The shift has been dedicated to the tuning of the DAS powers for the working point of the ITF.

21:00 UTC - Commissioning mode stopped and ITF left locked in Low_Noise_3 AUTORELOCK_ FAILSAFE.

DAQ
(19-10-2019 19:15 - 19-10-2019 19:45) From remote
Status: Ended
Description: The DET cameras were not visible and the process DET_Img_CEB crashed.
Actions undertaken: rtpc11 has been restarted by the expert from remote.

ITF found in LOW_NOISE_3. The shift was dedicated to the planned Injections, carried out at the WE Building by the ENV Team.
The ITF unlocked multiple times during the shift, some of these events seems to be caused by the activity itself. It was possible to reach LOW_NOISE_3 after few attempts without major issue. The longest lock was from 8:10 UTC to 11:48 UTC.
Activity still in progress, ITF left locked.

DET
SPRB_B4_PD1 VBias was off since 00:46 UTC. Rearmed and opened the related shutter at 11:46 UTC after an unlock.

SUSP
WE LC opened twice during the shift ( 7:25 UTC, 7:39 UTC ). They were closed before relocking.

TCS
During the night the DMS triggered several alarms related to TCS_CO2__REL3 and TCS_CO2__REL7. These flags refers to two flip mirrors used to engage the Central Heating when the ITF unlocks and disengage it when reaching the Dark Fringe ( see reports #47206 ). The alarms were caused by a delay of the Coils Relay signals of around 8 minutes ( see attached plot #1 ). But after a restart of the process TTL_co2 at 4:35 UTC due to a crash the delay decreased to less than 1 minute ( see attached plot #2 ). During the shift the delay increased again ( see attached plot #3 ) and triggered the alarms. To be monitored.

The afternoon shift has been dedicated to the seismic injections in the LB and tappings on SIB2.

VPM / TCS - 15:41 UTC: TCSMoni restarted in order to update the related flag configuration/alarm conditions.

Suspension / WE - LC control opened by guardian, closed properly at 15:46 UTC.

21:00 UTC -  Commissioning mode stopped and ITF left locked in Low_Noise_3 AUTORELOCK_ FAILSAFE. 

On-line Computing & Storage
(18-10-2019 18:00 - 18-10-2019 20:00) On site
Status: Ended
Description: I called the expert oncall because the operator console crashed and we were not able to restart it.
Actions undertaken: Console restarted.

Today we continued the tuning of TCS parameters.

We tried to increase power on NI CP to increase CMRF adjusting both NI DAS_OUT and NI DAS_IN. We found at the end a no good status of the entire ITF, both sideband power and CMRF were bad. We are not sure this was up to our work, but in any case we decided to restore the situation obtained on 17th at 14.00 UTC.
New DAS parameters are:
NI DAS_IN :     400 mW
NI DAS_OUT : 2.38 W
WI DAS_IN :    159 mw
WI DAS_OUT: 191 mW
Trend will be shown.

Figure 1. Romain found that the OMC correction are more noisy than before the input laser power increase.

Figure 2. Looking at the OMC error signal they are also increased, which is due to the wide noise bump around 10kHz on the B1 PD spectrum.

Figure 3. This noise is coherent with B2 8MHz I so it is probably due to the bad CMRF.

We have changed the metatron configuration OMC_LOCK.ini to increase the OMCs length dither by a factor 10 (similar to what was during the O3 commissioning), and changed online by cm command around 14:26 UTC

ITF found in LOW_NOISE_1. Activity started at 6:48 UTC, the shift was dedicated to ITF recovery and TCS Tuning, carried out by M. Mantovani, E. Cesarini and M. Lorenzini.
Activity still in progress.

Air Conditioning
From 7:25 UTC the flag mean_INF__TB_AIR_PRES_10 became red ( see attached plot ). I was contacted by D. Soldani that guided me in the Technical building to verify that the second compressor started working. This second unit resulted operative. Soldani then restored the main compressor at around 8:25 UTC.

DAQ
TTL_co2 crashed after an unlock, restarted at 10:30 UTC.

DET
SPRB_B4_PD1 VBias was off since yesterday from 11:30 UTC. Rearmed and opened the related shutter at 7:34 UTC after an unlock.

SUSP
WI and WE Local Control opened multiple times due to the unlocks during the shift. They were properly closed thanks to the help of P. Ruggi.

In addition to the 56MHz sideband gain, the B1p power / contrast defect, and the CMRF, the power on B1s2 when in LN3 is an important figure of merit. The B1s2 power when both OMCs are locked is dominated by the 56MHz TEM01 mode, which causes OMC lock acquisition issues and couples the 56MHz RAM to the sensitivity. From that perspective the best state was at around 9UTC with about 0.075mW, this is better that other times, but still higher than the 0.05mW we had during O3a.

