1235 - 1255 hrs. local -> To and from Y-mid Opened exhaust check valve bypass valve, opened LLCV bypass valve 1/2 turn -> LN2 at exhaust in 1 minute -> Restored valve configuration to as found. Next CP3 overfill to be Monday, July 18th.
While Sheila was exploring the input alignment my moving PR3 and letting the initial alignment follow the beam, I recorded the required drive vector for PR3, PR2 and IM4, in both pitch and yaw.
This combination of alignment sliders keeps the spot positions on PR3 and IM4 the same, but moves the spot on PR2 (and therefore somewhat on PRM).
The matrix to convert PR3 alignment slider counts into corresponding alignment slider counts of PR2 and IM4 is
pitPR3toPR2=-9.2;
yawPR3toPR2=+9.2;
pitPR3toIM4=56;
yawPR3toIM4=11;
The attached script continuosly reads PR3 slider values and updates PR2 and IM4 slider values, walking the beam on PR2 only.
The script still doesn't have the PRM matrix elements - we will have to addd this in DRMI or full lock.
=========================================================================
Also updated the script to include PRM. This was done in the PRM align state.
pitPR3toPR2=-9.2;
yawPR3toPR2=+9.2;
pitPR3toIM4=56;
yawPR3toIM4=11;
pitPR3toPRM=1.5;
yawPR3toPRM=2.2;
Note that this should work in-lock once the PRC1 pointing loop is off (despite the fact that PR3 is WFS controlled).
Last night I took a bunch of pictures using PCal EX and EY camera to check the large angle scattering at 20W.
This morning, while I was taking pictures while IFO was unlocked to capture images without 1064 light, EX worked fine but EY stopped working after 10 or 11 pictures. Camera control software stopped responding, I cannot trigger shutter, see the live view, see the camera settings, really nothing.
Rebooting the camera, the camera interface and h1pcaly multiple times didn't help.
The other day when verifying rack layout Filiberto felt like something was wrong. Upon further reflection he came to me this morning thinking he saw the HV shutter drive off. No readbacks on the lcd screen. I went out on the floor this morning and sure enough the +-15VDC switch was in the off position. I turned it on and knocked the IFO out of lock. At this time I am not sure how long it has been off (will run Dataviewer on a channel.) or why but will investigate. When we work on this chamber for vacuum reasons we switch off the High Voltage supply so no one should have shut this unit off. Investigation will continue to see how long we have been running in this state.
Further investigation looks like the unit has been off since early April. The HV Enable was also in the off position and the Trigger PD was disconnected. Someone really wanted this off.
So the ITMX did not break lock on this smaller EQ. Attached are the ISI T240 local channels--a couple hours around the EQ, for both ITMY & X. Differences in the responses at these inloop T240s for these minute trends is uninformative so don't bother looking. It is cool that the onboard vertical T240s shows the EQ shape better than the ground STS; related to the sensor correction? The two ITMs respond ~the same in the regard. When I compare the ITMs during the larger EQ Wednesday when the ITMX tripped, the response appears similar before the trip in the local basis from this vantage point. Again, I attach the trends I've looked at but I think zooming in, the frequency space and cartesian basis is next.
The old geophysicist in me believes the trip is occurring with the earlier arriving s-wave, not the large amplitude surface waves.
Subsystems report all is working as expected, at this time. Commissioning will be focusing on: - Understanding the recycling gain loss - Lowering noise in the system - Improving the reliability of the IFO
As Jenne wrote, it does look like POP90 should be a god signal to use for an SRM dither lock, but somehow POP90 is not actually going to the ADS input matrix despite what the medm screens say, and what my old alog says. I wonder if I frogot to check something into the svn after doing the model restart and someone reverted to the old version for some reason?
Anyway, it seems worth getting the model with OPO90 in ADS back since a dither using POP90 might be an easy way to get SRM alignment under control now that the PD is no longer saturated. Tonight it looked like the bounce mode rings up when SRM is misaligned, which makes sense if the SRM alignment causes SRC detuning and changes the DARM gain, as people have suggested.
Sheila, Kiwamu,
Tonight, we continued working on searching for the missing recycling gain. We opened up PRC1 (which uses POP_A_QPD and feeds it back to PRM) and moved PRM manually to study the effect of the spot position on PR2. This seems to work well and we were able to get back to a high recycling gain of about 34 when PSL was 25 W. The attached shows the time series of us moving PRM in pitch toward the negative side by about 90 urad. The first drop in the power recycling gain is caused by the power-up from 2 to 25 W. We had to move SRM by hand in order not to loose too much sideband build up in SRC. Our alignment action seems to pull the common soft (or something similar) to one direction in which all the test masses goes to negative pitch values (see the second attached). Therefore, as we were in O1, at this power level, the recycling gain can be still found under some alignment land. We did not try this test at a higher PSL power.
