ALS COMM was even more noisy tonight than last night, and now it is not stable at all. This could be because of the increased seismic noise, or some other reason.
The X arm guardian is working (there is no WFS control yet), I am leaving the arm locked with WFS on and no slow feedback to the ETM. It also seems like the COMM guardian is also working, but I'll need to test it sometime when the ALS COMM loop itself is working.
I started some excitations for Keita, screenshot attached.
written by Kiwamu I stopped the excitation at 6:10 AM local.
Seems like the green transmission is not on the same spot as last night, maybe the TMS moved.
We had some difficulty engaging the WFS because the whitening gains were all reset to 0dB and white/awhite were mismatched and Lock path for the ITM was off.
After these were fixed the WFS started working, today the transmission power doesn't go to 850 cts, it stays 750-800.
I'm leaving WFS on for the night, and when Sheila is done she'll start the excitation for the ITM and ETM.
[Keita Jeff Arnaud]
Several measurements were taken on ETMX for top mass length to test mass pitch decoupling as well as for the pum tf inversion (cf Keita's aLog), and are summarized in the attached plot.
Page 1 and 2 : Top mass to test mass TF compared with the "fiber" model, length to pitch (p1), pitch to pitch (p2)
Page 3 and 4 : Pum to test mass TF compared with the "fiber" model, pitch to pitch (p3), yaw to yaw (p4)
Points with an uncertainty lower than 10% have been removed from the transfer functions.The data for the length to pitch (2nd page) is a combination of a swept sine measurement, and a white noise measurement.
Measurements slightly differs from the model, especially for the top mass pitch to test mass pitch (p2), thas has its first pitch mode shifted down. It would be interesting to plot those measurements against the correction that is beeing made to the current model.
The data for the pum pitch to test mass pitch doesn't have any coherence below 0.6Hz, and those points have been deleted because of a low uncertainty, so it is currently hard to do a fit with the few points we have. I would like to try few more times tomorrow to do this TF again.
In the meanwhile, I will use Keita's happyVecfit function to do the top mass fitting, and calculate -L2P/P2P
Yuta, Kiwamu, Daniel, Sheila, Dave, Evan, Stefan After we had the PRMI drive properly balanced (alog 10559), we fine-tuned the 3f-locking again: - Added a 60Hz (triggered) notch in MICH to reduce the RMS drive to BS and PRM. - AC-coupled the PR2 drive at 2Hz - it was saturating at DC. - Reset the REFLAIR_B_RF27: two stages of whitening, 27dB of whitening gain, digital gain of 1. - With that the 3f lock by itself was fairly stable - lock duration was limited only by user impatience. Next we brought up the green arm - for some reason ETMX-HEPI seems to have moved - we had to re-align both TMS and ETMX suspensions. - After manually tweaking the green arm alignment the COMM_handoff worked fine. - We parked the about 12kHz avaiy from resonance and locked PRMI. - PRMI locking im this configuration seemed more difficult. Attached is a plot of the control outputs once locked - we were again frequently saturating the BS and PRM. - Ignoring all that we switched to 3f-locking, and brought the red on resonance - it stayed locked.... .... ... but we don't understand the red build-up. It's wasn't any higher than single shot.... something aint right quite yet, but I need a beer. EDIT (after the beer): The build-up actually makes perfect sense: The reflectivity sign flip of the ITMX due to arm resonance means the Michelson is exactly anti-resonant, redirecting all the light to the dark port. effectively no light gets back to the PRM. Hence the incident power on ITMX is equal to the (PRM misaligned) single shot power. We also had rockIFO playing the PRCL signal while we put the arm on resonance, and couldn’t hear any rumbling. To do next: measure length sensing matrix in PRMI and in state 3 (which we should lock on red using the 1f-signal).
The attached plot shows time traces for the combined 3f and x-arm cavity lock. The top left pad shows the frequency dialed into the COMM VCO. It is the calibrated green laser frequency in kHz. The red beam is resonant when the green frequency was near -37 kHz and was parked 10 kHz above. This is a 5 kHz offset for the red laser frequency. We verified that this is the direction away from the higher order transverse modes.
The bottom left plot shows the green transmitted power as measured in the corner station. The x-arm was locked to the green ALS laser most of the time.
The bottom right plot shows the POP_B demodulated signal for 2f. When high, it is an indication that the 1f-sideband is resonant in the PRMI. The lock was achieved with the 1f-signals and then switched over to the 3f-signals.
