Have remained locked. Took out of observing briefly to change the end X PCAL frequency. H1:LSC-POP_A_LF_OUTPUT has been slowly falling and is now around 15,750. There is a slight indication that it may be leveling out (see attached), so I'll give it some more time to see.
Per Rick's request I changed the end X PCAL frequency from 3501.3 to 4501.3. I also changed the sine amplitude from 35,000 to 40,000. I went out of observing from 12:00:23 - 12:11:29 UTC to do this. I accepted the SDF differences from doing this (see attached). Per Rick's schedule this should be left to run for 12 hours.
Ops Eve Summary: 00:00-08:00UTC, 16:00-23:59PT
State of H1: Observe, range around 79Mpc, POP_A_LF at 16000
Incoming Operator: Patrick
Shift Summary: Calibrations continue, and H1 was saved from a lock loss when POP_A_LF dropped to 14800, by adjusting the alignment of TMSY
Shift Details:
H1 in Observe when I took over
02:33:40UTC - H1 out of Observe for adjustments to TMSY alignment, POP_A_LF at 14800, see alog 24818
03:14:46UTC - H1 back to Observe, POP_A_LF at 16100
04:32:32UTC - H1 out of Observe to change the frequency of the PCAL line, see alog 24819
04:33:26UTC - H1 back in Observe
TITLE: 01/10 [OWL Shift]: 08:00-16:00 UTC (00:00-08:00 PDT), all times posted in UTC STATE Of H1: Observing @ ~ 79 MPc. OUTGOING OPERATOR: Cheryl QUICK SUMMARY: From the cameras: The lights are off in the LVEA, PSL enclosure and end Y. I can not tell if the lights are on or off at mid X, mid Y or end X. Winds are less than 10 mph. From pinging: CDS WAP is off at the LVEA, end X and end Y. CDS WAP is on at mid X and mid Y. Screenshots of the seismic bands and ISI blends are attached.
PCALX line moved from 3001.3 to 3501.3Hz at 04:32:32UTC (intention bit commissioning).
H1 back in Observe at 04:33:26UTC.
Screenshot of SDF attached.
This cal line frequency, 3501.3Hz, should run for 6 hours.
When this run is complete, the cal line should be changed to 4501.3Hz and then requires 12 hours of data.
See JeefK's alog for details: alog 24802
Summary:
TMSY pitch and yaw alignment tweaked to increase POP_A_LF, saving the lock.
Details:
After JeffK and I saw that POP_A_LF was dropping, I looked at the H1 alignment over the 27 hour lock, and focused of PR3, ETMX, and TMSX.
After looking at the trends, I proposed the idea to JeffK that we might be able to save the lock by aligning TMS, and he suggested TMSY, based on previous work by Kiwamu.
When I compared TMSX and TMSY alignments over the 27 hour lock, I found that TMSY alignment changed by -1urad in pitch and +0.5urad in yaw, when measured from the start of the decrease in POP_A_LF.
To compare, TMSX alignment changed 0.02urad in pitch and 0.05urad in yaw.
I started by raising POP_A_LF with TMSY yaw, and stepped the alignment slider by 0.05 at first and then increased that step size to 0.2, with a ramp time of 20 seconds, and made a step about every 4 minutes, which gave the ASC time to recover.
After some improvement in POP_A_LF, I switched to aligning TMSY pitch.
When I was done, POP_A_LF had increased from 14850 to 16100.
