The IOC for the MetOne 227b dust monitors was accidentally running. I stopped this one and started the one that Jim B. wrote for the new MetOne dust monitors.
(John W, Daniel S, Keita K)
The viewport on the HAM5/6 septum was rotated by 120 degree clockwise. Torqued to 16 lbs.foot as per the spec. We didn't have to use a new O-ring as the old one was intact.
Corner Station - we are currently vented. OpLev - Doug and Jason doing work at End-X (waiting on CDS to begin ) Also, the mitigation of equipment damage was brought up End-Y - Borja may me taking more measurements using higher resolution. HEPI currently NOT running. IRIG-B not running cleanly after power restored End-X - Krishna will be connecting the tilt-meter to the LIGO DAQ B&K Hammering will be taking place at end stations. Still unclear as to where it will start.
Late entry for Wednesday's restarts
model restarts logged for Wed 13/Aug/2014
2014_08_13 00:39 h1fw1
2014_08_13 01:39 h1fw1
2014_08_13 09:29 h1fw1
2014_08_13 10:48 h1nds1
2014_08_13 10:57 h1nds1
2014_08_13 13:15 h1iopoaf0
2014_08_13 13:15 h1oaf
2014_08_13 13:15 h1odcmaster
2014_08_13 13:15 h1pemcs
2014_08_13 13:15 h1tcscs
2014_08_13 13:24 h1iopoaf0
2014_08_13 13:24 h1oaf
2014_08_13 13:24 h1odcmaster
2014_08_13 13:24 h1pemcs
2014_08_13 13:24 h1tcscs
2014_08_13 18:31 h1ioppsl0
2014_08_13 18:31 h1pslfss
2014_08_13 18:31 h1pslpmc
2014_08_13 18:33 h1psldbb
unexpected restarts of h1fw1. Some DAQ min trend work required nds restarts. Stuck PEM ADC channels required h1oaf0 restarts (residual from Tue power glitch). PSL restart part of power outage shutdown prep.
Power outage day. All CDS computers were restarted. The total number of restarts is 138. Full log in attached text file. No unexpected restarts.
In /ligo/cds/lho/h1/burt/2014/08/13/08:10 I ran: burtwb -f h1ecatx1plc1epics.snap burtwb -f h1ecatx1plc2epics.snap burtwb -f h1ecatx1plc3epics.snap burtwb -f h1ecaty1plc1epics.snap burtwb -f h1ecaty1plc2epics.snap burtwb -f h1ecaty1plc3epics.snap
Jeff K., Krishna V. I analyzed the data from the T240 (labeled as STS atm) on the ground (about ~ 1 m away) and have attached three plots. To make them I used 7k seconds of tiltmeter and T240 data from Tuesday night, roughly between 9:10 PM to 11:10 PM. I resampled both data sets to 1 Hz to get them on the same time base. Then I compared the phase of the T240 signal (converted to acceleration) against the tiltmeter signal in 10-100 mHz range and time shifted them until the phase lined up, as shown in the second figure. I've also shown the residual, obtained by subtracting the time series of the two signals. We are limited by being on different clocks and DAQs, but even so, the coherence is decent. Just as I saw in our lab at UW, there is good coherence when both see tilt and very little coherence when the seismometer is seeing horizontal displacement while the tiltmeter still sees tilt. Edit: I've attached the data in the format: columns [time T240*w^2/g tiltmeter]
During the power outage yesterday, verified CPS units for BSC1, BSC2, and BSC3 had modifications per E1300251. This was to address a high frequency oscillation in the return path. BSC1 SN 12867 BSC1 SN 13525 BSC2 SN 13449 BSC3 SN 12951, 12892, 12959, 12958, 12961, 12895, 12904, 12945, 12948, 13430, 13235, 13444 All units had modifications complete. Filiberto Clara
J. Kissel, T. MacDonald While waiting for the CDS system to be restored after the scheduled power outage, and for B&K to release licenses, Tim and I spent some quality time with a 3 [lb] sledge hammer, a GS13, some concrete, and SR785. We found a large (~45x12 [sq.ft]) area of empty open LVEA floor (don't tell Jodi!), set up a GS13 in the middle, and used a B&K hammer to hit 10 locations along an 40 [ft] line, each separated by 5 [ft]. Sadly, I think the result is null, but still -- null results can be interesting results. If one squints their eyes, one might be able to claim some coherence with a feature between 5-10 [Hz], but we would need more data to determine if anything in this measurement real slab