Purge air has been mysteriously throttled to a "trickle" going into HAM6 (no longer my battle)
Looking at the voltage of the AS port trigger PD as seen by the shutter controller, we have just under 2V at 50W input. The maximum threshold is 2V, so this won't work. The QPD chassis of AS_C has serial number S1301506 indicating an R23 of 26.7K in the sum output. This needs to be reduced to 12.4K, or we have to reduce the modulation depth of the 45MHz before we reach 50W.
The current configuration of the BSC-ISI's does not use any high gain ST2 RX/RY isolation loops because we dont have a blend filter design that rolls off fast enough at low frequency to avoid injecting GS-13 noise back into the X/Y dofs. However, above 1 hz the suspension point motion is dominated by ST2 RX/RY motion. One way we can improve this is by using relatively low gain AC coupled loops on ST2 RX/RY. With RichM's help I've tested this some on a couple ISI's and it seems to work, but there are trade-offs. First attached plot shows the improvement in ETMX longitudinal motion I've managed to get. Green and brown are the ground for both measurements, pink traces are RY loops on, light blues is off. In the 2-10 hz band I can get about a factor of about 2.5 reduction (I've seen up to ~5) , but at about 1 hz the motion is worse by about a factor of 2. Not spectacular, and I'm not sure that the improvement at a few hz is worth the loss at 1 hz. I've looked at the oplev (second plot, pink is on, light blues is off) and the oplev doesn't seem to see a substation difference at 1 hz.
One reason to do this would be to speed recovery after a lock loss. The ISI RX/RY loops get rung up by the impulse from the suspension, and it seems plausible that the ISI would settle down quicker is the RX/RY loops had more gain.
The designs for the ETMs RX/RY loops are included in the attached pdf. The lugfs are around 1 hz, uugfs are around 10-20 hz, phase margins are around 40-50 degrees. Higher gain loops didn't seem to provide much more isolation, but made the motion around 1hz worse.
21:00 - 23:00 UTC 21:05 UTC Joe and Chris done working on wind fence at end X 21:13 UTC Jeff realigning ITMY, misaligning ITMX for crew at HAM6 22:03 UTC Robert and Ansel at end Y. Do not change ISI from damped to isolated. 22:16 UTC Jason done in optics lab
J. Kissel ECR E1500045 FRS 6014 WP 6051 While attempting to quickly install the ITM updates in between the HAM6 work, I found that ITM models would not compile as is. Errors in the compilation process reminded me that their upgrade would not be as simple as copying and pasting from the ETMs (thanks for all the great error messages, Rolf! They're, sincerely, very useful nowadays). This complexity a result of the differing BIO control of the ESD stage drivers, so the BIO block as a whole cannot be the same. Further, I discovered that I forgot to switch the EUL2OSEM matrix on the FOUROSEM_DAMPED_STAGE_MASTER_WITH_DAMP_MODE.mdl library part that controls the L2 stage of both the QUAD_MASTER.mdl (for ETMs) and QUAD_ITM_MASTER.mdl (for ITMs) from a regular matrix to a ramping matrix. This means that the ITMs are ready for install, and the ETMs need a re-install to gather this bug fix. I'll coordinate with the HAM6 install/repair team and with the CDS crew. I've since made the model changes, and all QUADs now successfully compile, but all QUADs will need to have these models installed and rebooted. For future reference, changes have been made to /opt/rtcds/userapps/release/sus/common/models/ QUAD_MASTER.mdl << Modified the BIO block to use the new Individually Controlled PUM library block, arranged connections from BIO to L2 blocks accordingly QUAD_ITM_MASTER.mdl << (same as above) Modified the BIO block to use the new Individually Controlled PUM library block, arranged connections from BIO to L2 blocks accordingly STATE_BIO_MASTER.mdl << Created new library block for individual control of PUM driver FOUROSEM_DAMPED_STAGE_MASTER_WITH_DAMP_MODE.mdl << Modified COILOUTF bank to accept individual control and switched EUL2OSEM matrix from static to ramping and they've been committed to the userapps svn repo. I attach screenshots of all the relevant parts that have been modified, named after the simulink blocks, respectively.