Also, we have noticed at some point that some segments of the POP QPDs already used up the half of the ADC range at 25 W. So we decreased the whitening gains to 0 dB (which were 6 dB before). Sheila changed the guardian accordingly so that it maintains the 0 dB gain as a nominal setting. Also when we had a good recycling gain with this alignment, the beams on both POP QPDs were almost falling off of the photodiodes
Another interesting point to note:
In a different lock stretch with a 25 W PSL, I was allso able to recover the recycling gain back to 32 or so by moving the soft offsets mostly in yaw. As reporeted above, it was surprising to me that for some reason moving PRM in pitch also regained the recycling gain.
Weigang Liu tried folding DARM and selected magnetometer channels during the three days of ER9 (UTC days) with results posted here: https://ldas-jobs.ligo.caltech.edu/~weigang.liu/advancedLIGO_3/ER9/files/ER9_magnetometers_and_strain_list.html, where DARM folding focused on the 10-50 Hz band, and the magnetometer folding was low-pass below 50 Hz, choices motivated by O1 experience. As expected, the magnetometer channels still exhibit a variety of pathologies with 2-second periodicity, but the elevated low-frequency noise in DARM make it hard to see this 2-second contamination in that channel with this method. Attached are a few sample magnetometer fold summaries over the three days.
Quick Summary: We stay at relatively low power because ITM PI damping modules are not ready.
(All time in UTC)
23:45 SUSPI model restart
00:45 DAQ restart
1:47 Kiwamu to the floor aligning REFL camera
2:35 Kiwamu out
2:45 Kiwamu request for higher power. Going to 30W.
Today turned into an impromptu maintenance day
DAQ
Maddie, Jim, Dave:
h1fw0 work, please see previous alog. DAQ broadcaster reconfigured to add H1:SUS-ETMY_LKIN_P_LO_DQ channel to DMT for Maddie. DAQ restarted to support all of today's model changes
OMCPI Beacon Alignmnet [WP5991]
Shiela:
h1omcpi mode changed, required DAQ restart (slow chans change only).
SUS-PI DAMP quad PD data transfer to ITM-PI [WP5997]
Matt, Carl, Ross, Dave:
A new RFM channel was created on each arm to transfer the mux'ed ETM QUAD-PD I&Q phase to the corner station. The 8 channels are sent from the SUSETM-PI model via Dolphin to the PEM model. This model muxes the 8 channels to one new RFM, which is sent to the corner. The h1omc model receives the RFM channel, demuxes it to 8 channels and sends these (via Dolphin) to the h1susitmpi model. The h1susitmpi model receives 16 new channels, 8 per arm.
The following models were restarted: h1susetmxpi, h1pemex, h1susetmypi, h1pemey, h1omc, h1susitmpi. The DAQ was restarted.
h1susetm[x,y]pi has some fast channel renames:
+: fast channel H1:SUS-ETMX_PI_DOWNCONV1_DEMOD_I_OUT_DQ added to the DAQ
+: fast channel H1:SUS-ETMX_PI_DOWNCONV1_DEMOD_Q_OUT_DQ added to the DAQ
+: fast channel H1:SUS-ETMX_PI_DOWNCONV2_DEMOD_I_OUT_DQ added to the DAQ
+: fast channel H1:SUS-ETMX_PI_DOWNCONV2_DEMOD_Q_OUT_DQ added to the DAQ
+: fast channel H1:SUS-ETMX_PI_DOWNCONV3_DEMOD_I_OUT_DQ added to the DAQ
+: fast channel H1:SUS-ETMX_PI_DOWNCONV3_DEMOD_Q_OUT_DQ added to the DAQ
+: fast channel H1:SUS-ETMX_PI_DOWNCONV4_DEMOD_I_OUT_DQ added to the DAQ
+: fast channel H1:SUS-ETMX_PI_DOWNCONV4_DEMOD_Q_OUT_DQ added to the DAQ
-: fast channel H1:SUS-ETMX_PI_DOWNCONV_DOWNCONV1_DEMOD_I_OUT_DQ removed from DAQ
-: fast channel H1:SUS-ETMX_PI_DOWNCONV_DOWNCONV1_DEMOD_Q_OUT_DQ removed from DAQ
-: fast channel H1:SUS-ETMX_PI_DOWNCONV_DOWNCONV2_DEMOD_I_OUT_DQ removed from DAQ
-: fast channel H1:SUS-ETMX_PI_DOWNCONV_DOWNCONV2_DEMOD_Q_OUT_DQ removed from DAQ
-: fast channel H1:SUS-ETMX_PI_DOWNCONV_DOWNCONV3_DEMOD_I_OUT_DQ removed from DAQ
-: fast channel H1:SUS-ETMX_PI_DOWNCONV_DOWNCONV3_DEMOD_Q_OUT_DQ removed from DAQ
-: fast channel H1:SUS-ETMX_PI_DOWNCONV_DOWNCONV4_DEMOD_I_OUT_DQ removed from DAQ
-: fast channel H1:SUS-ETMX_PI_DOWNCONV_DOWNCONV4_DEMOD_Q_OUT_DQ removed from DAQ
which may cause some issues replaying old data.