Yesterday morning we have started the tuning of the DAS actuators following this procedure:

starting point: CMRF = -0.4e-9; 112MHz mag = 0.081; CD = 19;

- 1st step: increase of the sideband power using the WI outer ring (from 170 mW to 200 mW)

CMRF = -0.5e-9; 112MHz mag = 0.087; CD = 18.4;

- 2nd step: decrease the CMRF using the NI outer ring (from 2.36 W to 2.41 W). This tuining is not differential thus the common mode (the one that maximise the sidebands)  is spoiled  and has to be compensated

CMRF = -0.12e-9; 112MHz mag = 0.083; CD = 18.6;

- 3rd step: increase back the sidebands using the WI inner ring (from 154mW to 124 mW)

CMRF = it was @ LN1.  112MHz mag = 0.085; CD = 18.6;

the trend is visible in Figures 1 and 2  to highilight the pourpouse of the actuator tuning).
Figure 3 shows the tuning to reduce the CMRF.

Among the shifts devoted to PSL durong the oct. break one was focused on the beam realignment on the PSTAB quadrants (and accordingly on the PSTAB photodiodes).

It seems there is no need anymore for that.

Don't know what triggers the trend. Possibly the alignment of the IMC wrt the ITF ?

A PyALP algo was added to perform an automated tuning of the AA working point. The script will tune the setpoints of six angular degrees of freedom (DOF) to find the point of minimum simplified contrast defect (CD_simple = B1_DC/B4_112MHz). The six DOFs are:

• BS_TX
• BS_TY
• PR_TX
• PR_TY
• DIFFp_TX
• DIFFp_TY

1. Begin the alignment: Press the 'Enable AATuning Servo button.  The algo will begin to align the DOFs in the order listed above. Once all the DOFs have been aligned the algo will loop again through all 6 DOFs to refine the alignment further (by default 8 loops). The algo will output status messages in VPM. NOTE: The script does not have protection for if the ITF goes down. You must manually stop the loop if the ITF is not in a condition where the working point can be tuned. There is a glitch protection to skip data that is suspected to contain a glitch but it is not designed to work in an unlocked condition.
2. Change loop iterations: Press the 'Loop Iterations...' button. An input dialog box will pop up where you may enter the number of loop iterations as an integer value.
3. End the alignment: Press the 'Disable AATuning Servo button. The algo will stop immediately and reset all internal counters. When the loop is re-enabled, it will begin again as if for the first time.

This version of the script is stable and has no known bugs. It was developed on VPM: Virgo Online Test and was migrated to VPM: Virgo Online this morning. If you encounter any please contact one of the authors of this entry. As separate development script is still available on Virgo Online Test (On the VPM page click Tools at the top then VPM:Virgo Online Test). This script will be actively developed and the stability of the script at any given time is not guaranteed. This test script will be used to add further features to VPM or to improve the efficiency or accuracy of the algo.

Michal has updated the OMC node in METATRON to improve the logic concerning the setting of the OMC2 initial temperature.

Previously, the OMC2 temperature set in the ini file was updated only after locking OMC2, but not in the case only OMC1 was locked. Thus in case of a modification of the locking temperature of OMC1, this was not propagated to OMC2, which could lead with some bad luck, to having the OMC2 resonant at the same time as OMC1.

Now when the OMC1 is locked, the OMC2 temperature will be updated in the ini file, by the same delta T as the change of OMC1 temperature. This should reduce the odd of having the two cavities resonant at the same time.

This morning the OMC1 lock acquisition was failing because the OMC2 was resonant as the same time as OMC1 (see attached figure). Thus when reaching a resonance on OMC1, the power on the B1 photodiodes was reaching 3 mW, which (as expected) triggered the safety closing the fast shutter.

I reduced the temperature of OMC2 from 23.326 to 23.314°c and the OMC2 is no longer resonant at the same time as OMC1.

