Robert, Christina, Dave
Robert and Christina reported some corner station PEM channels with ADC values of zero. I power cycled the h1oaf0 front end computer and its IO Chassis and these channels are now non-zero.
Sequence was:
Investigation reveals that all (3) running turbos on site (YBM, XBM and X-end) shut down simultaneously at 1603 hrs. local time yesterday -> Coincidentally, this was only a few minutes prior to my opening of GV2 -> As such, my earlier theory of the XBM turbo tripping off on its safety valve pressure setpoint turned out not to be the case -> This is confirmed also by the fact that PT170A never came on scale. Today at ~1230 hrs. local -> I spun-up to 100% rpm the troublesome XBM turbo by employing the technique of "loading" the rotor. To do this, I maintained the turbo inlet pressure at ~ 0.2 torr by adjusting the "up-to-air" needle valve at the turbo's inlet while it spun-up -> Once at full speed, I shut off the air and eventually valved-in the turbo to the XBM volume. ERRORS in PRESSURE GAUGES Also, the LHO vacuum equipment is beginning to show its age as we have been experiencing an increase in the failure rate of the site cold-cathode gauges (lifetime maturation) -> most noticeably PT180B, PT120B and PT170B are reading bogus values now for portions of their nominal range. As of this writing, both the YBM and XBM turbo inlet CC gauges are reading 1.8 x 10-6 torr which is what I would expect for the recent history -> I would then guess the pressure at BSC2 to be 5-7 x 10-6 torr
There were 167 restarts due to storm related power glitching. Also some unexpected restarts of h1fw1. I have attached the list to spare the casual reader the gory details.
It was the best of lasing. It was the worst of lasing. It was the best of shipments (although still bad). It was the worst of shipments. It was the season of single mode. It was the season of no modes. TCS has now received back the lasers that were sent to Access for repairs. Unfortunately the first shipment containing Lasy-50 20510-208160 and RF50 208160-20510 was sent to Caltech (an oversight in shipping) and then to Hanford. It was worse for the wear and the RF driver was busted, it was blowing fuses. Using the RF50 210040 to drive the laser proved more fruitful, while RF50 208160-20510 will be sent back to Access for repairs. Here are the laser/RF driver pairs and their outputs (running at 99% duty cycle) as tested:
Laser | RF Driver | Watts | Notes |
---|---|---|---|
20510-208160 | 210040 | ~53 | RF50 208160-20510 Broken |
20708-208100 | 20708-208100 | ~53 | RF50 water inlets bent |
210020-20710 | 20710-210020 | ~52 | RF50 broken water inlet, replaced. Break in wire sheath, fixed. |
Attached are some of the photos from the process, including the beam on an IR card.
Replaced insulating film used to isolate the -15V regulators to the chassis metal wall for OP Lev Whitening chassis at end stations. EY - S1101539 EX - S1101552 Filiberto Clara
The daqd process was restarted on h1nds1 to test the copied raw minute files. WP 4790.
The x1work computer won't boot to Ubuntu 12.04. It is up temporarily in Ubuntu 10.04 to allow access to user's home directories. Don't do anything serious on it, it will be rebuilt after the power fail scheduled for tomorrow. Home directories and /ligo will be preserved since they are on separate disks.
HAM6 alignment work ongoing IMC ring down measurements PR3 oplev almost done ISS arrays being finished up Power cabling for PEM chassis Pulling cables for UW tilt meter EX tilt meter testing ongoing EY charging measurements ongoing Safety Meeting this aft
(Borja)
First of all let us compile the chronological time information of the measurements and the discharge runs:
* Measurement 1 (before any discharge): took place between UTC (2014-08-07 07:00:00) and UTC (2014-08-07 08:25:00).
* Discharge run 1: between UTC (2014-08-07 19:15:00) and UTC (2014-08-07 19:44:00).
* Measurement 2 (after 1st discharge run): between UTC (2014-08-8 18:00:00) and UTC (2014-08-8 19:30:00)
* Discharge run 2: between UTC (2014-08-08 20:48:00) and UTC (2014-08-08 21:38:00).
* Measurement 3 (after 2nd discharge run): between UTC (2014-08-10 00:00:00) and UTC (2014-08-11 02:00:00)
* Ion pumps in ETMY station was turned ON at UTC 2014-08-11 16:44:00. Notice from this entry that the ion pumps were OFF between discharge 1 and discharge 2 maybe that is why we see a charge reduction after each of these discharges. However the first charge measurement done above (before any discharge run) was done after several days of having the ion pumps on so why the ETMY had not bigger charge?