After the addition of the RFM channel, so far we have not seen any IPC errors.
PEM changes [WP5998]
Robert, Dave:
PEMCS was changed to read three unused ADC channels into the model. Due to broken ADC channels, on the fifth ADC the last two channels (30 and 31) were upgraded to 16kHz and added to the science frame. The broken channels were removed from the frame, resulting in a nett reduction of science data.
At EX and EY additional ADC channels were ingested into pem (was ADC0 chans 8-13, now chans 8-23). At EY, four of these new channels were added to the DAQ science frame at 2kHz.
Jim, Dave:
Overnight h1fw0 has been stable while writing only science frames. Today we reconfigured it to also write second and minute trend frame files (so it is only not writing commissioning frames). It crashed on the next 10 minute point. We replaced the fiber optics cable connecting h1fw0 with the Fujitsu switch, it still crashed. Finally we power cycled h1ldasgw0 (solaris machine) and it fw0 has been stable since 12:20 PDT.
Title: 07/14/2016, Day Shift 15:00 – 23:00 (08:00 – 16:00) All times in UTC (PT)
State of H1: IFO unlocked. Relocking not successful. Need to do initial alignment.
Commissioning:
Outgoing Operator: None
Activity Log: All Times in UTC (PT)
15:00 (08:00) Start of shift
15:23 (08:23) Bubba – Driving in desert between the arms "to look at stuff"
15:51 (08:51) Bubba – Back from drive in desert
16:00 (09:00) Robert – End-Y electronics bay
16:05 (09:05) Kyle – Going to Mid-Y and Mid-X for vacuum work
16:40 (09:40) PSL tripped – Called Peter
17:10 (10:10) Peter & Ed – In PSL to diagnose oscillator & check alignment (WP #5996)
18:06 (11:06) Dave – Restart FR0
18:07 (11:07) Dan – Going into the CER to make PEM measurements
18:30 (11:30) Filiberto & Dan – Going into the CER
19:15 (12:15) Robert & Filiberto – Out of the CER
20:00 (13:00) Robert – Going to End-Y
20:15 (13:15) Robert – Out of End-Y
21:43 (13:43) Robert – Going to End-Y
23:00 (16:00) Turn over to Nutsinee
End of Shift Summary:
Title: 07/14/2016, Day Shift 15:00 – 23:00 (08:00 – 16:00) All times in UTC (PT)
Support: Peter, Sheila, Ed,
Incoming Operator: Nutsinee
Shift Detail Summary: Started locking. Problem getting past DRMI_1F. Run initial alignment.
Having trouble getting through initial alignment SRC_ALIGN step.
Restart initial alignment. While doing the INPUT_ALIGN step, laser tripped off.
Peter reset the laser, checked and adjusted PMC alignment.
Now during INPUT_ALIGN stops at LOCKING_XARM_IR. While here, the ICM will break lock and the Guardian restarts the INPUT_ALIGN. Sheila fixed the INPUT_ALIGN problem. Completed INPUT_ALIGN of the initial alignment.
Commissioners working on the locking sequence.
Sheila, Matt, Kiwamu, Carl, Stefan
Earlier today we tried heating TCS CO2 X-arm with 2Watt (0 Watt into Y), and all we saw was a futher drop in recycling gain.
Tonight (07:29 UTC) we tried the opposite TCS: TCS CO2 Y-arm with 1Watt (0 Watt into X). (Half of what we put into x, because we broke lock on the first try.)
Result: Absolutely nothing - all recycling gains remained the same or further dropped.
Conclusion: TCS CO2 cannot get any recycling gain back.
============================================
Log:
UTC 20160713 23:08:09 all TCS CO2 completely off
no effect on any sidebands
UTC 20160713 23:16:22 TCS CO2 X to 1 W
UTC 20160713 23:21:21 TCS CO2 X to 2 W
recycling gains drop, lock loss
UTC 20160714 07:29:00 TCS CO2 X to 2 W, TCS CO2 Y to 0 W
recycling gains drop, lock loss
This lock was brocken by ITMY bounce mode. It looks like we may need to double check all the bounce mode phases to see that they are unchanged at 40 Watts.