Plots attached:
Time history of this lock:
- Cheryl, JeffK, Kiwamu
ASC during the drop and recovery of POP_A_LF
plot 1 - MICH P and Y and POP_A_LF - MICH noise and a glitch in POP_A_LF, just right of center in the plot, starts the POP dive
plot 2 - PRC is unperterbed by all events
plot 3 - SRC moves away from zero as POP_A_LF drops
PRM, PR3, BS, ITMs, ETMs, and TMSs during the POP drop and recovery
J. Kissel, R. Savage, E. Goetz, D. Tuyenbayev We've replicated a study similar yesterday (LHO aLOG 24784) to what LLO has done in early December (see LLO aLOG 23184), in order to confirm/demonstrate the Delta L = Lx - Ly (i.e. no factors of 1/2) convention for the calibration. In addition, we confirm the accuracy of the relative calibration between PCALX and PCALY by comparing a true CARM and true DARM excitation. In summary, - When the PCALs are driven exactly 180 [deg] apart at equal amplitude (i.e. "true DARM," where amplitude is predicted by PCAL), the DARM displacement (i.e. Delta L = Lx - Ly, as measured by the IFO) is twice the amplitude. Convention confirmed! - Comparing the ratio of a true CARM (driven hard enough that some residual DARM is visible in DELTAL_EXTERNAL) and true DARM excitation (again in DELTAL_EXTERNAL), the ratio of CARM / DARM = 0.004, or 0.4%. This indicates that the relative calibration between the two PCALs better than 0.4%. Very good! Details %%%%%%%%%%%%%% ----------------------------------- Setting the amplitude of excitation ----------------------------------- In order to set up PCALX to drive at exactly the same amplitude as PCALY, we followed Shivaraj's procedure from LLO aLOG 24043, but not perfectly. In step 3 of his procedure, he determined the amplitude ratio between digitally requested counts of drive at PCALY vs PCALX end assuming the Optical Follower Servo (OFS) PD has been well-calibrated into force on the test mass (Newtons), such that the relative offset in the PCALY and PCALX servo represents the relative amplitude ratio of digitally requested drive to impose the same amount of force on the test mass. LHO's OFS PDs have not yet been into Newtons. Instead, we use each PCAL's main TX PDs, which *have* been well-calibrated into Newtons, and take the ratio of the TX PD (in Newtons) to OFS PD (in Volts), TX PD(X) N (X) TX PD(Y) N (Y) ------ = --- and -------- = --- . OFS PD V OFS PD V These transfer functions are frequency independent. Further, the OFS PD Volts are proportional the DAC counts [ct] of the oscillator such that TX PD (Y) OFS PD X Drive [ct] . ------ . ------ = Y Drive [ct] OFS PD TX PD (X) These transfer functions, taken at 36.7 [Hz] are shown in the right-most panels of the two attachments 2016-01-08_H1PCAL_TrueCARM_Drive_TXRX_TF.png and 2016-01-08_H1PCAL_TrueDARM_Drive_TXRX_TF.png. Conveniently, the ratio of transfer functions happens to be 2.00. So, we need to drive the X end exactly twice as hard as as the Y end. Once the amplitude is identical at both PCALs, the phase is tuned as described in Shivaraj's procedue, such that the PCALs are driving at the same phase to achieve pure CARM excitation. Remember, both the Sine and Cosine amplitudes must be ON and the same, for this whole phasing thing to work. For simplicity, we -- like Shivaraj -- just chose the same three frequencies that are on PCALY all the time, 36.7, 331.9, and 1083.7 [Hz] and at roughly their normal amplitudes (100, 1500, and 7500 [ct] respectively). The phase is tuned manually on the PCALX oscillator to 0.0 +/- 0.2 [deg]. This is demonstrated on the middle three panels (one for each line) in 2016-01-08_H1PCAL_TrueCARM_Drive_TXRX_TF.png. We then compare the transfer functions between both end station's estimate of thier respective test mass' displacement. This is shown in the left-most three panels of 2016-01-08_H1PCAL_TrueCARM_Drive_TXRX_TF.png. Recall that at the PCAL X-end, there are troubles with clipping in the RX PD (see, e.g., LHO aLOG 24774), so the PCALX TX PD is the best source of calibrated displacement at that end, and is what is used