* Measurement 4 (2nd measurement after 2nd discharge run): between UTC (2014-08-11 20:31:17) and UTC (2014-08-11 23:00:00)
* Measurement 5 (3rd measurement after 2nd discharge run, this measurement is reported in this entry below): between UTC (2014-08-12 17:17:00) and UTC (2014-08-12 19:09:00)
Yesterday ETMY charge results (Measurement 3) were telling us that the ETMY was charging again since the 2nd discharge took place. What is charging the ETMY then? There are 6 mechanisms (that we can think of at the moment) that would cause ETMY charging and that they are different to the original MIT charging measurements, these are:
1) Green light (through 2 photon process). MIT measurements did not include green light tests. But the green light is not on at the moment.
2) The rubber stops on the reaction mass (seismic stoppers) with silica tips through silica to silica friction. The MIT tests did include these rubber stops and they did not observe any noticeable effect on the charging. Certainly they are installed in the ETMY reaction mass so they could be playing a role in the ETMY charging but it is hard to believe they would charge the mass so much in only 1 day.
3) LED illuminators that were turned on from the first discharge run to see effect on the ETMY surface. They have not been turned off since (although there is no reason to keep them on at the moment apart of testing the current charging hypothesis). Again it is hard to believe that a broad angle not extremely bright LED can charge so much a mass in only 1 day.
4) First contact. For the original MIT tests the masses were not covered with first contact. It is not secret that then removing the first contact layer the mass is charged. But no first contact has been applied or removed for the length of the discharging measurements.
5) Discharge gage (aka 'cold cathod') this is the very low pressure sensor at the ETMY tank (BSC10). It operates in a similar fashion to an ion pump in which uses ion to pump out some air to sense the pressure. The amount of ions this one uses is considerably smaller than the ion pump (see below) however it is much closer to the ETMY mass.
6) Ion pumps. This is a new game player. The MIT ion pumps were not used during their tests, so there is not measurement of their effect on the mass charge. Torsion experiments with fused silica fibres indicate that ion pumps do cause considerable charging. As I have describe above this is consisten with what we have seen. The ion pumps were OFF between discharge runs and in both cases we measured charge reduction on the ETMY, however after they were turned on they ETMY started to charge again (notice that the LED iluminators were ON during the 2nd discharge and never turned OFF yet so if it was the iluminator charging the ETMY then why we saw charge reduction one day after the discharge?). The only think that may not agree with the ion pump being the charger is why the first measurement (Measurement 1) shows such small charging levels while the ion pumps were ON for days before that measurement took place?
However let's see now the results of today's measurement (nothing was changed respect to yesterday's measurement) there is only 1 day different in the measurement, see notes with the measurement values attached here as well as the plots of Normalized pitch and yaw deflection vs. VBIAS comparing measurements of today (identified in the legend as ending with 23) with the ones from yesterday (identified in the legend as ending with 22), I have also plotted a zoom version looking at the Veff (or zero crossing of the deflections. I show next the table summary with red for yesterday's results and green for today's. It is clear that the EMTY has charged considerably, which again seems to confirm the ion pump charger hypothesis.
|
UL after22 |
UL after23 |
UR after22 |
UR after23 |
LL after22 |
LL after23 |
LR after22 |
LR after23 |
Veff PITCH [V] |
123 |
228 |
65 |
-169 |
61 |
-130 |
85 |
153 |
PITCH slope [10-7 µrad/V] |
-2.565 |
-2.606 |
2.1605 |
2.305 |
2.154 |
2.053 |
-2.612 |
-2.635 |
Veff YAW [V] |
177 |
320 |
70 |
-11 |
54 |
-39 |
100 |
227 |
YAW slope [10-7 µrad/V] |
-2.178 |
-2.226 |
2.428 |
2.489 |
-2.271 |
-2.215 |
2.374 |
2.308 |
Still there is one more variable in all this charging measurement game. Remember that one of the ESD quadrants could not be driven (quadrant LL) because its driving signal was going through the ESD low pass filter box, designed to filter the ESD BIAS frequencies above 1Hz, and the driving signal for these measurements is at 4Hz. This was solved before yesterday’s measurements, however are we sure that this change has not caused a different electric field configuration on the ETMY therefore making it incomparable with any other previous measurement? I had tried to run another measurement by bringing the cable configuration to the original case however between some power glitches bring the CDS down and affecting the suspensions controls at end station together with the higher seismic noise due to the sand storm has so far made this measurement impossible as the SNR is too low.