dynamics. Plots attached, details below. Details ------- Timeline - Hooked up S13 and Hammer to SR785 - Tried several binwidths and frequency ranges, looking for interesting features - Didn't get any consistent coherence; looked at spectra of S13; decided its spectra looked too featureless and time series didn't look too promising - Grabbed GS13s 568V and 574V, some power supplies, and freshly made power cable - After some struggles with a wonky power supply, confirmed that both GS13s 568V and 574V are fully functional (looking at time series alone) with different power supplies - Hooked up GS13 to SR785, with hammer - Played with bandwidths and frequency ranges again (went as low as 62 [mHz]), settled on 1 [Hz] binwidth, from 1-400 [Hz], with 10 avgs. We justified the range with "the sledge wouldn't be able to move enough mass to excite anything lower than 1 [Hz], and we're already getting crap coherence, might as well make the measurements quick." Cable construction: - Used D0902011 to determine pinout of signals coming out of GS13 military connector. The GS13 needs +/- 15 [V] on pins D and F, respectively with pin R as GND. The +/- legs of the differential GS13 signal are on K and L, respectively. - Followed signals through Signal / Power pigtail breakout cable that came already attached to the GS13. D to pin 1, F to 3, and R is connected to both 2 and 4. The + GS13 signal is connected to the inner coax of the BSC, and - to the sheild (*tsk*tsk*). See attached pictures for connector pinouts of mil and power connections. - Aaron/Richard fabricated a power-to-banana cable. (+15 [V]) 1 to Red, (GND) 2 to Green, (-15 [V]) 3 to White, (GND) 4 to Black, though the banana ends of both Green and Black are Black. Calibration of TF: - B&K Hammer: - gain = 0.2356 [mV/N] (from calibration datasheet) - assumed frequency independent - GS13: - 2200 [V/(m/s)] (from eLIGO knowledge, confirmed by a few web sources) - assumed frequency independent, since instrument response is flat above 1 [Hz] natural frequency of the instrument - D050358 preamp gain of (1+R2/R1)*(1+R4/R5) = 40.2 [Vout/Vin] - Integrated once (divided by 2*pi*freq) to get displacement units - Divide by 2 for only reading in one leg of differential signal Data and analysis script live in ${SeiSVN}/seismic/Common/Data/2014-08-14_LHOLVEASlabCharacterization/
Peter, Jim, Dave
Peter K found that he could not open the PSL shutter. We did the usual MEDM button presses which used to work to no avail. After restarting the FE and beckhoff ends several times, we reverted the PSL from the new RCG2.8.5 to its old 2.8.3. In this configuration we were able to open the shutter (though not cleanly). Tomorrow we will investigate (using the DTS) if it was the upgrade which caused this problem. For now we will leave h1psl0 at RCG2.8.3 and the laser is operational.
After power was restored, I brought the laser back up. A process not quite as simple as pushing a button but was relatively a painless process. All the servos were locked and running. When all of a sudden I noticed that the pre-modecleaner reflection and transmission were zero. However the laser was up and running. The Beckhoff computer indicated that the external shutter was closed and that the EpicsAlarm was set. Doing a system reset and a computer reboot did not fix the problem. Neither did doing the steps Dave mentioned that were done previously to open the shutter. The laser is up and running with the watchdog set. All servos were left locked.
The following suspensions had their IOP as well as USER watchdogs reset. A burtrestore was done, restoring a snapfile from yesterday afternoon (August 13 at 17:10). Then guardian was set to "DAMPED" for all of them (except the IMs since no guardian) : MC1 MC2 MC3 PRM PR2 PR3 SRM SR2 SR3 BS ITMX ITMY ETMX ETMY TMSX TMSY OMC.
Currently, no alignment offsets are engaged.
Richard, Filiburto, Patrick, Cyrus, Jim, Dave
VME crates for vacuum controls were restarted, medm screens were operational in the control room immediately after power was restored.