Removed temporary extension cord for GV15 AIP329 and replaced it with new extension cord, work done under WP 6060, and this closes FRS ticket 5950.
State of H1: OMC work continues, progress on PDs
Activities:
~1330 hrs. local Opened exhaust check-valve bypass-valve, opened LLCV bypass-valve 1/2 turn -> LN2 @ exhaust in 45 seconds -> Restored valves to as found configuration. Next CP3 overfill to be Monday, August 8th.
Chris B. and Ryan F. reported that during the hardware injections in ER9, the H1:CAL-PINJ_TRANSIENT_OUT channel did not fall below 1e-200 counts until about 1 hour after the last BBH injection. He also said this was not observed in O1. One needs to consider the following two points:
First, one cannot directly compare the TRANSIENT_OUT channel from O1 and ER9 because they had fundamentally different units. In O1, the channel had units of strain, in ER9, it had units of counts. In ER9, the strain time series has passed through the inverse actuation filters.
Second, if one instead looks at the HARDWARE_OUT channel in O1 (after the IAF during O1), any impulse response to the transient signal ending would be completely masked by the continuous wave signal that has been added to the HARDWARE_IN time series. Thus it doesn't make sense to have a 1e-200 threshold on the HARDWARE_OUT channel during O1.
To invstigate the ER9 IAF, I have separately computed the impulse response of the different filters. The biggest impulse response comes from the f^2 filter module because it has high gain at high frequencies (see first attachment). This is to be expected.
I also compare the impulse response of the IAF used in O1 and the IAF used during ER9 (second and third figures, respectively). The O1 actually has a bigger impulse response and longer duration. I would surmise from this that any oscillations are actually worse in O1 than they are now.
Conclusion:
From this investigation, there is nothing to suggest that the ER9 IAF are worse than O1. In fact, the new IAF provide improved signal fidelity and the impulse response suggests that the new IAF is actually better than the O1 version.
J. Kissel, K. Arai, C. Gray, T. Shaffer, B. Weaver After a great deal of inspection of the top mass, where I thought the OMCS was locked last night (see LHO aLOG 28883), the in-vacuum team list above identified that an earthquake stop on the bottom mass / breadboard / M2 stage was rubbing slightly. Once this EQ stop was backed off, all DOFs of the suspensions top to top mass dynamics returned to the expected results akin to the former OMC. Nice work team! Attached are screenshots of drawings of the OMC with the offending EQ stop highlighted with a red square, as well as the transfer functions themselves. On the transfer function plots, BLACK shows the former OMC reference measurements, BLUE is yesterday's measurement with the EQ stop rubbing, and RED is the current OMC suspended, free of rubbing. Templates for today's measurements live here: /ligo/svbcommon/SusSVN/sus/trunk/OMCS/H1/OMC/SAGM1/Data 2016-08-05_1909_H1SUSOMC_WhiteNoise_L_0p2to50Hz.xml 2016-08-05_1909_H1SUSOMC_WhiteNoise_P_0p2to50Hz.xml 2016-08-05_1909_H1SUSOMC_WhiteNoise_R_0p2to50Hz.xml 2016-08-05_1909_H1SUSOMC_WhiteNoise_T_0p2to50Hz.xml 2016-08-05_1909_H1SUSOMC_WhiteNoise_V_0p2to50Hz.xml 2016-08-05_1909_H1SUSOMC_WhiteNoise_Y_0p2to50Hz.xml Notes: - Because I wanted yesterday's measurements in the template as a reference, and we were in rush, I did not head my own advice and use the templates from 28736. Just keep this in mind. At least, this time, I'd paid attention to the frequency resolution and made it the same 0.02 Hz BW across all DOF's templates, but I did not add in the low-frequency boost. Again, one can stil get the information needed out of these templates, but they could be better. - I only took a few averages. This is the luxury of having reference transfer functions and suspensions with very repeatable dynamics: we know immediately -- even after one average -- that if the resonance frequencies, zero frequencies, and their respective Qs line up with the reference then the suspension is healthy. More averages will just reduce the incoherent noise on the TF at all frequencies, and since these measurements are in air, we expect a good deal of incoherent noise at all frequencies anyway.