At the moment the ESD cables are as per the old configuration.
(Dan, Alexa)
Since the MC cavity length was adjusted, we repeated the MC cavity length measurement as described in alog 9679.
Data_refl9_short.txt is the data collected using REFL9. ArmCavityLength_v2.m is the script that determines the length given the zero crossing of the projection. The attached plot show the results with a linear regression included.
The cavity length is determined to be L = 16.471698m ± 4um assuming 2 Hz accuracy. The 2 Hz accuracy comes from the accuracy of the IFR plus the extreme rattyness of the transfer function.
Compared to the previous measurement the delta L = L_old - L_new = 0.001914 ± 6um. This is very close to the expected 2mm reduction in length as mentioned in alog 12654.
If desired, we can repeat the measurement and include a zero crossing for REFL45 as well. However, we just wanted to make a measurement and get the result public...
Looking at the PRC length measurement from alog 10642, we can reevaluate the relative length mismatch:
Parameter | Value | Unit |
---|---|---|
FSRPRC | 2.600075 | MHz |
LPRC | 57.6508 | m |
FSRMC | 9.100229 | MHz |
LMC | 16.471698 | m |
FSRMC / 3.5 - FSRPRC | -5 | Hz |
(1 - FSRMC / 3.5 FSRPRC) LPRC | 0.1 | mm |
(1 - 3.5 FSRPRC / FSRMC) LMC | -0.03 | mm |
Right on target! 9.100230 MHz should become the new modulation frequency.
Jeff K., Krishna V. The pressure at the Turbo was 1.2E-6 torr this morning. The transfer function measurements have been less than convincing today. For now, I think d may be (+12.5 +/- 5) microns. I think cross-couplings (twist to tilt) may be introducing larger errors in this measurement than what I'm used to in the lab, where I have a much sturdier platform. The above value of d gives a displacement rejection of (9E-5 +/- 4E-5) rad/m. I have attached an ASD plot of 10k seconds of data from late last night/early morning, showing a very quiet ground, at the level of 0.1 nrad/rt(Hz). The blue curve, labeled tilmeter, is the measured tilt and the reference mirror curve is a ~ proxy for the autocollimator noise. We were hit with a dust storm around 4 PM this afternoon which produced lots of tilt noise, shown in the second graph, which was taken from ~5:25-6:30 PM. The noise near 0.1-0.5 Hz appears to be a factor of 10-100 worse. Wind speeds in that time frame were in the 30-40 mph range. The pump has been turned off and I'm doing another overnight measurement. Tomorrow we will decide how much mass to add to try and reduce d as much as possible.
(Dan, Koji, Masayuki, Alexa)
We measured the HAM6 septum angle using a laser pointer. We confirmed that there was no observable vertical component to the wedge angle, and then proceeded to measure the horizontal angle. We pointed the laser pointer such that the retro-reflected beam off the surface of the septum returned approximately directly back. Then we measured the distance from the second reflection to this point. This distance was 17mm. The distance from the laser pointer to the septum was measured to be 360mm.
This gives: wedge horizontal angle: 17/360 * 180/pi /2 /1.45 = 0.93 deg
In the equation above the factor of 2 comes from the optical lever effect. Meanwhile the factor of 1.45 comes from applying snells law with the index of refraction for glass and assuming the small angle approximation (see attached drawing).
This measurement was not extremely precise, but was close enough to the expected value of 0.75 deg.
In the attached picture, you will see the retro-reflected beam, which is almost ontop of the outgoing beam, and the second reflected beam. We used the ruler below to measure the separation.
Koji
As the things are getting more precise, I pulled out my old raytracing calculation for an wedged angle.
This gave me the wedge angle of 0.91deg.
This includes the new effect of
- Refractive index of fused silica at 632.8nm (n=1.457)
- Average thickness of the window ((0.948+0.870)/2 = 0.909" = 23.1mm)
- Non-orthogonal input angle
The primary beam is distant from the laser diode by -8mm while the secondary beam from the backsurface is at +9mm.
This condition was fullfilled when the wedge angle is 0.91deg.
The attached plots are:
Attachment1: The overview of the rays
Attachment2: Zoomed view of the optic part
Attachment3: Zoomed view of around the source
[Alexa, Masayuki, Dan, and Koji]
A beam found on one of the OMC QPDs. The fast shutter beam dump elevated.