All front ends a back online. A few started with high IRIG-B timing numbers which gradually drifted good. The longest recovery was h1seih16 which took several hours.
All front ends are now running RCG2.8.5 with the exception of h1psl0, please see my separate alog.
Only one front end is not running correctly, h1susauxex has a very large IRIG error which is not fixable by a hard power cycle of the FE-CPU and the IO Chassis. We also performed a power cycle of the timing fanout and IRIG-B fanout at EX. We will continue the investigation of h1susauxex tomorrow, all other EX systems are running OK.
The DAQ is back up and running after the rack reconfiguration.
After the power was returned, we took the opportunity to clone the boot disk of h1boot and perform an fsck and upgrade of cdsfs0.
web snapshots of medm screens were started for offsite monitoring of vacuum vent.
tomorrow we will start the secondary systems and test stands.
Swapped the M4 terminal EL3104 with a spare in the corner 4 chassis. This fixes the problem seen in alog 13364.
Verbal approvals from Mike L., John W. and Andres R. as I was unable to access eform -> I will replace PT180B, PT120B and PT170B tomorrow while John and others re-clock the HAM5/6 septum viewport. NOTE: I tested the three new CC gauges that will be installed on an aux. cart and using PT114B's cable.
Negative 18V Power was damaged upon restoring power. Chassis with no power switches were temporarily powered with only positive 18V. No damage to the chassis was observed. Damaged power supply S1300278 was switched with a spare located at EY. Everything appears to be functioning now.
12:00 – Power is back On. CDs recovery process ongoing 13:25 - Fire Department on site (Maintenance) 13:45 - Heading to End X to power-cycling electronics- Filiberto 13:45 - Heading to End Y to power-cycling electronics- Aaron 13:48 – 15:35 Going into the LVEA (B-K Hammering) – Tim/Jeff K. 14:03 - Transitioning the LVEA to Laser SAFE – Richard 16:00- CDs recovery process continues ...
Richard, Cyrus, Jim, Patrick, Kyle, Dave.
The site was without power between the times of 08:32 and 09:05 PDT.
We have recovered the vacuum controls systems. h0vemr, h0vemx and h0vemy rode through the outage and needed no recovery. h0velx, h0vely, h0veex and h0veey required a reset button press, reboot, code start and burt restore (from 23:00 8/13) because they lost power. Robert reports the LVEA VME lost power only seconds before the power came back.
Reracking of MSR is ongoing.
We took the opportinity to perform an fsck on cdsfs0's /ligo file system as this was showing errors.
The clone of h1boot's boot-disk is ongoing. workstations and front end restarts will occur after the clone is completed.
(Koji, Alexa, Dan)
We examined the beam path in HAM6 to OM1 in order to figure out the angle of the beam. We made measurements at four different points. Using an (x, y, z) coordinate system with z = up, y = East, x = South, we find at (all in mm):
Edge of table: (20.32, 0, 98)
Intermediate point 1: (0, 552.72, 95)
Intermediate point 2: (-25.4, 1219.2, 93)
OM1: (-40.64, 1574.8, 91)
The error of the measurement in height is ±1mm, and the error along the x, y axis is ±2.5mm. The attached layout shows the original (red) beam path and the new (green) beam path. From this layout, one can see the actual vs. measured angle deviation in height and along the hoirzontal plane. Using the points above, we made a linear regression and determined the vertical angle of the beam to be 4.3 mrad. The attached plot shows the data with error bars and the linear fit.
So the situation now is this:
| angle of the beam [mrad] | position of the beam at OM1 design center [mm] | |
| PIT (positive=up) | -4.3 | -10.6 |
| YAW (positive=North) | (-39.7, though the absolute number is not that important here.) | -50.8mm |
Because of this, Koji had to tilt the OM1 up by about 4.3mrad, which is big, and I'd say that there's a high chance we will want to fix the beam angle some time in the future (e.g. larger bounce to alignment coupling). YAW is not that much of a problem because there's enough space to absorb -50.8mm.
We've been discussing how to alleviate this, and the simple hack is to rotate the septum window, which is supposed to have a 0.75deg horizontal wedge which causes 5.9mrad deflection.
According to ICS (via Joe), we should have D1101092 S/N assembly, which should have D1101005 window S/N15, which has dimension measurement that suggests 0.745deg wedge.