as the reference in the TFs. This compared against the PCALY's TX PD and RX PD, both of which are also well-calibrated into [m] of Y-end Test Mass displacement. As can be seen, the ratio of TX PDs is better that 0.002 (or 0.2%). Awesome. Though they only looked at it quickly, the PCAL team claimed to understand why PCALY's RX PD is 2% larger than both PCALY and PCALX's TX PDs, but I don't recall why. I'll ask them to comment on this log about it. Though above details of how the amplitude and phase of each excitation is matched may be tough to follow, the take-away is that we "dead reckon" the amplitude with each end stations PCALs calibration (which, admittedly are both based on LHO's working standard, derived from NIST's gold standard). We do not "cheat the result," for example, by driving in CARM, and minimizing the amplitude and phase by minimizing the line height in DARM. Finally, we flip the X-end PCAL excitation phase by 180 [deg], such that we're now driving pure DARM. A remeasurement of the relative phase between X and Y confirms the 180 +/- 0.2 [deg] difference, and the amplitude ratios between end stations's TX and RX PDs are identical (also awesome). Both the DARM and CARM configurations were measured at the normal amplitude, driving all three lines at once. While driving in DARM, the DELTAL_EXTERNAL displacement is exactly twice the predicted test mass displacement at all three frequencies These are shown in the attachments 2016-01-08_H1DARM_ASD_PCAL_TrueCARM_Drive.pdf and 2016-01-08_H1DARM_ASD_PCAL_TrueDARM_Drive.pdf. However, while driving all three lines simultaneously in CARM, we did not have enough actuation range on PCALX to see any residual DARM motion. As such, we repeated the test only driving the 331.9 [Hz] line, but much harder in both end stations (12e3 [ct] at Y, 24e3 [ct] at X). At this level of drive, the residual DARM motion during true CARM excitation was visible. Comparing that ratio of CARM to DARM, we find the residual, relative drive calibration between the end stations, (true CARM EXC) / (true DARM EXC) as measured by DARM is 0.0042683 at 331 [Hz]. A large fraction of this residual DARM is likely limited by the precision to which we've measured the ratio of requested actuator strengths (which we rounded to 2.00, instead of higher precision; upon later inspection with a cursor, the ratio is 2.0037675). However, the calibration has been demonstrated to better than the total uncertainty claimed by PCAL, 0.76%, so we need not spend the time to be more precise. The location of the measurement templates are restated here for convenience: /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/PostO1/H1/Measurements/PCAL/ 2016-01-08_H1DARM_ASD_PCAL_TrueCARM_Drive.xml 2016-01-08_H1DARM_ASD_PCAL_TrueDARM_Drive.xml 2016-01-08_H1PCAL_TrueCARM_Drive.xml 2016-01-08_H1PCAL_TrueDARM_Drive.xml and have been commited.
Investigation of ~1.8% discrepancy between TxPD and RxPD calibrated channels
During "H1 PCAL True DARM / CARM Excitations" study (see original alog above) we discovered that H1:CAL-PCALY_RX_PD_OUT_DQ / H1:CAL-PCALY_TX_PD_OUT_DQ transfer function deviates from 1 by about 1.8 % (see top-left subplot in the 2nd image attachment to original alog).
These are the channels calibrated in meters (except for zpk( [ 1, 1 ] : [ ], 1) whitening) of PCALY actuation and we expect them to give the same actuation level.
We found that a N / V calibration factors installed in H1CALEY foton file correspond to a calibration results from May 22, 2015 given in Pcal end-station calibration report (the first column in the report correspond to May 22).
Our recent PCALY end-station calibration measurement show that since May 22 these calibration factors have changed (see DCC T1500131-v2). Between May 22 and Aug 11 an overall optical efficiency of PCALY system increased from 0.977 to 0.989, which is probably the most significant contributor.
Our analysis show that currently installed PCALY TxPD calibration factor is off by 0.89% and RxPD factor is off by 0.78%.
