The DAQ is back up after it was shutdown for a reconfiguration restart at 19:32 PDT. For some reason the monit autostarter failed to restart daqd on h1dc0. This was further compounded by the maintenance changes made this morning which left duplicate slow channels in the configuration. So after a few false restarts I was able to get the system stable.
Here is the data gap on both frame writers.
-rw-r--r-- 1 controls controls 750947268 Aug 19 19:32 H-H1_R-1092537088-64.gwf
-rw-r--r-- 1 controls controls 753765993 Aug 19 21:49 H-H1_R-1092545280-64.gwf
J. Kissel, T. MacDonald We measured a final exploration of ETMY HEPI today to identify the exact mode shape of the 8 [Hz] resonance, given that there were still some ambiguities in yesterdays results. Peppering the piers, boots, housings, crossbeams, and support tubes with hit points, we've nailed it down. The bending point is the crossbeam foot (D020195, see green highlighted part in 2014-08-19_HEPIBoot.pdf attachment) -- the connection point of the suspended stage with the Euler springs. Inward of this foot, everything else is moving in concert -- the crossbeam, the support tubes, and everything else inside the chamber. We took two sets of measurements, one involving tons of points on one pair of piers and crossbeam (that's where we identified the foot as the problem, see HitPattern.pdf), and the other with all four cross beams involved focused more on the crossbeam foot -- as many data points into which we could squeeze the hammer. The "single_side.avi" is the first measurment with only one crossbeam system, and the four "*_both.avi" show the measurement with the whole system. Check out all the visual aides, and be sure to loop your video, since the B&K software doesn't export more than once cycle.
[Alexa Dan Koji]
- We aligned the OM1 transmission path again. The AS_C QPD alignment was reestablished. We can use this QPD as the alignment reference.
- The alignment of the fast shutter was revisited and now we can accommodate the beam without the extra legs. We removed the 20mm leg extentions and the placed the beam dump at the designated place.
- OM2 and OM3 were moved so that we can see the spots on the QPDs.
- We think the beams are on the diodes. In fact the diode SUM outputs are responding to the beam.
However these QPD outputs do not give us linear response to the input alignment.
- We checked the electronics outside of the chamber including the cables from the feedthrough.
They don't show any misbehavior. (Basically we swapped the electronics between OMC QPDs and AS_C.
- We'll investigate the QPD functionality tomorrow.
- We'll want to shake the PZT of the OMC sometime soon.
J. Kissel, K. Venkateswara, K. Ryan We finally figured out the issue behind the increasing current in the ion pump. In the morning, the current was near 1.5 mA, so we decided to do a leak check. There was a small leak through a flange but not enough to account for the ion pump behaviour. I fixed it by tightening the flange further and found no other leaks. Kyle then pointed out that we had the wrong polarity setting on the controller. The Starcell ion pump needed negative voltage not positive. With that sorted out, we restarted the pump. The current increased to ~10 mA, but has since been going down steadily. After ~1.5 hrs of pumping, it was at 0.49 mA at 7 PM. With the reversed polarity, we were probably just observing the leakage current, which may have been proportional to the pressure. I've also attached an ASD plot of some data from last night, when the ground was fairly quiet. The red curve shows the BRS/tiltmeter output while the blue shows the T240 (incorrectly labelled as STS) converted to angle by multiplying by (2pif)^2/g. The coherence between the sensors is also shown.
J. Kissel Was looking to restart the frame builder now that ISC folks no longer need it (for ~5 minutes while the h1dc0 reboots), telnet'ed in, ]$ telnet h1cd0 8087 and restarted it, daqd> shutdown and unlike what happens normally, everything with the data concentrator stayed white, and all front ends DAQ lights went red, and neither came back. This is what happens when you get greedy, and assume the CDS system will behave as you expect it to -- especially if you're me! I've left a voicemail with Dave.
Keita, Kiwamu (WorkPermit 4806)
We placed the IOT2R table back in place (i.e. East side of HAM2). We have re-installed the light pipe that connects the lower left viewport of HAM2's east side to the table. The lexan guard of this particular viewport was then removed.
There was some mess on the optical table of IOT2R because of the PRC length measurement. We need to clean them up at some point. Also the GigEs on the table need the interior ethernet cables and feedthrough patch panels. They probably will be useful. The attached is a picture of the IOT2R after we placed it.
Jim and Dave.
at 15:57 PDT both h1seib2 and h1seih45 suffered a timing fault. The IOP IRIG-B timing channel showed a number of 336685 and 336687 respectively. All user models on these front ends stopped running. The IOP SWWD tripped, zeroing the DAC output voltages.
we were able to ssh onto the machines, and verify the IO Chassis were visible.
After getting permission from the HAM6 commissioning teams, we rebooted the two front ends using the sequence:
Both computers rebooted and restarted all their models with no timing errors. Commissioning time constraints due to staff at/in ham6 did not facilitate any lengthy diagnostics of the broken system.
Arnaud, Dave
we just noticed that the safe.snap link from the target area to the userapps area for the ISI systems: ITMX, ITMY and BS were missing. This meant the settings for h1isibs were not restored after the reboot of h1seib2. We put the symbolic links back into the target directories and manually burt restored h1isibs. Why the links disappeared is a mystery.
Keita, Daniel, Kiwamu
In response to the too-low-power at HAM6 (see alog 13472), we revisited the IM4 trans power in order to make sure that there is no large loss in the input optics chain.
As a result, we confirmed that the incident power on IM4 is about 1.4 W. This corresponds to a throughput of PSL-IMC-Farday chain of 88%. This is consistent with the number that Giacomo and Paul gave us (see for exmaple, alog 9954). So we conclude that the IO chain is healthy.