This afternoon we carried on the iterative ASC/TCS tuning of the ITF, with a couple of extra issues:

• we had a very strange unlock at 15:03:43 UTC; after some digging, we found that DET_MAIN asked for it because it could read the SDB2 PDs as working, but they were; Alain fixed some parameters on SDB2_Fb, which was restarted at 18:17:20 UTC;
• we did the first tuning of ASC in LOW_NOISE_1;
• the power on the DAS on NI OUT was changed to 2.4 W at ~ 17:08 UTC; later, the power of NI IN was changed to 0.44 W (~ 17:49 UTC);
• we repeated the tuning starting from 17:53 UTC, following the thermal transient;
• later, we experienced a few unlocks while going to dark fringe; we cured them by increasing the DARM gain by +33% (in two iterations) during the hand-off to B1p_56MHz; later, once the UGF servo did its job, the resulting gain was lower than the one in the previous locks, so this is unclear;
• with these settings we had some periods where we were unlocked (WI/WE LC opening, unsuccessfull lock acquisitions, etc...), nonetheless when we got in LOW_NOISE_1 the good working point was re-reached in a quite shorter time than before, despite the higher levels of sidebands (this could already have started in the morning shift, I am not sure);
• once we got back in LOW_NOISE_3 the ASC working point was found to be slightly different, so we adjusted it; we kept the old reference for LOW_NOISE_1 and onwards, and this later one for LOW_NOISE_3; ITF_LOCK is able to do another intermediate step during DC Readout, but this seemed not necessary;
• we had an unlucky unlock at 20:12:33 UTC, where a BS glitch interposed itself between the engagement of the ASC loops and a stronger filter for DARM in LOW_NOISE_1;
• MICH gain was found to be too low after the hand-off to the normalized signal after locking OMC1, given the higher B4_112MHz_mag we have now; one time it caused an unlock (20:24:03 UTC), which was more visible as a 10 Hz oscillation on DARM rather than a MICH one (at lower frequency); this was cured in-lock during the following (and last) time we did a lock acquisition; a tentative mich_gain = 300 is set in the [OMC1_LOCK] section, to be verified;
• The summary of the status of the ITF, looking also at h(t), is the following:
  • the BNS Range oscillates between 35 and 44 Mpc; a good part of the oscillation is due to the BS glitches;
  • given that we are with an outdated Alpha subtraction, no PRCL->SSFS feed-forward and no squeezing, this is not bad;
  • there is a big line at 10 Hz, I am not sure this is real given it is at the border of the range for the reconstruction, but something is also visible (although smaller) on DARM;
  • there is a breathing bump between 15 and 45 Hz;
  • the PRCL coupling is higher (by roughly looking at the 150 Hz line and around it), but this is expected;
  • the CMRF is still quite bad; what we call "contrast defect" (actually B1p_DC/B4_112MHz_mag) is around 18, which is not bad but it could be smaller;
  • B4_112MHz_mag reached 0.0089 but, at the time of writing, a small declining trend is visible;
  • B1p_DC is around 0.16 mW; the trend here seems much fainter.

We left the ITF in LOW_NOISE_3, with the following changes to be noted:

• we only tested manually the gain for MICH after OMC1 is locked (see above);
• since the PRCL loop seems fine, the UGF line is again turned off in the usual way;
• since MICH has to be checked, its line is turned off as well, but only well later (ACQUIRE_LOW_NOISE_3) and by a hardcoded command, instead of using ampl = 0 from the config file alongside the PRCL one;
• ITF_STATUS has been set in AUTORELOCK_FAILSAFE, but the target state (in case it gets triggered). is not DOWN but LOCKED_RECOMBINED.

The activity on ITF tuning went on for the whole shift, without major problems, at LowNoise1 and LowNoise3 with BNS Range ~42 Mpc.
It has been carried out by Ruggi, Mantovani, Bersanetti.
At 16:00 UTC Chiummo went in Laser Lab to make a PMC alignment tuning.

DAQ
- 15:05 UTC - TTL_co2 process crashed / restarted
- 21:00 UTC - FbTrend_10800 process crashed / restarted

DMS
TCS_WI_CO2_PWROUT thresholds changed.
Logic of TCS_CO2__REL7, TCS_CO2__REL3 changed in TCSMoni config file

SUSP
At the unlock event of 20:24 UTC, guardians opened LC on WI and WE. WE LC properly closed activating the CRS ENBL.
For the WI, after first attempt to close LC, also the F7 controls have been opened by the guardians. The expert, from remote, provided to close both loops.

# Effect of the M8 mirror mount swap (last light green square @ 2/10/2019)
- it seems it triggers a slow down of the horizontal drift (yellow line, AMP_Cam_PosX: horiz)
while there is no effect on the newly appeared (end of august) vertical drift (violet line, AMP_Cam_PosxY: vertic)

- The effect on the PMC alignment is not obvious

plot 1 gives the trend on neoVAN & PMC signals after the M8 mirror mount swap (02/10/19) and the neoVAN alignment adjustment

(plot 2 gives the legend of the plot)

After the swapping of a mirror mount on the laser bench (see 47088) the slow drift of the PMC power seems to be less evident (first picture), as already reported by M. Gosselin in the last weekly meeting. Unfortunately, there has been a misalignment event during latest maintenance, so i took advantage of a lock loss to try and restore the alignment of the beam injected to the PMC. It was a quick tuning, just looking at the reflection of the cavity without second iteration.