However, Koji measured the wedge using laser pointer and got 0.89deg which should cause 7.0mrad deflection. His measurement also suggests that the thickest side is facing south.
Now, when we rotate the septum window by X (positive=clockwise), PIT deflection was zero before but now the beam is deflected vertically by sin(X)*5.9mrad (or 7.3mrad).
Horizontally, the deflection is -5.9mrad (or -7.3) before rotation, and -cos(X)*5.9mrad (or 7.3) after, so the change in the angle would be 5.9mrad*(1-cos(X)).
If we optimize the septum rotation (which only changes by 30deg steps) for 0.75deg septum we need to rotate the septum by 120 deg clockwise.
For 0.89deg septum wedge, it would be 150deg clockwise. See below.
(The beam position change at OM1 is calculated by using 1.93m as the distance from OM1 to the septum.)
| Septum rotation (deg) |
Septum wedge (deg), and deflection (mrad) |
PIT deflection change (mrad) | PIT beam pos at OM1 (mm) | PIT beam angle (mrad) | YAW deflection change (mrad) | YAW pos at OM1 (mm) |
| 120 |
0.75, and 5.9 |
+5.1 |
-10.6+5.1mrad*1930mm |
5.1-4.3=+0.8mrad |
+5.9+2.95 =8.85 |
-50.8+8.85mrad*1930mm |
|
0.89, and 7.0 |
+6.1 |
-10.6+6.1mrad*1930 |
+6.1-4.3=+1.8mrad |
+7.0+3.5 |
-50.8+20.3mm |
|
| 150 | 0.75, and 5.9 | +2.95 |
-10.6+2.95mrad*1930 = -4.9mm |
+2.95-4.3=-1.35mrad |
+5.9+5.1 = +11.0 |
-50.8+21.2mm |
| 0.89, and 7.0 | +3.5 |
-10.6+3.5mrad*1930 = -3.8mm |
+3.5-4.3=-0.8mrad |
+7.0+6.1 = +13.1 |
-50.8+25.3mm |
Anyway, there's not much difference, but since the ICS says 0.745deg wedge, we need to rotate it by 120 deg clockwise if we decide to do it.
Keita and I concerned about the AR reflection from the septum. We thought we should at least check where the AR reflection goes.
This required to make a 3D version of the ray tracing. The result is, in short, the rotation of the wedged window(by 120 or 150deg)
makes the returning beams closer to the arrangement with the nominal beams. They fly about 30-40mm North of the aperture on Faraday.
In this entry, the wedge angle of 0.75 deg is assumed.
The "nominal" beam means: "Use the HAM6 dawing. Assume this incorporates the wedging effect by the septum window."
The "actual" beam means: "Use the measured beam geometry in HAM6."
The "actual+120" and "actual+150" means: "The beams expected by rotating the septum by 120 or 150 deg in CW. The "actual" beam used for the calculation.
1st attachment is an example view of the ray tracing result.
2nd attachment shows the spot positions on OM1 viewed from the back side of OM1.
Rotation of the septum by 120 deg makes the spot close to the "nominal" beam position.
"+120deg" gives us better result than "+150deg".
Note that the result I obtained here are consistent wth Keita's handwriting calculation for the OM1 spots.
3rd attachment
The beam was back-traced to HAM5. We expect that there is a 20mm aperture (iris) at 315mm from the septum window.
It is assumed that the apertue is located at the beam properly. The primary and secondary reflections are located about 35~40mm North of the aperture.
According to D0900623, these beams might be hitting the beam dump for the PBS, but not so clear.
4th attachment
This time, the actual beam was traced-back. Without rotation, the secondary beam definetely hits the apeture structure.
The primary reflction is ~30mm away from the aperture. The rotation moves the secondary reflection further away to North.
Vertical displacement is 5~10mm. So, we can say that the rotation makes the spots close to the original positions.
In all of these cases, it seems like all ghost beams will fall on the Faraday Isolator Refl Baffle which is mounted on the suspension cage.
https://dcc.ligo.org/LIGO-D0900136 (Output Faraday Assy)
https://dcc.ligo.org/D0902845-v5 (Faraday Isolator Refl Baffle)
The photo of the septum window attached