- We went into the cavity and spent some time to align OM2 and OM3 to have a beam aligned to the OMC.
- We confirmed the beam is hitting the QPDA (short arm one).
- The beam is still misaligned (mainly in yaw) at QPDB.
- We want decent damping of OM1. OM1 has too much tilt and requires adjustment on OSEMs. We are working on this.
- The beam dump for the fast shutter need to be raised by an inch. Betsy provided us a set of suspension addon masses
to make the post longer. Using one of them, We successfullt elevated the height of the beam dump by 20mm. This was
enough to accommodate the beam including the possible wobbling of the mirror on the fast shutter.
In trying to re-center the BOSEMs for OM1 (which were off thanks to the large pitch necessary to center the beam on OM2), we found that the flag of LL was contacting the inside of the BOSEM (either the LED or the PD barrel, can't tell which). This is the same problem that was observed previously on OM2 in alog:12352 and alog:12382. We managed to twist the LL BOSEM by turning one of the PEEK adjustment nuts in, and turning the other out; this appeared to free the mirror, but the power glitches struck before we could check if the flag was appropriately centered in the PD/LED path. Checks from the control room after CDS was recovered were not promising. Probably the pitch correction is so large that we'll need to make some adjustments to the balance of this mirror. To be continuted tomorrow.
Noticed that the turbo at the X-end station was also tripped off (electrical?), also its QDP80 -> Restarted and vavled-in -> resuming pumping at X-end
Jim, Cyrus, Dave
The large wind event was preceded by several power glithches which impacted on the DAQ and killed the front end computers.
After waiting to ensure that the power was stable again, we remotely (via management port) power cycled the FE computers at the end and mid stations and reset the MSR computers. Generally, computers not on the dolphin networks started themselves, some needed a power cycle. Once all the computers were all booted, they all started their models. At that point we discovered the Dolphin IPC in the MSR was non-operational. We suspect the glitchy nature of the outage put the Dolphin switches in a bad state. We stopped all models running on MSR computers attached to the Dolphin network (all but the PSL and SUSAUX). Power cycled the Dolphin switches. Rebooted the FE computers via front panel RESET button. Some models did not autostart and needed their "BURT_RESTORE" button pressed, which we did.
The DAQ was showing bogus data for slow channels (e.g. wind speed below 10mph when it was 50mph outside), so a clean restart of the DAQ was done. The NDS machines took many minutes before they got started, not sure why at the momemt.
Two systems started with a IRIG-B timing signal about 400 (should be 15), which then drifted down to nomimal over 20-30minutes. These were h1sush34 and h1iscey. We allowed these to become good rather than power cycle their IO Chassis.
Once the models were running, stable and had communication, I enabled the SWWD systems to drive the DACs.
Shiela is handling the recovery of the Beckhoff and PSL.
I opened GV2 and disconnected the leak detector -> noticed YBM and XBM turbos had tripped -> I assumed this was due to the gas bump from the gate annulus volume being too high for the, relatively low, safety valve set points of 5x10-2 torr -> Increased set point and restarted YBM turbo -> valved-in YBM turbo -> Increased set point and tried to restart XBM turbo but its QDP80 was also off(?) -> Restarted QDP80 but now turbo trips on vibration at about 75% rpm -> attempted to spin it up a few times but no luck -> This symptom has happened before and has been bypassed by introducing a gas load at the turbo inlet via cracking opening the "up-to-air" needle valve -> I didn't try this now will revisit tomorrow Pumping YBM, Vertex and XBM with YBM turbo tonight
Gains to decouple bottom mass (pitch and yaw) angular motion from top mass longitudinal drive at DC were implemented in L2P and L2Y components of the top mass drivealign matrix of MC2 using the dc magnitude of the top to bottom L2P (urad/Force cts) and P2P (urad/Torque cts) for length to pitch decoupling and L2Y, Y2Y for length to yaw decoupling. This should be tested.
alpha [urad/Force cts] |
beta [urad/Torque cts] |
alpha/beta [Torque cts/Force cts] | |
in pitch | -3.90E-03 | 5.37E-01 | -7.26E-03 |
in yaw | 2.20E-03 | 4.11E-01 | 5.35E-03 |
Attached are the plots of the undamped top to bottom TFs with cross couplings terms. The data cursor shows where I get the numbers from the table above from. The minus sign for Presp/Ldrive comes from the -180 degrees phase.
restarted it at 7:38pm since it crashed after the first power glitch
Photos of the violent dust storm approaching are here:
https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=13366
and here:
https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=13365