Upon arrival, the planned activity on ISC (carried out by Boldrini, Ruggi, and Casanueva) was already started.
It went on without major problems for the whole shift.
Activity still in progress.

Parallel activity:
SGD - FC alignment Recovery + visibility with FCIM (De Laurentis, Sequino)

DAQ
13:55 UTC - FdRawFull1 crashed / restarted
13:58, 14:27 UTC - FdRawFull crashed / restarted

SUSP
13: 58 UTC - WE Guardian opened LC after the unlock. Properly closed.

The goal of this shift was to investigate the effectiveness of the new 50MHz demodulation ferquency as error signals for the SR alignment.

We checked the demodulation phase of DIFFp, but the one saved in ITF_LOCK.ini (-1.6 rad) was optimal for CARM offset zero.

Despite numerous unlocks caused by SSFS engagement misfires, or an uncontrolled excitation of the optical spring at CARM offset zero, we managed to obtain some longer locks that allowed to investigate the new demodulated signals. In particular we observed the behaviour of B1p_QD1_50MHz and B1p_QD2_50MHz while tilting the SR on TX.

It appears that B1p_QD1_V_50MHz behaves like a good error signal for SR_TX (Fig.1), even if the demodulation phase is still mistuned (for this plot we set B1p_QD1_50MHz_phi0 = -2.4 rad).

Another attempt failed due to the optical spring getting excited. The oscillation induced by it on COMMp is visible on both quadrants on B1p_50MHz, and the movement of the SR is not clearly discernible anymore.

The work on SR alignment will continue in the next shifts.

We leave tha arms locked on the IR
 

During the discussion after the talk about the commissioning of the IMC baffle,  an hypotesis to explain the presence of co-resonating HOMs was put forward, involving the possibility that the heat dissipation of the baffle could be responsible for a deformation of the mirror (ring-heater-ish effect) which in turn could put some HOMs  in resonance. Then Vincenzo spotted that from time to time there is a large transient on the temperature reported by embedded baffle sensors, which Mario explained to be due to accidental switch off of the baffle electronics. As suggested by Raffaele, these times were a good moment to look for the possible effect described before.

In the first plot, an overall trend of the data related to the aforementioned switch-off events. The normalized power transmitted by the IMC does not show evident correlation with the times when the baffle allegedely switched off.

I picked up some images of the MC camera to check whether the apparent mode contents visible on the MC end mirror could change during the transient of the baffle temperature:

pic 1) MC cam during baffle ON - steady state before switch off;

pic 2) MC cam during baffle OFF - first grey area on the trend data;

pic 3) MC cam during baffle OFF - first grey area on the trend data, later on;

pic 4, 5, 6, 7)  MC cam during the transient;

pic 8) basically steady state again after the transient.

So again, from the evolution of the camera pics there is no clear correlation with the status of the baffles.

After the replacement of the RF driver, the switch on of the NI CO2 laser affected also signals acquired by IMMS and ENV systems.

In particular (see jump3.gif) the jump is visible in INF_TCS_CHILROOM_TE (AD590 temperature probe inside TCS chiller room).

Also temperature sensors (AD590) installed into NI superattenuators see a similar jump.

In jump4.gif we see the effect of switch on of the NI CO2 after the replacement of the RF driver with the original one.

The INF_TCS_CHILROOM_TE signal doesn't  change because in the meanwhile we replaced the sensor with a PT100, while NI superattenuator temperature sensor are still affected by the jump.

 

ITF State: Infrared Cavities locked.
Quick Summary: IMC and RFC locked; all Suspensions loops closed, all SBE loops closed
Activities ongoing since this morning: Commissioning activity ongoing

No DMS Event Monitor to report

The goal of this afternoon's shift was to test the new available signals SDB2_B1p_QD2_{V,H}_50MHz_{I,Q} (see entry #54597) for the BS and SR alignment.

