patrick.thomas@LIGO.ORG - posted 12:16, Tuesday 20 September 2016 (29841)
OPS: reset of HEPI L4C Accumulated WD Counters Tuesday 20 September 2016
Reset HAM2 and ITMX.
Reset HAM2 and ITMX.
The 56.8406 Hz comb that Keith pointed to in recent data shows up in End-X magnetometers, and seems to be an isolated spike repeating at that rate. Keith's recent alog 29783 points to a comb with a spacing of 56.8406 Hz that has been seen before in ER9 and in the one-arm tests. TJ's Bruco results in alog 29828 show that the 56 Hz region of DARM is coherent with the SEIRACK and FLOOR magnetometers. The first plot shows the coherence just at the fundamental frequency. The line seems strongest in SEIRACK_Y, so I tried folding the channel. The frequency doesn't seem to be an exact multiple of anything digital. But taking 1153 samples at 8192 Hz comes very close to 8 times the comb spacing, so I used that. The result is in plot two, which is four hours of mag data from today folded over 1153 samples. Eight spikes show up very clearly, so there does seem to be some kind of spike happening every 0.01759 seconds. There are some other channels showing up in Bruco that may not be connected to anything - PEM-EX_ADC_{8,12,13}. I think these are spares. At EY there are some magnetometers on these channels but these look too dead for that. They seem to see the same line, so maybe this really is some kind of DAC problem, or maybe there's just some electronic pickup. We should get an independent magnetometer out to EX and look around for this comb, and also just confirm that it's not native to the DAC.
We can rule out any mechanical mode around 17.7 kHz causing instability. Any mode around 17.7 kHz does not have good overlap with 3rd order mechanical mode. It most probably aliased mode from ~ 47.7 kHz. Interestingly, according to my FEA model of ETM, there are three candidates: 47.767 kHz, 47.811 kHz, and 47.827 kHz mode. These modes corresponds to the mode number #542, #544, and #545, respectively. Mode shapes are attached. All of them have reasonably high overlap factor: # freq, kHz TEM Overlap ITM Overlap ETM 542 47.767 02 0.028 0.026 544 47.811 02 0.057 0.050 545 47.827 02' 0.026 0.024 '-see figure The overlap with 2nd order confirms that one of these modes must ring up. The parametric gain for tuned ETM is shown in attached figure. It is not possible to pinpoint which of these modes is observed. According to the overlap value there is slightly better chance that this mode belongs to ITM and it's mode #544. However, mode #542 is the closest to the observed frequency (I do not expect to see error of computed frequency larger than 0.1% at 50 kHz range). Using ring heater we should be able to rule out at leas one of the three modes.
WP 6166 This gauge pair has been replaced with a Inficon BPG402 gauge. The ITMX ESD and HAM6 HV supplies have been reset. Dave will remove the channels from the DAQ.
I have restarted the medm web capture program for the updated medm screens.
PSL tripped at 16:14:52 UTC (9:14:52 PDT). Apparent cause according to the interlocks is a glitch in the power meter flow. The first attachment shows the power meter flow dropping just before all the other flows fall off. The second attachment shows the power meter flow dropping just before (<1 second) the NPRO turns off. It should be noted that there has been work around the PSL racks this morning, although the NPRO does not at this time appear to be the cause of this trip.
When restarting the laser, I had to reset the Beckhoff PC to get the control software to accept input from the mouse. I haven't seen this occur before. In turning the crystal chiller on and off (restart after the trip, then restart after the Beckhoff PC restart), I had to add water in 2 stages: 250mL after the initial trip, and 220mL after the Beckhoff PC reset. This is standard procedure after the crystal chiller is reset; when the chiller shuts off the pressure in the line causes water to leak out of the chiller reservoir fill port. There is no indication that there is a leak in the PSL.
The PSL is now back up and running.
WP 6164 When h1brsex reboots, it appears that the C# code starts before the PLC and therefore fails. This does not appear to be case for h1brsey. I have taken out the automatic starting of the C# code and EPICS IOC at h1brsex. I have left the PLC to automatically restart. I restarted h1brsey twice to check that it had not worked by random coincidence, and it worked both times. I restarted h1brsex more than once and each time it failed.