Comparison with recent measurements of PCALY N / V calibration factors
Since RxPD and TxPD calibration discrepancy must be due to non frequency dependent gain factors (the V / W absolute power calibration of TxPD and RxPD) we will use a single line values of RxPD and TxPD readouts. We use TxPD and RxPD outputs in volts to compare how calibrated signals from these two photodetectors will differ from each other.
Amplitudes of voltages measured by RxPD and TxPD at 331.9 Hz (we have a Pcal calibration line at this frequency) at 12/01/2016 01:03:32 UTC are:
VTx = 0.75448 [ V ]
VRx = 1.09274 [ V ]
Currently installed calibration factors are based on following V / W absolute power on ETM calibrations that correspond to May 22 meausrements;
TxPD: ρTxe = 4.3572 [ V / W ]
RxPD: ρRxe = 6.1953 [ V / W ]
The amount of laser power incident on ETM calculated from currently installed calibration factors give values that are discrepant by 1.82 %:
calculated from TxPD: P = 0.17316 [ W ]
calculated from RxPD: P = 0.17638 [ W ]
If instead we use most recently measured calibration factors (from Dec 22, 2015, LHO alog 24398):
TxPD: ρTxe = 4.3187 * 10 [ V / W ]
RxPD: ρRxe = 6.2438 * 10 [ V / W ]
the amount of laser power incident on ETM will be consistent (discrepancy < 0.19 %):
calculated from TxPD: P = 0.17470 [ W ]
calculated from RxPD: P = 0.17501 [ W ]
Additional notes:
There's no indication of when exactly overall optical efficiency changes between May 22 and Aug 11, surfing through alog gave two occasions when some work was done on PCALY: on May 26 (LHO alog 18638) and on May 28 (LHO alog 18665).
There was no indication that an overall optical efficiency of PCALY changed after Aug 11.
Filter archive show that PCALY calibration factors in H1CALEY foton file did not change since Aug 08 2015 03:52:15 UTC.
PCALX line moved from 1001.3 to 3001.3Hz at 23:59:53UTC (intention bit commissioning).
H1 back in Observe at 00:02:15UTC.
Screenshot of SDF attached.
This means that the 1001.3 [Hz] line was on from 22:38:27 UTC to 23:59:53 UTC. No appreciable RF45 noise during this data stretch. I attach the usual screenshot demonstrating the amplitude and absense of any other features around the 3001.3 [Hz] line. We'll leave the 3 [kHz] line on for about 4 hours, as per Rick's schedule. Also for future reference, I've checked 3.5 and 4.5 [kHz], and they're also featureless, so no need to worry about injecting their later (see second and third attachment). The next scheduled change (if the IFO is kind enough to stay locked) is at ~8:30 PDT (~4:30 UTC), and I've instructed Cheryl to make the change.
This configuration, cal ine at 3501.3Hz, should run for 6 hours.
After the data at this frquency is collected, the last configuration is the cal line at 4501.3 Hz, which needs to run for 12 hours.
H1 in Observe
JeffK running calibration lines at various frequencies
TITLE: Jan 9 DAY Shift 16:00-00:00UTC (08:00-14:00 PDT), all times posted in UTC
STATE Of H1: Obser ing
SUPPORT: Jeff K, Rick S.
INCOMING OPERATOR: Corey
SHIFT SUMMARY: IFO locked fo the duration of the shift. Environmentally, we were more or less sound. µSeism extended past the 90th% line to .8Mm/s occassionally. RF45 noise dominated the lock time. PCal frequencies were changes and RF45 offsets were “bumped” in an effort to terminate glitching. For a little over 2 hours our range seems to be back, glitch free, at ≈78Mpc. Jeff is still on site doing some work. Handing off to Cheryl.
ACTIVITY LOG:
18:26 out of Observing to turnoff P-Cal line(s) at EX to observe 45Mhz noise in DARM for subsequent effects
18:30 Jeff changing the P-Cal frequncy to 2001.3K 35Kamp.