While we were waiting for the recovery of the NE suspension (see #54598) we did a few preliminary actions, also on other items:

• we propagated the new signals, acquired as ACL_ADC_CH inputs in ASC_Acl, to the current loop structure: demodulation phases, rotated signals, inclusion in the TX and TY global sensing matrices, push to the DAQ of phases and rotated signals; signal names are coherent with the current naming scheme;
• we updated the LSC_Acl_Moni process: we changed the new channel for the monitoring of the DARM TF high frequency pole, LSC_DARM_HF_pole_mag, to LSC_DARM_HF_pole_mag_raw; we added a new constant LSC_DARM_HF_pole_cali (defaulted to 888.0), so now the channel LSC_DARM_HF_pole_mag is the product of these two and is calibrated in Hz and should give directly the pole frequency (we just implemented what was developed already in #54590);
• we restarted the SusDAQBridge process, in order to expose to the DAQ the recent setpoints GNAMEs for the dithering ({NI,WI,NE,WE}_MIR_{X,Y}_SET); in order to do this, I had to manually remove many duplicated GNAMEs from the /virgoData/Sa/PySatServer/SAT_GAINS.ini file (named TEST_COM) which have been left over there for over a year.

Once the recovery of the NE suspension was over, we relocked the arms and waited a bit with the Y damper on (thanks to Paolo) in order to reduce a ~ 7 Hz mechanical oscillation which arised. Then we started to relock the ITF, with the first goal of retuning the B1p_QD2_56MHz demodulation phases:

• we did not manage to lock the DRMI for the first long part of the shift, with very low peaks on the B4 DC and SB signals, and no locks; we then discovered an issue with the WI Central Heating power (Figure 1, as reported in #54602); we let the TCS crew to check the issue and we restarted after some time, also waiting for some re-thermalization of the ITF;
• we had the first DRMI lock at 19:49:30 UTC; we had quite mistuned figures of merit (high B1p_DC, low B4_DC and sideband powers); we tried to reach the first viable state for the tuning of the QPD demodulation phases (CARM_OFFSET_STEP_3_OF_3, meaning 480 mW of arms powers, where the DIFFp is engaged in drift control); in order to do the tuning, we disabled the engagement of the loop (lines 2213-2214 of ITF_LOCK.py); unfortunately, we unlocked during the ramp from 160 mW to 480mW, due to a high frequency oscillation on DARM (Figure 2);
• we did the same a couple of other times, then we decided not to change all the lock acquisition parameters, as they would soon need to be reverted, but to stay in the previous step for a while (CARM_OFFSET_STEP_2_OF_3, meaning 160 mW of arms powers), hoping to hasten the thermalization of the ITF; this started at 21:00:42 UTC; some trend is visible in Figure 3;
• while we were at this step, we noticed that DARM gain was quite high and resulting to a UGF of around 62 Hz; as we suspect that some of the (very few, especially at 480 mW) unlocks we usually have at this stages also with a stable ITF could be related to a high DARM gain, we decided to change it and save it in the configuration file (from 0.03 to 0.022);
• then, at around 21:23 UTC we moved past this step and reached again STEP3, 480 mW; here we noticed the same issue with DARM so we changed also the gain here, from 0.035 to 0.022;
• these are the only two changes we did to the lock acquisition; the extent of these changes could be reduced once the ITF is in the same configuration as before, but probably some tuning of these gains is needed nonetheless; we looked at yesterday's long lock at the same step (Figure 4), but unfortunately we never stay in these intermediate steps long enough to let the UGF monitors to settle, so this should be checked more; however, we can see in the Figure a different trend of the SBs and B7_DC and B8_DC swapped;
• at this step we started to tune the 56MHz demodulation phase but we unlocked quite soon, after a small WE_TX step, due to a ~6 Hz oscillation (Figure 5);
• at around 21:38 UTC we relocked the DRMI; profiting to the time spent at 160 mW and 480 mW in the previous lock, now the usual figures of merit of the DRMI lock were much more similar to the usual ones, although not completely there; we went again to STEP3 and unlocked again quite soon, this time because of a ~ 90 Hz oscillation visible on most signals, particularly on DARM and the CARM_MC correction;
• we did another lock a little time later, but the usual figures of merit of the DRMI lock already degraded a little; we unlocked basically in the same way as in the lock before, at the same step (Figure 6).