The results are shown in the second plot. After the alignment, I adjusted the power sent to the IMC in order to restore the transmitted power (within 100mW accuracy). The PMC power increased by roughly 3%, but seems now less noisy.

For the SSFS and SQZ rtpcs, the software architecture is the following

• realtime CPU0: the control server running at 50KHz  with all the datas at 500KHz
• realtime CPU1: the noise server running at 50KHz and providing  the noise datas at 500KHz
• relatime CPU2: the extension server running at 10KHz and extending the 10KHz datas to 500KHz datas, mainly the phase and  the DC power channels provided by others servers at 10KHz .The servers are respectively SSFS_phi for the SSFS and SQZ_PLL_AA  for the SQZ_CC.

The extension was performed by a server running at 10KHz, meaning that every 100us it provides a buffer of 50 filtered samples and the control server accesses 5 times this buffer by reading 10 samples each time. Unfortunately It was not the case  and the control server was accessing 5 times the buffer and reading each time the same 10 samples: as consequence a lot of imaging (aliasing) was present in the extended channels. The reasons of this malfunctionning are well identified and will fixed

To improve the situation,  this morning  the extension was moved in the noise server  and the extension filters tuned.

The plots show the INJ_IMC_REF_I_POST equal to the sum of the INJ_IMC_REF_I_POST and the SSFS_Corr channels, blue before the upgrade and purple after where one can see that all the 10KHz lines have disappeared

• ITF_LOCK_index at 90 and the SSFS working the O3_SSFS_base_flt filter and a zoom for Freq > 10KHz
• ITF_LOCK_index at 160 and the SSFS working with the O3_SSFS_boost filter, zoom around 10KHz

After a call from the control room, I looked at two unlocks at the following time:

• 12:22:15 UTC - ITF in LN3
• 12:38:25 UTC - ITF in LN2

The two unlocks are due to the triggering of the fast shutter because of B1_PD1/2_DC channels reached 3 mW.

The B1_DC reaches 3 mW because of an oscillation growing in DARM around 4.5 Hz in the first case and around 10 Hz in the second case.

One can see that the audio channels of B1_PD1/2 started to saturate before the fast shutter trigerring causing the OMCs error signals to be spoilled.

Picture 1 shows the unlock at 12:22:15, where the DARM oscillation is at 4.5 Hz and growing very rapidly.

Picture 2 shows the unlock at 12:39:25,  where the DARM oscillation is at 10 Hz and growing less rapidly than the first studied unlock.

In both cases, the fast shutter triggered a few 10s of milliseconds after the plot.

I don't know the origin of the 4.5 Hz and 10 Hz oscillation in DARM.

This morning I found the ITF in "locked recombined", the whole shift has been dedicated to the ITF recovery and tuning, the ISC crew is still working at 13:30 UTC. All the subsystems worked properly, no particular problems or events to report.

After the tuning of the injection system (#47259), we started working on the ITF.

Before reaching LN1, a check of the SSFS demodulation phase was performed, but it was off only by ~0.05.

In order to compare the SSFS OLTF before and after the INJ intervention, a noise injection on SSFS was performed at PRITF MICH offset 0.5, and the two measurements are compared in figure 1. No change in the delay was observed, but only a small change of gain.

Then we locked to LN1 and we scanned the OMC1 in order to look, as an additional figure of merit for the TCS working point tuning, the sidebands amplitude balancing, the amplitude of the CAR order2 mode and the relative amplitude of the order2 mode of the sidebands.

The first step was performed with the initial tuning set by Ilaria, which was

WI --> 162 mW INNER, 172 mW OUTER

NI --> 353 mW INNER, 2.36 W OUTER

The first scan (figure 2, GPS = 1255288220) shows that in this configuration the sidebands are well balanced, and there are 0.61mW in the CAR TEM02 mode.

Then Alessio peformed another step on the WI INNER DAS, bringing it to 220mW. The second scan (figure 3, GPS = 1255291518) shows that there is an unbalance of the sidebands of about 30%, that's why we decided to go back to the previous point. However, in this configuration the CAR TEM02 is reduced to 0.43mW.

We finally went back to the initial configuration for the DAS, and from the scan (figure 4, GPS = 1255295868) we got an unbalance of 13% and 0.51mW in CAR TEM02. Notice that the thermalization could be not yet completely finished when the scan was performed.