18:37 changed RF45 offset by .4(neg). This changed the noise characteristics of DARM (as expected). We will wait and see what effects this will have.
18:43 RF45 offset returned to nominal value.
18:44 PCal X value changes accepted in SDF.
18:44 Intention Bit set back to Undisturbed
18:54 MICH and PRCL live have settled back down to their respective references.
18:58 well, that was short lived……. the noise seems to be sharper in PRCL between 15 and 25 Hz. The correlating noise in MICH is much broader. It seems to be glitching now rather than constant as before. THe only other ‘heads up’ noise correlation I can see is the unsteadiness of RF45 Coherence between 1 and 10Hz.
19:10 the noise is still glitchy but diminished in nature.Range back to 78Mpc range, save for one large ETMY saturation.
CONT…. aaannnd we’re back to crud.
21:19 Kyle heading down to MY to do the daily overfill of CP
21:49 15 minutes now; Range is back and glitchiness is down. It appears to be random.
21:51 Intention bit set back to Observe
22:05 Kyle back from MY
22:42 Intention bit set back to Observe after PCal freq was moved to 1KHz
23:06 Tour group into the Control Room
Wanting to eliminate CP3 as the source of the leak that appeared on 12/27/2015, I applied still more Vacseal to the area of the previously stopped leak which had been stopped months ago (see various aLOG entries) the difference is that this time a stayed away from the "cut/gouge" area that I had attributed as being the site of the leak but, instead, applied it only in the area where the spool roll seam weld interfaced with the stiffening ring stitch weld (also in the gap beneath the stiffener) -> WAHLAH! see attached pressure response
Plot of PT210
J. Kissel Continuing on schedule(-ish, see LHO aLOG 24802), we've moved the PCALX line from 1501.3 [Hz] to 1001.3 [Hz]. I'll collect all of these times, amplitudes, and frequencies once we're done, but I'll post a mini-alog every time we change it to explain the drops to commissioning mode. Through the magical powers of awesome, Rick somehow picked a frequency that perfectly avoided the an-harmonic, 1st harmonic of the QUAD's violin modes. Nice work! This change means the 1.5 [kHz] line was in place from 21:02:11 UTC to 22:38:27 UTC (padded by a 30 second ramp time on either side) We've now resumed observation intent, and it appears we might be done with RF45 noise for the day (crossing fingers).
~1315 - 1355 hrs. local LN2 observed at GN2 exhaust after only 4 minutes -> Considering the exact procedure was followed as was during the previous manual over fill (2 days ago) and that then it had taken 29 1/2 minutes for LN2 to be observed at the GN2 exhaust, this demonstrates that opening the LLCV bypass valve 1/2 turn ccw is insufficient -> As such, from now on we should use 1+ full turn ccw (exhaust check valve bypass assumed to always be opened before LLCV bypass is opened) -> Today's data supports the choice of using 15% open for the LLCV, which was based upon the historical average needed to maintain the proper pump level- meaning that the pump level decays only slowly over the course of days if reliant only on the filling that occurs via the 15% open LLCV Next scheduled manual over fill of CP3 will be Monday, Jan. 11th before 4:00 pm
J. Kissel Continuing on schedule(-ish, see LHO aLOG 24802), we've moved the PCALX line from 2001.3 [Hz] to 1501.3 [Hz]. I'll collect all of these times, amplitudes, and frequencies once we're done, but I'll post a mini-alog every time we change it to explain the drops to commissioning mode. This change means the 2 [kHz] line was in place from 18:29:45 UTC to 21:02:11 UTC (padded by a 30 second ramp time.) Also note there has been terrible sub-1[kHz] noise through out that ~1.5 hour period, but it has remained at low frequency, so I don't suspect that it will impact the estimate of the transfer function at these super-1[kHz] excitation frequencies.