Conclusions:

• the demodulation phases for the DIFFp signals (ASC_B1p_QD2_{V,H}_56MHz_phi0) are still to be re-tuned;
• the same for the new 50MHz signals;
• we changed the gains for DARM in the [CARM_SET_STEP2] and [CARM_SET_STEP3] sections of ITF_LOCK.ini, related respectively to the 160 and 480 mW arms powers' step in the CARM offset reduction phase; to be verified/adjusted;
• the ITF is nominally in its usual state, but the variability of the DRMI figures of merit and its lockability suggest some additional check; also, the PyTCS loops were still quite tuning the DAS powers, but we don't have the grasp of the numbers so we don't know if the changes are significative.

We left the arms locked on the IR.

The DAQ servers running on the olserver52 hosts have their CPU affinity well defined meaning that a each server remains running on a given CPU . If  the affinity of a task is not well defined  the task runs on the different CPUs according the system policy.

During the week-end the CPU affinity of the Metatron servers has been fixed.  This operation was complete around the 2022-01-23-11H-UTC . The detail of the CPU affinity for the olserver52 host can be find in  the ~virgorun/bin/DaqOlserver._taskset.sh script..

The following plots show according different CPU affinity policy applied

• the DAQ olserver52  frame merger servers latency and the olserver52 global cpu_idle value
• the olserver52 load reported by ganglia .
• the cpu_idle value for each one of the 32 olserver52 CPUs

There is 3 main periods called

• "fixed taskset" meaning that the CPU affinity for each of the DAQ and Metatron server is fixed
• "free taskset" meaning that each of the DAQ and Metatron server can run and move on all the olserver52 CPUs
• "mix taskset" meaning that the CPU affinity was not fixed only for the FbmFFE,FbmFE, FbmMain frame mergers and all the DAQ 50Hz servers

From the plots one can see that

• the latency at the FbmMain level is lower by 0.5s  in the case  of the "free taskset"  period ,compare to the  "fixed taslkset" period
• the global cpu_idle value is reduced by 5% in the case  of the "free taskset"  period ,compare to the  "fixed taslkset" period, meaning that all the CPUs are more used (32*0.05=1.6 CPU)
• the olserver52 ganglia load value is reduced  from 24 to 18 in the case   the "free taskset"  period ,compare to the  "fixed taslkset" period,
• in the case of the "mix taskset" , the latency remains at the same level as the "free taskset" period while  the global cpu_ilde value remains at the same level as the "fixed taskset" period .
• the effect observed in the transition between "fixed taslkset" period and "free taskset" period seems repeatable

According to these observations,

• in the first time  the CFG_SCHEDAFFINITY of the olserver52 tasks could be commented to run like the  "free taskset"period
• in case of need one can run the olserver52 tasks as the  "mix taskset" period .
• Now all the olserver52 servers are running as in  "free taskset" period but the olserver52 task configuration files are not yet updated

Storage host  CPU affinity state:

• the CPUs affinity for the stol01 tasks remains as "fixed taskets" . The detail is available in the  ~virgorun/bin/DaqStol01._taskset.sh script.
• the CPUs affinity for the stol02 tasks remains as "fixed taskets" . The detail is available in the  ~virgorun/bin/DaqStol02._taskset.sh script
• the CPUs affinity for the stol03 tasks is now as "free taskets" . The detail is available in the  ~virgorun/bin/DaqStol03._taskset.sh script

In today's shift we tried to improve the alignment of the SC into the filter cavity and check the losses of the path towards the filter cavity.