==== (some details) ====
We measured the IM4 trans beam at the east side of HAM2. The power was measured to be 220 uW using a Ophir Vega without the filter on it.
According to the IO documents (E1300756 and E1300160), the relevant optics have the following properties:
Using these information together with the measured IM4 trans power, we estimated the power incident on IM4,
Power incident on IM4 = 220 uW / T_IM4 / T_RH4 / R_LH1 = 1.432 W
At the time of the measurement, the PSL power was at 1.620 W according to PSL_PERISCOPE_A_DC.
Therefore the throughput of the PSL-IMC-Faraday chain is
throughput = 1.432 / 1.620 = 88.4 %
In HAM6, we measured 3.6mW with the Ophir Vega power meter, with the filter off. (With the filter on, we measured 3.8mW.) The configuration during this measurement was ITMX aligned, ITMY misaligned, 1.620 watts into the IMC.
This is 81% of Daniel's prediction of 4.45mW in HAM6 -- note, the difference between yesterday and today is in the power meter that was used. Kiwamu compared the Vega meter with the Ophir Nova we used yesterday, and found that the Nova, with the filter on, gave bogus measurements. So, yesterday's measurement is not valid.
8:15-8:30 Morning Meeting 8:33-Heading to End Y for B-K hammering work - T. MacDonald 8:35-Crystal chiller has been refilled (added 250 ml of water to bring the level up to maximum) – Operator 9:22-Checking mobility camera in HAM3 - Travis 9:24–Heading to work in the optics lab - Nathan 9:55-LN2 delivery – Praxiar 10:09-Reset HEPI overflow counters for HAM3, HAM4, ITMX, and BS – Operator 11:16-Water delivery - Paradise 11:35- Heading into the LVEA (HAM2 area) – Kiwamu/Keita 11:52-Working at End X – Krishna 12:55-Heading to End Y for cleaning – Karen 13:17-Back from the LVEA – Kiwamu/Keita 14:23-Returned from End Y – Karen 14:40-Working in the LVEA (craning) – Kyle 15:01-Finished working in the LVEA – Kyle 15:14-Moving Leak Detector Setup from the LVEA to End X – Gerardo/Filiberto Note: Analog video cameras were not working today (see Alog# 13487)
Further to making the Beckhoff Ethercat PLC systems more visible to the control room, I have created "GDS" looking screens for the nine PLC systems and linked them to the CDS OVERVIEW screen in the "SLOW CONTROLS" section.
The NTP clients on the three Windows-based Ethercat computers has been set to use a server at 10.20.0.6, and the interval for checking shortened from one week to 1 hour after finding the clocks to be off by several seconds.
[Jim Arnaud]
This morning the quad iop and sus models of the staging building were restarted in preparation of future testing of QUAD09. The system would restart with wrong filters engaged (the 'norm' filters of the damping were on, whereas the input is not calibrated in um), so a new burt was saved with the good config.
Attached are some notes for testing QUAD09 in the staging building, as well as how to start the models
Jim, Dave
All systems (except h1psl0) were upgraded to RCG2.8.5 last week after the power outage. RCG2.8.5 has the bug fix which permits the IOP to disable the SUS DAC anti-imaging chassis via a keep-alive signal outputted from the 18bit DAC card.
Today we tested the control of the AI chassis using the ETMY/TMSY suspensions system via the IOP model h1iopsusey. Jeff K and Arnaud P gave us permission to use the EY suspension systems.
All the tests passed.
Here is what we did:
This concluded the test. The BYPASS on h1seiey was removed. Guardian put both ETMY and TMSY into the DAMPED state.
Maintenance: Reset HEPI overflow counters for HAM3, HAM4, ITMX, and BS.
Andres reported that the analog video feeds in the control room weren't working. The matrix switcher in the MSR was plugged in, but did not appear to have power. I unplugged it, waited a few seconds, and plugged it back in and it powered up. There was nothing that looked amiss with the power supply module or any of the boards accessible from the front of the chassis. It may have been a power glitch, or the power supply is getting flaky in it's old age. It appears to be a modular type under the cover, so it may be easy to replace if needed if it is still available commercially.
Crystal chiller has been refilled (added 250 ml of water to bring the level up to maximum)
J. Kissel, K. Venkateswara Because we continue to find the reference mirror signal -- a proxy for the autocollimator noise -- to be a useful signal to judge where the Beam Rotation Sensor's noise floor lies, we've output this channel out of the laptop's DAC, and piped it into the H1 ISI EX ADC on Channel ADC_0_29, and stored it in the science frames at 256 [Hz]. We expect the reference mirror to be stable, so we've not applied the very-low-frequency 1 [mHz] high-pass, but otherwise the calibration of the signal is identical to that of the beam balance mirror (see 13448): 3.5e-6 [rad/pixel] * 1 [pixels/V_sing] * 1/2 [V_sing / V_diff] * 40/2^16 [V_diff/ct] * 1e9 [nrad/rad] = 1.0681 [nrad/ct] This gain, called "cts2nrad" has been installed into FM1 of the H1:ISI-GND_BRS_ETMX_REF filter bank, as with the beam balance mirror. This required a recompile, reinstall, restart, and restore of the H1ISIETMX. We have not yet reboot the frame builder, so it remains with a 0x2000 status until the next DAQ restart at a more convenient time. I've committed the changes both the top-level h1isietmx.mdl model, and the H1ISIETMX.txt filter file to the userapps repo. We haven't yet added the filter bank to the BRS screen, but we'll get there.
WP 4790 - The default NDS server has been changed back to h1nds1 for the control room workstations. This affect new logins or new shells. Existing shells and programs should not be affected.