State of H1: OMC PD electroincs pulled for testing - issues are ongoing, OMC suspension issues are resolved
Activities: all times UTC
Dust:
Alignment:
J. Warner Jim has put gathered some data on the wind fence. Check out SEI aLOG 1050.
Also advertising a comment to Jim's log with the detchar tag: SEI log 1052
Chandra, Gerardo, Patrick Degassed the two nude ion gauges in corner station - PT170 & PT180 on BSC 7 & BCS 8. First degassed PT170 (700degC for 3 min. according to manual). We monitored the adjacent PT120 CC gauge, where pressure barely rose (8.48e-9 Torr to 8.52e-9 Torr). Pressures at the two ion gauges rose significantly and caused a verbal alarm at the operator console; pressures spiked to ~2e-7 Torr and then settled down to base pressures. Note: degassing draws ~ 70 W and thus caused PT120 pirani to flat line and CC to read bogus pressure reading during degassing. This happened during PT180 degas, but not with PT170. We also noticed that PT140A periodically flat lines - linked to PT110 (?).
System Check-In and issues discussed:
OMC:
All other systems / issues:
Visitors next week: Aug. 8-12
The OMC scan was performed last night after the chamber was covered. While the scan showed noimnal resonant features of the slightly misaligned cavity, I saw negative trips of the DCPD outputs (both) when the photocurrent is of the order of ~0.1mA.
The attched plot shows the refl QPD sum (CH1), scanned PZT Voltage (CH2), and the DCPD A/B outputs (CH3/4). The output of the DCPDs are usually positve, while they trip to negative occasionally when the cavity hit the major resonances. The positive voltage segments show nice resonant features when they are plotted in a logscale.
The refl QPD sum shows that the light was nicely absorbed by (transmitted through) the OMC. So it seems it is not optical.
I quickly checked the power supply condition of the field rack and confirmed that the related chassis are on.
We want to figure out this feature by the time we close the dorrs next week as the DCPD preamps are located in HAM6.
First thing that comes to mind would be to check the status and operation of the OMC piezo shutter for anything spurious... The dips are perfectly common to both diodes, so I would also look at the Split Whitening for: Bias to the PDs, DC power as supplied by the OMC Split Whitening Interface, and give special scrutiny to the most recent changes to the split whitening involving AC coupling of the OMC DCPD Chain for PI use.
Robert, Richard, Koji, Stefan
Bottom line: We think that this is simply the due to acousting vibrations modulating the OMC transmission in air - leading to ADC saturations. We thus decided to go ahead with the install.
Details;
We noticed that the drop to negative voltages is due to ADC saturations. When reducing the light to avoid saturations, we find that the RIN on the light is ~0.25 RMS. It is dominated by a peak around 1 kHz, and is coherent with the microphone.
Note that the OMC fringe width is only on the order of 1 nm, and its body mode is around 1kHz. It is not surprising that the thing vibrates more than that when in the horrendous acoustic environment of an open HAM chamber.
==============================
We verifyed with a light bulb that the electronics woks well up 0.3mA of current, which is higher than the previously observed problem. Case closed.
The HV power cable for the Y2-8 ion pump has been repaired. The damaged section was cut off, and the cable was spliced together. Cable was tested with the HI-POT tester to 5KV.
Connected cable to controller and turned ion pump on. Pump is pumping now, see attached for 3 hour trend data.