The top right plot shows the red transmitted power in the x-end station. With the green locked, it will typically dither around resonance, when the COMM PLL frequency is set to the red resonance. The first region of red arm cavity locking is with the PR mirror misaligned. During the 3f-lock the arm cavity was first parked away from resonance to facilitate locking, and then brought near resonance. The PRMI stayed locked using the 3f-signals during the entire transition! The transmitted power was about equal to the power when the PR mirror was misaligned. With the PRMI locked on sidebands and only one arm locked, there is a 180º phase shift for the carrier in the Michelson. Hence, all the carrier power is guided towards the AS port.
I've been updating the MEDM screens to be sure that all the BSC-ISIs currently used are consistent with eachothers.
Right now, BS, ITMX, ITMY & ETMX have:
- the Start, Tstart and Tcrappy blends installed respectively in blend banks 1, 2 and 3
- a working controller (lvl3 on ITMX & ITMY, lvl2 on BS & ETMX)
- tilt decoupling on ST1 for all them, except BS.
Putting on the controllers with Tcrappy blends on all DOFs for ST1 and ST2 provide a robust configuration with good performance on all these ISIs.
In other words, everything looks good into the seismic world!
There is of course some room for improvement. We're currently in an aggressive configuration, sensitive to varation of the input motion. Adding tilt decoupling on stage2 will help pushing the performance. I'll work on that as soon as I have access to a platform.
One more pair of WHAM4 HEPI Actuators attached--NE(corner 4.) Sure and steady's the word.
The BIO chassis that controls the ISC Whitening was shifted over from H2 without having its IP addresses shifted over, so it couldn't communicate with the rest of the slow controls system. Since we like being able to use the slow controls to control things, I changed the IP addresses from the H2 10.80.xx.xx convention to the H1 10.40.xx.xx convention.
Greg, Mitchell, Greg and I were able to attach the lift hook receiver onto the first 3IFO BSC. This is in preparation to transplant the ISI into it's long term storage home.
Betsy & Jeff The glass optic was installed in H1-SR3. There were no problems or issues during the installation.
It was realized yesterday that TMS still had some "head-banger" hardware left from the cartridge install. This was removed today, which necessitated a rebalancing of St2. While working on this, I also looked at a CPS that was acting like it was broken. I poked around the cable for the CPS and it fixed itself, so I'm hoping it was just a loose connection. I'll look at this again tomorrow, but at this point it looks like the in-air cable MAY be bad. For now, I will be running TF's tonight. from OPSWS5.
J. Kissel Following the same procedure outlined in LHO aLOGs 9453 and 9079, and similar to results from LHO aLOG 10493, with virtually the same demod parameters (only the amplitude was decreased from 1.25e5 to 1.2e5 [ct]) I balanced the coils on the UIM stage of H1 SUS ETMX. The final balanced gains in the L1_COILOUTF bank are H1 SUS ETMX L1 Channel Balanced COILOUTF Gain UL -0.966 LL +1.004 UR +0.996 LR -1.034 The precision is still within +/- 0.5%; the ground motion while tuning the gains was not much better than yesterday, even though I tried several configurations of the ISI. See details of balanced vs. unbalanced signal SNR in comments below. This balancing has reduced the L3 P and Y caused by a L1 pringle excitation at 4 [Hz] by DOF Reduction Factor @ 4.0 [Hz] P > 11.9 (peak below the noise, and totally incoherent) Y > 57.6 (peak below the noise, and totally incoherent) The attachment shows the result from which these values were obtained, comparing the optical lever ASD at 4 [Hz] driven from L2 at the same amplitude for both balanced and unbalanced configurations. Important note -- I found and was foiled by finding the L1_COILOUTF_UR gain sign set to Found Expected L1 UL - - L1 LL + + L1 UR - + L1 LR - - where expected comes from T1200015, and E1000617. Note these signs are exactly opposite from the PUM, because the UIM uses BOSEMs and the PUM uses AOSEMs, whose coils are wound in a different direction. Thus, the same magnet polarities are compensated by opposite signs. I did a cursory check on the other three QUADs, and the same mistake is there too. This might just be a systematic, copy and paste error that we've propagated every where, and didn't notice until now because we haven't needed to use the UIM.