According to the documentation the C# code should be starting after the PLC.
LHO WP 6151 has been completed. The external links have been upgraded from 1Gb to 10Gb transceivers. Monitoring and tuning of the new connections will be performed in a non-disruptive manner.
Current transceiver light levels:
plugged: SFP/SFP+/SFP28 10G Base-ER (LC) module temperature: 62.46 C Voltage: 3.25 Volts RX: 0.13 mW (-8.80 dBm) TX: 1.28 mW (1.09 dBm)
After settling for a bit: RX: 0.13 mW (-8.81 dBm) TX: 1.61 mW (2.09 dBm)
This laser tripped this morning at around 03:00. Found the NPRO power supply was off. Suspected problem is either a mains power glitch or the UPS for the NPRO power supply had a hiccup. System came up okay except for the ISS, which I am guessing is related to changes made to the sensors as described by Keita in a previous entry.
Attached is a plot of the temperature rise of the laser heads when the laser tripped. The sudden rise in temperature corresponds to when the laser tripped.
Richard had me trend the PEM power channels and it definitely looks like there was something going on around that time. Oddly, I couldn't trend the full data though, it just gave me a line. So this is second trends of the mean.
"System came up okay except for the ISS, which I am guessing is related to changes made to the sensors as described
by Keita in a previous entry."
This change was to the out of loop PD (PDA) and was not the cause of the ISS problems. The change was also reverted last night, so the ISS still needs fixing.
Sheila, Jenne, Terra, Matt, Lisa More complete entries will follow, but the bottom line is that we are still missing the main source of noise below 200 Hz (above 200 Hz jitter/intensity noise should explain most of the noise we see). We had another long lock today at 50W (broken by a new PI ) and we tried several things: - the reduction of the flow rate done in the PSL this morning didn't really last (see here ), so the PSL table motion is only very very marginally better than it was; - Jenne played more with MICH and SRCL FFs, the subtraction is not perfect, but their contribution is negligible at 100 Hz; - we did AL2, and that reduced a lot the noise below 30 Hz; - we checked that all the actuators were in their nominal low noise state state; the ITMs ESD had the low pass OFF, so we engaged them (no impact on the noise) - the ESDX bias was at zero as expected, all coils where in nominal low noise states; - Jenne lowered the 9 MHz modulation depth (see https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=29817), getting rid of the 900 Hz peak; - Sheila did some exploration of POP offset and IMC WFS offset to minimize jitter (to be continued) - OMC alignment to be checked as well; - I believe the calibrators have checked the low frequency calibration, and they are confident that it is not off by more than 20%; - Matt tried to close the ISS with the photodiode in the new location (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=29812), but that didn't work. Data around Sep 20, 6 UTC should be good to look at, close to the best low noise state achieved so far. Bruco, actuator noise estimates, checks of couplings with (new wrt O1) TCS-related electronics, and other suggestions are very welcomed.
TJ, Andy, Josh
Bruco results are here using OMC-DCPD_OUT_DQ as the main channel: https://ldas-jobs.ligo.caltech.edu/~thomas.massinger/bruco_1158386417_OMC_DCPD/
The targeted time was 10 minutes beginning at 6:00 UTC on September 20th.
OMC Null shows up pretty strongly at very low frequencies, maybe there's a slight detuning between the OMC DCPDs?
It looks like most of the noise below 100 Hz is correlated with vertex length control signals (MICH error/control, PR2/PRM/BS drive signals).
Attachment 1 zooms in on the coherence between MICH/PRCL and DARM. Coherence with MICH peaks at around 30 Hz and falls off as it approaches 200 Hz.
From 100 - 200 Hz it looks like the IMC alignment error signals start to show up more strongly.
I'll happily rerun with a different target channel if it would be useful.