In order to improve the IR alignment, we first checked the green alignment situation compared to the data before all the hardware work. We have about 11% of misalignment and 5% of mismatching, which is  similar to what we had before. Just in case we tried to shift the beam horizontally inside the cavity with SQB1 GM11 and SQB2 M1, then recover the cavity alignment, the transmitted power was still around 8.6V. 

Then we improved the alignment of the SC by moving the SQB1 mirrors and the IR BPC. We could reach a maximum around 11V. In this situation, we measured the power in reflection on IR_PD_MONI with the cavity and the delay line. The delay line reflection we measured 1.03V, cavity reflection we measured 1.01V. But we knew that the wave plate in front of the HD needs to be tuned for different configurations to reduce the spurious polarization. So we went to the lab to check the condition of the waveplate. 

Entering the lab, we created a lot of noise, the cavity alignment moved with the green beam, so we lost our good alignment on the IR. We managed to tune the wave plate to minimize the spurious polarization of the beam reflected by the delay line. The maximum reflection of the delay line measured on IR_PD_MONI was 1.069V. Then we rotated the waveplate again to the position where the FC reflection spurious polarization minimized. By better aligned the cavity we could reach1.016V on IR_PD_MONI.(pic 1) (SQB1 HWP2 was rotated to have minimum transmission). According to these two values the losses we have between EQB1 and FC input mirror was about 5%. 

Under the best alignment condition, we checked the beam shape on EQB1 HD cams, the beam seems not clipped.(pic 2) But on SQB1 cams, the beam shows a strangely round shape seems from some optics.(pic 3)

At the end we also managed to measure the SC power on EQB1.

Before entering FI  17.59 mW +/- 0.03 mW

Before SQB1 14.10 mW+/- 0.03 mW

In front of the HD 13.22 mW+/- 0.03 mW

So this gives us a loss of about 6%, compatible with the measurement using PD_MONI. 

We measured also the TEM01 mode's amplitude in transmission, it was about 0.55V.(pic4) So we have about 5% of misalignment.

The issue of the NE suspensions, reported in the previous operator report, was fixed at the very beginning of the shift.

The commissioning team started to work at the planned ISC activity but it was not possible to lock the ITF in DRMI; we spent some time to investigate the problem and finally we discovered that the WI CH power injected into the ITF decreased; we contacted the TCS experts on site and they fixed the problem.

After their intervention, at around 20:00 UTC, we were able to relock the ITF in DRMI but the ITF did not behave properly (see next entry from ISC cerw) and CARM_NULL_1F has not been reached yet.

ISC activity still in progress...

Parallel activity for the SQZ in DET LAB and in control room.

Diego contacted us after having observed a strange behaviour of the signal TCS_WI_CO2_POWER_CH_PICKOFF measuring the sum of powers of CH, DAS IN and DAS OUT.

A mysterious step  at around 11 UTC caused the decrease of the WI CH from 80 mW to ~ 15 mW (see figure). One hypothesis to explain this behaviour is the change in the polarization of the laser but the trigger is not clear and has to be investigated.

We spent 10 min on the bench to verify the proper operation of the power meter and everything was working.

We restored the WI CH nominal power injected into the ITF at 18.50 UTC.

Today we were in squeezing mode.

1. Alignment of SC and LO on HD

We checked the alignment of the SC and the LO on EQB1_HD_DIFF_RF_4MHz_mag.

After the alignment with M13, M4 and M6, we reached a magnitude oscillating between 0.00226 and 0.00658, which corresponds to parametric gain of 2.9 (9.25 dB of squeezing producted).

We adjusted the weight of the EQB1_HD_DIFF_AUDIO_RMS_1_3kHz:

ACL_SUM_CH    HD_DIFF_AUDIO_RMS_1_3kHz_norm    ""    1    273973      HD_DIFF_AUDIO_RMS_1_3kHz

2. Attempt to close the CC fast and coarse on LO_M2 and M4/M6

We worked with the dither line on M4_Z off.