Thank you!!
Work done under WP #6047 and FRS ticket 5992 closed
The shot noise level from last night seems higher (worse) than the O1 level by 6%. Here is the spectrum:
You can see that the red trace (which is the one from the last night) is slightly higher than the (post-) O1 spectrum. The 6% increment was estimated by dividing the two spectra for frequencies above 1200 Hz and taking a median of it.
Evan H. suggested looking at the null and sum channels to see if the excess in shot noise is from an addition technical noise or not. The attached shows the spectrum of the null and sum channels at the same duration as the spectrum in the above entry.
From this plot, it is evident that the excess is not due to technical white noise.
I have checked the calibration of the DARM signal by comparing it against the Pcal excitation signals. I used the same lock stretch as the above entry. The height of the Pcal line at 331.9 Hz in the DARM spectrum was found be too high by 13% relative to the Pcal TR and RX PDs. See the attached. This means that we have overestimated the DARM signal at 331.9 Hz due to a calibration error. If we assume this is all due to an inaccurate optical gain, actual shot noise level should be smaller by the same factor of 13% that what we thought, corresponding to a ~7% smaller shot noise level than that in O1. We need to nail down whether this is an error in the optical gain or cavity pole in order to further evaluate the calibration error.
Note that the Pcal Y uses a fresh set of the calibration factors that was updated a month ago (27983). The ratio of RX PD over TX PD was found to be 1.002 at 331.9 Hz and this makes me think that the Pcal Y calibration is reliable.
Here I have attached plots of the optical gain during this lock as well a few locks randomly picked during the month of July. I used O1 model as reference (wasn't not quite sure whether there was new time zero reference after O1 with all kappas set to 1). The first plot showing kappa_C over a few locks during July show that kappa_C values were close to 1. However here we note that the gain in the inverse sensing function during July was set to 1.102e-6 compared to 8.834e-7 during O1 (the referene model has changed). At high frequencies, the relation between corrected h(t) and h(t) recorded in front-end is,
corrected h(t) ~ h(t) / kappa_C ~ inv_gain * DARM_ERR / kappa_C
So for same DARM_ERR, kappa_C of 1 during July 2016 corresponds to 0.8 * h(t) (= 8.834e-7 / 1.102e-6) as that of during O1. This assumes that there wasn't any change in the gain of the electronic chain on the OMC side. The second plot show trend of kappa_C during the lock Kiwamu was looking at. An interesting thing to note here that there was ~10% change in the optical gain during this lock. Kiwamu's plot correspond to time of the second peak we see in the plot (a coincidence!). The kappa_C value of 1.15 suggests that the measured h(t) in the above a-log would correspod to 0.70 ( = 8.834e-7/1.102e-6/1.15) times that of h(t) we would be measured during O1. Since the trend plot show that there were times in the same lock during which the kappa_C values were different, I tried to compare the power spectrum between those times. The third plot show that comparison. The mystery is that eventhough the ratio between the 331.9 Hz photon calibrator line and DELTAL_EXTERNAL line is ~10 % different between the times compared (and hence corresponding to ~10% different optical gain), the shot noise level looks same! We couldn't get the exact cavity pole frequencies because at this point I don't have the new LHO DARM model function, but the trend indicated that it didn't change during the lock. For completeness we also added the acutation strength variation during this time. The values are close to what we expect. Since 35.9 Hz ESD line we used during O1 wasn't available, for actuation strength comparison we used 35.3 Hz ESD line.
EDIT: We corrected the earlier estimate of high frequency h(t) level change.