Matt, Terra
We have 2 modes which are very close in frequency (both around 15540kHz) and we have had some trouble damping them. To help with this, I modified the PLL filters for these modes to support a lower bandwidth loop, which I hope will be less easily "distracted" by the nearby mode. The low-pass in FREQ_FILT1 (usually a 1Hz pole) is a modified elliptic which provides significant attentuation at 0.5Hz (ELF0.5), and requires a UGF of about 100mHz. To support this, the integrator in FREQ_FILT2 is moved down to 30mHz and the gain is reduced to 0.3. (Note that ELF0.5 has a gain of 0.5 below the cutoff. This helps to move the UGF down when this filter is on.)
So far this configuration has been working, but should it prove problematic the old filters can be moved back from FM2 to FM3 (which is the FM operated by the guardian).
While working on these modes, we found evidence for coupling between the ETMX ESD drive and Trans QPD signal.
I injected a sine wave at 15540 Hz (where there is no known mechanical mode resonance) to each ETM ESD drive through the PI damping loop of Mode17 (ETMX) and Mode25 (ETMY) and watched the response in the QPDs (H1:SUS-ETM?_PI_DOWNCONV_DC1_INP_IN1). I turned off all other noise and injected 10 000 and 50 000 counts set amplitude. We find that the X-arm TransMon QPD sees greater signal even when excitation is to ETMY.
In spectrum below, dashed and solid lines of the same color are the X-arm and Y-arm QPD responses, respectively. Blue and green are with 10k count excitation and orange and red with 50k counts.
To this end, we found a more reliable response from our PLL damping scheme by bringing the error signal for both modes (17 and 25) from Y-arm QPD, despite Mode17 being an ETMX mechanical mode. We should have a look at the cabling in the end stations for ESD and QPD signals.
Terra - can you repeat these ESD -> QPD coupling measurements with no light (IFO not locked)? This would help disentangle electrical cross-coupling from optical.
Took a 12 hours trend of the PSL Chiller vital signs after this mornings flow adjustment. The various pressures and flows changes in the manifold, chillers, AMP and PWRMETER are not out of the range of expectation. The change in the PWRMETERFLOW is a bit larger than the changes in the other parts. Will monitor this over the next few weeks to make sure all is well. The interesting result is the changes in the four head flows and temperatures. On heads 1 through 3 the flow was reduced by about 0.1 l/m. On HEAD1 the temperature remained constant. On HEAD2 and HEAD3 goes up by 0.1 degree. What is interesting about this is the change in temperature actually smooths out the variation of the high/low temperature range. Need to monitor to see if this is holds up over time. HEAD4 has always been the PSL poster child of different, and continues this trend. When the flow was first adjusted it took a relative large drop (0.2 l/m) and over the next three hours the flow recovers to within 0.02 l/m of its original value. The temperature at first goes up and then settles down to its normal value. Whereas on HEAD2 & HEAD3 the temperature goes up and smooths out the high/low fluctuations; on HEAD4 the temperature goes up, then down and then smooths out the high/low fluctuations. It should be noted these changes are actually relatively small. The Crystal Chiller flows around 22 l/m and the Diode Chiller flows around 32 l/m.
Looking at Jeff's plot it looks like after a while all the head flow numbers came back almost exactly where they were before the work in the PSL... The plot shows the noise measured by the PSL accelerometer last Friday (blue, Sep 16) compared with today right after people left the PSL table (purple) and a couple of hours ago (cyan - by this time the flow rate came back pretty much were it was). Red is the reference from ER9 (July). Any chance we can lower the head flow and keep it low?
Whilst in the enclosure the flow rates were adjusted to be as low as possible without tripping the flow rate watchdog. It's hard to predict, at least for me, where the eventual flow rates will settle out. We could increase the amount of water by-passed at the water manifold to reduce the flow rates further, which I would prefer to be done with the laser off, and then allow the flow rates to settle down for a long period of time (like half a day or so).
Posted are the ISI HAM & CPS Spectra Checks. Did not notice anything that appeared out of the ordinary.
One thing to note: short 3 BSCs and 2 HAMs. Also, I'd say ITMX H3s are elevated and BS St1 H1 is higher relative to previous check Sept 7. Call em like I see um.