There were several problems with PI damping this evening, the recovery of the safe.snap after ETMY's failures today resulted in H1:SUS-ETMY_BIO_L3_PI_ULLL_SW selecting the UL quadrant rather than the LL quadrant resulting in no ETMY damping for the entire evening.
In the 50W lock an ITMY mode (purple trace -80min) reached saturation level, the resulting unstable bounce mode at lock loss may have been associated. The mode was still at elevated amplitudes at the beginning of the next lock but was damped. The damping phase needs careful monitorring a changing sign regularly. I suggest moving to QPD error signals. Finally at -15min I set H1:SUS-ETMY_BIO_L3_PI_ULLL_SW to LL and was able to damp ETMY modes again.
For the bounce mode phasing: is the DARM gain now high enough that the cloop is close to +1 at the bounce mode frequencies?
I've put together some plots of the common and differential lenses as measured by the Hartmann Sensors.
The only change I made to the Hartmann sensor data is to scale the ITMY spherical power (H1:TCS-ITMY_HWS_PROBE_SPHERICAL_POWER) by (17.5/7.5)^2. This is because the magnification has defaulted to 17.5x on HWSY instead of 7.5x which is the actual level. Additionally, I reset both HWS measurements to zero at t=0.
The first plot (LHO_diff_CO2.pdf) shows the HWSY and HWSX measurements as well as common ([X+Y]/2) and differential (Y-X) lenses along with corresponding CO2 laser powers.
The second plot (gain_vs_time.pdf) shows an approximation of the gain: the TR-X_SUM / IMC_OUTPUT. Perhaps someone can point me to a better approximation. You can see the reduction in gain powering up. The very sharp initial spikes can be ignored.
The third and fourth plots show the gain vs common and differential lenses, respectively. Remember, there is an offset in lensing in both of these plots due to resetting the HWS values to zero.
It's certainly not conclusive, but:
Nutsinee, Sheila
There is something wrong with the REFL and POP cameras, we see no image from them although we shoudl be seeing something on REFL.
Actuallythe POP camera is just misaligned (probably has been since we inserted the beamsplitter monday), but there is still nothing on refl which I think there should be.
The issue with the POP camera was found to be a clipping at the newly installed beamsplitter. This was fixed today. The REFL camera was completely misaligned for some reason, perhaps it was mistakenly moved around. The REFL camera was also realigned. So both are back to functional.
1440 -1510 hrs. local -> To and from Y-mid Opened exhaust check valve bypass valve, opened LLCV bypass valve 1/2 turn -> LN2 @ exhaust in 60 seconds -> Restored valves to as found configuration. Next CP3 overfill to be Friday, July 15th. Also, demonstrated MTP Safety Valve functionality when backed by local scroll pump (rediscovered that the "NORMOAL" light LED on the turbo controller is burned out) -> Let MTP brake phase complete but am leaving turbo rotor levitated, i.e. controller energized, overnight -> will de-energize tomorrow.
~0900 hrs. local -> De-energized Y-mid Turbo controller (Foreline 1.2 x 10-1 torr, turbo inlet 2.6 x 10-1 torr seams odd - gauges reversed?)
The CW injection amptlitude, as measured at the HARDWARE_OUT, appears to have changed on restart May 12. So far, I have not been able to trace down the problem, and conlog is currently down so I cannot see a history of chagnes in the HARDWARE_OUT bank.
Figure 1 observations (10 minute trend of HARDWARE_OUT):
Figure 2 observations (10 minute trend of CW_OUT):
Need conlog to investigate further.
On Jan 20, there was a site-wide power outage (alog 25041), and after everything was restored, one of the filter modules (FM3) was not turned on, but it should have been. This was the 1/AI2 filter with design:
zpk([1.2627e3 - i*2.6133e3;1.2627e3 + i*2.6133e3],[7000;7000],1,"n")
Turning this off flips the phase of the injected signal, but it doesn't really explain the drop in the signal amplitude that I would have expected.
One thing interesting is that there was considerable high frequency noise on the PINJX_HARDWARE_OUT time series that is reduced by turning off the 1/AI2. I plotted the spectrum of the HARDWARE_OUT channel before and after, separated by 24 hours so that the pulsar injections are approximately the same amplitude. Sure enough, with the 1/AI2 filter on, there is a large peak at 7 kHz with an amplitude nearly as large as the highest frequency pulsars. Is this why the pre-power outage time series is about twice as large? It is certainly suspicious...
See attached figures for pre- and post-power-outage spectrum.
