Green resonating; IR only resonating, and Green only resonating. Attached below are two composite images composed of four separate images taken with the same camera settings: Upper Left: Xarm Green Lower Left: Xarm IR Upper Right: Yarm Green Lower Right: Yarm IR The images in the first composite were taken with the following camera settings: F8, ISO 200, 30 second exposure, WB-cloudy. For the second composite image the aperture was F29 (~13 times less light) The Yend camera was re-focused for the IR-only images, but the Xend camera was not re-focused.
Thomas Abbott at LLO applied the Pcal beam localization analysis the the LHO ETMY image to calculate the position of the center of the optic in the image from last Friday. The image below contains lines that indicate the center of the optic using the Pcal image analysis.
Attached is a picture of the original FirstContact (FC) sheet, circa ~Jan 2014, showing the "IAS window" which is a thinner film of FC in the central 3" of the larger sheet. To me, the shape of the FC window looks similar to the 3" ring showing up in green on the recent optic photo above. SYS is working with us to get our cleaning game plan together in order to remove the ring. As well, they are investigating other possible scenarios of where the ring came from if not the window. Note, there was a full FC sheet re-cleaning in March that apparently did not remove all of the ring that was left behind apon the removal of the first sheet. To be continued...
I've attached an overlay of (a) the SolidWorks CAD view of ETMy along the PCal camera path and (b) the PCal camera image of H1 ETMy (scaled and rotated). Since SolidWorks does not diffract the image viewed through the ETM optic, I indicate the shift in the ETM Telescope Baffle aperture as well. Three of the 4 bright areas are along the ETM Telescope Input aperture/baffle edge (a coincidence?). (The upper one is red.) One of the 4 bright spots does not correspond to any feature in the CAD image and is likely a spot of residual First Contact. As subsequently shown by the zoomed in PCal image using the Green Lantern flashlight (green LED) after venting (see entry #15635), it is simply a coincidence that the two prominent bright areas appeared to be along the ETM Telescope baffle aperture edge.
Dave, Alexa, Evan
We’ve now completed a scan of the equivalent ETMY loss as a function of spot position on the Y optics.
cdsutils.avg.Loss vs. alignment is given in the attached plot. The attached zip contains the data and the code used to perform the measurement. The measurement uncertainty is about 45 ppm, and comes from the uncertainty in the number of ASAIR counts with the cavity unlocked. Note that this plot is equivalent ETMY loss; i.e., all the observed loss (including power in the rf sidebands and mode-mismatched light) is assigned to the ETM.
We are repeating the measurement for the ITM. Then the script will try to get a bigger spiral on the ETM.
I’ve attached the results of Friday night’s ITM scan data (first attachment). The arm unlocked shortly into the ETM scan, so there is no data there. Note that the axes indicate the ITMY spot position, but the quantity plotted is the equivalent ETMY loss (for the sake of consistency with the previous measurement).
First, the data indicate that the spiral is centered around an especially lossy spot for the arm. Second, it seems that sweeping the spot on the ITM can change the measured arm loss by several hundred ppm. That magnitude seems comparable to the effect of sweeping the spot on the ETM, as we measured on Friday. There are several possible things we might conclude from this:
The ITMY coating also has anisotropic loss.
Moving the spot on ITMY is causing clipping somewhere (e.g., the edge of the optic).
The scan strategy is also moving the spot on ETMY. Since the ROCs of the ETM and ITM are similar (2.2 km and 1.9 km, respectively), we require the alignment sliders to be calibrated to better than 7 % of each other in order for this to work [since (4 km - ROC_I) / ROC_I = 1.07].
The unlocked value of ASAIR is changing with alignment.
To test the last of these, I locked the Y arm, turned on the WFS loops, and looked at the locked vs. unlocked values of ASAIR for several different arm alignments (separated from each other by 2 urad in pitch/yaw). I first measured ASAIR while locked. Then I held the outputs of the WFS loops, unlocked the arm, and measured the ASAIR value again. I found consistently that the unlocked value was 1370(15) ct. So it seems this is not the issue.
To test the idea that the spot location on ITMY is causing clipping, I adjusted the arm alignment to give lower loss. I ran the ITM sweep again (second attachment). Here the overall loss values are much lower (as low as 550 ppm), but again we find a variation of 100 ppm or so.
The limiting aperture for an ITM scan is probably the BS baffle. Not sure we can (yet) conclude that these losses come from the ITMY.
Here are the expected clipping losses as a function of offset from the mirror center. This assumes a 62mm radius beam on a 326mm radius optic coating. The pdf shows the clipping loss as a function of offset, and the other attachment is an animation showing the intensity spilled over the edge as the offset is increased (since gifs are all the rage these days).