It is possible to get some information also from the relative amplitude between the SB TEM02 and TEM00. For instance, the amplitude of LSB TEM02 with respect to LSB TEM00 in the three scans is 17%, 20% and 16% in the three scans, respectively. However, this latter measurement is not really accurate, as the TEM02 peak is in the shoulder of the USB TEM00 peak, and needs a deconvolution to get a precise information, although we can still look at the trend between one scan and the next.

After few unlocks in LOCKING_OMC1 happened at the beginning of the scan, the fast retuning of the working point was performed, and the ITF reached LN3. However, since the lock was not really robust and the relock failed many attempts, ITF was left in LN1 autorelock failsafe.

ITF found unlocked, the effects of the Eartquake were still present on the Suspensions. The first part of the shift was dedicated to the INJ Activity ( see report #47259 ), for these operations the ITF had only the PR mirror aligned.
From 17:20 UTC A. Allocca, A. Rocchi and R. Gouaty started performing scans of the OMC to tune the TCS working point. Activity still in progress.

Air Conditioning
Since yesterday afternoon from around 18:50 UTC the pressure inside SIB2 Room increased its oscillation range ( see attached plot ). The expert is aware of this behavior.

DAQ
InframoniAlarmed restarted at 15:17 UTC to update mean_HVAC__NEB_COLD_TE minimum threshold under request of the expert. Value changed from 12 to 14.5.

Other
I performed the periodic check of the shelved flags. Summary in the attachment.

Today I added an optional, stricter control for the Central Heating in LOCKED_PRITF_DF: the purpose is to prevent the ITF to stay in such state with the CH on while unattended, due to the crash of the TTL_co2 process. While a more robust and reliable solution will be pursued (emails will start soon about this), for the time being here is what is set up:

• at the top on the ITF_LOCK.ini file there is a new flag, tcs_ch_failsafe, which is set to False as default, with the same principle as the flags fmoderr_skip and sqz_debug, which is to switch something to a safer state;
• when this flag is set to True (before reaching LOCKED_PRITF_DF where the node reads it), the check on the CH turning off correctly will also include a counter; if the check fails a number of times (parameter tcs_ch_max_tries in the [TCS] section a little below in the same .ini file), the node will assume that the CH cannot be turned off because the TTL_co2 process has crashed;
• at this point, ITF_LOCK will go to DOWN and self-request LOCKED_RECOMBINED as a fallback state, where it is ok for the CH to be on and the alignment will not be lost;
• when people are around, the flag should be set as False, since ITF_LOCK will notify the problem and people can restart the TTL_co2 process and Lock Acquisition will then carry on without any other intervention; it would be unnecessary, in this case, to force an unlock.

The code is in place but it is not tested yet, please let me know in case the ITF is left in such configuration for the night.

We compensated for the increase of power on the error signal photodiodes by decreasing the impinging power by a factor 1.5

- IMC REFL PD: from 3.4 mW to 2.3 mW when the IMC is unlocked (plot 1) by tuning IPC8 (with EIBRot axis 1 ch1). We then had to compensate the gain (attenuation from 15 dB to 11 dB) in order to come back to an UGF of 130 kHz (plot 2).

- IMC Aa quadrant photodiodes: we tuned the power on the RF quadrants with IPC 5 for the FF (EIBRot axis 1 ch 2) and with IPC 6 for the NF (EIBRot axis 2 ch 2). The quandrant in transmission which is used in DC doesn't have a rotator, so we let it as it is (it is still far from saturation). We had then to remeasure the sensing matrix

-B2: with the PR aligned we tuned the IPC3 to decrease from 31 mW to 23 mW (in the minirack in the CB)

-PSTAB: we didn't find the actuator so we let it with the gain set this morning by F. Cleva and J.P. Coulon #47250

In the table are reported the parameters of the injections to measure the IMC Aa sensing matrix

The effect of these operations will be better evaluated on a longer period.

The last test for the lock of CARM with the signal SIB2_RFC_8MHz_I was done on August 1st 2019; those tests were done with a very rough filter, unconditionally stable but with very poor performance, just to have a starting point.

A trend of the meaningful channels is in Figure 1; of all the tests and configurations explored, probably two stretches of data can be useful before the next shift:

• GPS = 1248692748 (DUR = 300 s): CARM loop closed with "high" gain in LOW_NOISE_1;
• GPS = 1248700638 (DUR = 300 s): CARM loop closed with "low" gain in LOW_NOISE_3.

Just to have a comparison, I replicated in Figure 2 the first plot of the original entry with these data stretches.