J. Kissel, E. Goetz We've taken another DARMOLG TF and PCAL2 DARM transfer function in order to see if / how the SRC Detuning Optical (Anti-)Spring is changing from lock-stretch to lock-stretch. The bad (but not necessarily unexpected) news: it looks like the optical (anti-)spring frequency is moving around. One can tell by looking at the lowest frequency data points in the .pdf attachments. The model has a spring frequency of 9.381 Hz and both measurements are divided by it to form the residuals on the right column of subplots. Note also that Kiwamu and Craig's more sophisticated parametrization for the optical spring (which includes some Q between in the anti-spring poles, see LHO aLOG 28274) is not yet included. Jul 1 data points are ~ +15% discrepant in magnitude, where as Jul 09 data points are ~ +25%. More thoughts on this (what to do about it, how to track it, tests to try and control / reduce it) after the weekend. The pre-processed sensing function has been attached here, such that whomever gets to it first, can reproduce the fit using Craig's code from LHO aLOG 28274. Data sets live here: ${CalSVN}/trunk/Runs/PreER9/H1/Measurements/DARMOLGTFs/2016-07-09_H1_DARM_OLGTF_4to1200Hz_SRCTuned.xml ${CalSVN}/trunk/Runs/PreER9/H1/Measurements/PCAL/2016-07-09_H1_PCAL2DARMTF_4to1200Hz_SRCTuned.xml
C. Cahillane, K. Izumi See DCC T1600278 for an updated document on the sensing function detuning. I've taken the new ER9 sensing measurement from July 9th and plotted it next to the July 1st measurement for comparison, and done a fit on the July 9th measurement as well. Plot 1 shows July 9th and July 1st Sensing measurements at LHO. The detuning is visibly different for both measurements. Plot2 shows the July 9th fit. The fitting parameters are copied below. Plot 3 shows the fitting parameter cornerplot.July 9st Parameters ===================== Optical gain = 8.998599e+05 +/- 1.095765e+03 [cnts/m] Cavity pole = 3.206019e+02 +/- 8.580968e-01 [Hz] Time delay = 2.218270e+01 +/- 5.364895e-01 [usec] Spring frequency = 8.304667e+00 +/- 6.061604e-02 [Hz] Spring Inverse Q = 5.107011e-02 +/- 6.784018e-03The original July 1st parameter fits from aLOG 28274 are reprinted here for convenience:July 1st Parameters ===================== Optical gain = 9.124805e+05 +/- 8.152381e+02 [cnts/m] Cavity pole = 3.234361e+02 +/- 5.545748e-01 [Hz] Time delay = 5.460838e+00 +/- 3.475198e-01 [usec] Spring frequency = 9.975837e+00 +/- 5.477828e-02 [Hz] Spring Inverse Q = 1.369124e-01 +/- 3.522990e-03There is a difference of 1.67 Hz in the optical spring frequency. This represents a detuning phase difference of 0.316 degrees:July 1st Detuning Phase = 1.027 deg July 7th Detuning Phase = 0.711 deg
TF from DCPD sum to DARM IN1 at 2016-07-09 01:00:00 is 2.96e-7 ct/mA. Therefore, the peak optical gain (above the antispring, below the DARM pole) is 3.0 mA/pm for this lock stretch.
During O1, the peak optical gain was 3.2 mA/pm. However, this was for 95 kW of circulating power and 20 mA of dc readout photocurrent. For ER9, these numbers are instead 130 kW and 15 mA. Therefore, the naive expectation for the ER9 optical gain would be 3.2*sqrt(130/95)*sqrt(15/20) = 3.2 mA/pm, rather than the 3.0 mA/pm that was observed.
In terms of DARM shot noise, 3.2 mA/pm with 20 mA of dc photocurrent (the O1 situation) amounts to a shot noise of 2.5e-20 m/rtHz in the bucket. 3.0 mA/pm of with 15 mA of dc photocurrent (the ER9 situation) amounts to a shot noise of 2.3e-20 m/rtHz in the bucket; i.e., an improvement of less than 10 % over O1. This agrees roughly with Kiwamu's shot noise analysis.
We adjusted it on Friday in hopes to get a better OMC measurement. I thought that I had brought it back to its previous state but with how touchy that valvue is, I may have been slightly off.