Measurement Details ------------------- Coil Driver Configuration: State = 1 I was able to get plenty of SNR leaving the coil driver in the state I found it. Demodulator filters used: SIG band pass: BP4.0Hz = butter("BandPass",2,3.5,4.0) DEMOD I & Q low-pass: CLP50mHz = cheby1("LowPass",2,3,0.05) Demodulator Drive Parameters Freq [Hz] Amp [ct] Sin [ct] Cos [ct] 4.0 120000 10000 10000 4.0 120000 10000 10000 SEI Configuration: HPI: Level 1 Isolation, "Pos" position sensor only blend filters ST1: Level 2 Isolation, "TCrappy" blend filters (in all DOFs) ST2: Level 2 Isolation, "TCrappy" blend filters (in all DOFs) Note -- Sebastien had informed us that the Level 3 isolation filters and ST1-ST2 Sensor Correction are unstable for the time being, so I stuck with the above configuration. Resulting Demod Phases: Measured using a 200 second average (shorter than yesterday) of the demodulated signals, i.e. tdsavg 200 H1:SUS-ETMX_LKIN_P_DEMOD_I_OUT H1:SUS-ETMX_LKIN_P_DEMOD_Q_OUT H1:SUS-ETMX_LKIN_Y_DEMOD_I_OUT H1:SUS-ETMX_LKIN_Y_DEMOD_Q_OUT H1 ETMX L2 Demod Phase [deg] Unbalanced Value [ct] Balanced Value [ct] P 79 I +1.933 pm ~0.75 -0.103 pm ~0.75 Q -0.072 pm ~0.75 -0.060 pm ~0.75 Y 78 I +5.181 pm ~0.75 -0.012 pm ~0.75 Q 0.148 pm ~0.75 0.078 pm ~0.75 note that the quote pm values are the by-eye, peak-to-peak amplitude of the demodulated signal which oscillates with a ~25 sec period. I qoute values to a much higher precision because I'm averaging over 200 seconds. To perturb the PIT or YAW balancing by 1%: /ligo/svncommon/SusSVN/sus/trunk/Common/PythonTools/perturbcoilbalance_fourosem.py H1 ETMX L2 [PIT/YAW] 0.01 Exact balanced values: Measured using a simple command line caget, i.e. caget H1:SUS-ETMX_L2_COILOUTF_UL_GAIN H1:SUS-ETMX_L2_COILOUTF_LL_GAIN H1:SUS-ETMX_L2_COILOUTF_UR_GAIN H1:SUS-ETMX_L2_COILOUTF_LR_GAIN H1 ETMX L1 Coil COILOUTF Gain UL -0.965868 LL +1.003680 UR +0.995864 LR -1.003386 Of course, these values are set at arbitrary precession, they're rounded to the values quoted in the main entry (a) because the measurement uncertainty is no better than 0.5%, and (b) the MEDM screen does not display out to higher precession, so further precision would not be visible.
Kiwamu, Yuta, Stefan We finished installing the inverse plant filters for PRM and BS. Compared to yesterday, we lowered the Q of the <1Hz to 5. This effectively adds some boost at the main pendulum resonance, bud hardly affects the cancellation above 2Hz. Locking with the new filters required some reshuffling of limiters and boost filters. See snapchot. We then added PRM feed-back to the MICH loop, driving it at 99 Hz and cancelling its signature in REFL_45_I. We rescaled the BS gain (divided it by 16) to make the MICH output matrix the correct one from theory (1,-0.5). An updated all-included-snapshot is attached. New OLG measurements are also included. Attached are a screen shot of all relevant MEDM screens, as well as PRCL and MICH OLG measurements. The LSC Guardian was updated to reflect these changes. Inverse plant filters: BS: zpk([0.0382738+i*1.57475;0.0382738-i*1.57475;0.0647002+i*1.13667;0.0647002-i*1.13667; 0.0420175+i*0.418069;0.0420175-i*0.418069],[0.0163157+i*1.54679;0.0163157-i*1.54679],1,"n") zpk([],[0.5+i*0.866025;0.5-i*0.866025],1,"n")zpk([],[1250+i*4841.23;1250-i*4841.23],1,"n") PRM: zpk([0.130304+i*2.84281;0.130304-i*2.84281;0.092145+i*1.61388;0.092145-i*1.61388; 0.0683932+i*0.680504;0.0683932-i*0.680504;37.2961], [0.139329+i*2.35023;0.139329-i*2.35023;9.53503+i*13.2475;9.53503-i*13.2475;3.58083; 0.5+i*0.866025;0.5-i*0.866025],1,"n") MICH roll-off filter (to avoid BS saturation): zpk([-0+i*120;-0-i*120],[25+i*43.3013;25-i*43.3013;50+i*86.6025;50-i*86.6025;400],1,"n")
Exited LVEA 1630pst.
Jeff and Betsy will swap the glass optic into the SR3 suspension when Jeff gets back from end Y.
Betsy out of LVEA.
Work is done.