We needed to restart the EQB1_CC process since we were not able to remove a 3Hz fluctuations  on the CC corrections. After the restart the oscillation disappeared.

Then we tried to close CC fast and CC coarse but we noticed from the mag signal that the HD was misaligning very fast.

We tried with CC coarse gain -1 and CC gain of 10, 20, 30, 50, 100 but the system was always misaligning.

We tried to send the correction of the CC coarse only on M4 and not M6 but the corrections were drifting very fast.

3. Attempt to close CC coarse loop on FC input mirror

We sent the correction to FCIM_LC_MIR_Z, so the new correction signal is FCIM_LC_MIR_Z_ACT.

We sent a large noise on the input mirror in Z to check the range of the mirror:

NOISE (FCIM_LC_NOISE): sine at 1Hz up to 20 V of amplitude

MIR_Z: oscillation of 19 um (with 20V of noise amplitude)

This is shown in Fig. 1.

To send this noise we used the following function on python (since the other python functions were not working well):

FCIM_LC.sin_noise.AcSinChSet(frequency, amplitude, 0.0, 0.0, 1.0).

4. Diagonalization of MIR_Z, MIR_TX and MIR_TY

We notice that sending a noise on MIR_Z, the spots on FCEB and EQB1_GR cameras was moving a lot in VER and a bit in HOR.

We diagonalized the mirror driving matrix in order to reduce this coupling between different dofs.

We checked the signals MIR_Z, MIR_TX, MIR_TY, changing online via python the weights on the coils of the input mirror, to reduce TX and TY.

To do a final tuning we checked also the oscillation on the cameras X and Y. Even if minimizing TX and TY we had some residual motion on the cameras. This shows that there is a coupling in the optical lever signals. We minimized the beam motions on the cameras with the following driving matrix:

MIR_R = 0.115

MIR_L = 0

MIR_B = 1

MIR_T = -1

Today we performed further tests with the NEB coil

• we investigated a likely reason for the anomalous phase rotation noticed in the measured TF between the ENV_NEB_NOISE and magnetometer signals (see previous entry) (results to be posted).
• we practiced injecting current noise and voltage noise, at various levels, to get an idea of the maximum colored magnetic noise that we can deal. Both the amplifier saturation limit and the magnetometers saturation limit have to be taken into account. While at high frequency (above 100Hz) injecting in voltage mode allows to avoid amplifier saturation, below 100Hz we soon reach the limit of magnetometers saturation. We find that roughly that in the 0-1000Hz we can inject a noise level that is a factor 100-200 above the quiet ambient noise.
• we measured the TF between magnetometers in the reference location (N and W magnetometers) and one magnetometers placed in different locations in proximity of the NEB vacuum chamber (data to be analyzed and posted).

Details and gps times of single injections are in the attached log file

The shift was dedicated to the planned Maintenance. Here is a list of the activities reported to the Control Room:

• Cleaning of Central Building (Menzione, Ciardelli with external firm);
• Standard Vacuum Refill (VAC Team);
• DAQ SDB2_B1p_QD{1,2} demodulated at 50MHz (see entry #54597);
• From 8:15 UTC to 9:15 UTC, CERN tiltmeter maintenance at NE Building (Di Girolamo);
• At around 8:30 UTC, replacement of Nexus power supply in TCS Room (Tringali);
• From 8:30 UTC to 11:50 UTC, ENV Intervention at NE Building (Fiori, Tringali, Paoletti);
• From 11:15 UTC to 11:50 UTC, TCS Inspection at NE (Lumaca, Nardecchia);
• At 9:25 UTC, TCS chiller refill (Menzione);
• DAQ trend stream (see entry #54599).

From 14:15 UTC Boschi worked to restore the missing connection with Sa_NE (see Network and SUSP sections), activity still in progress.

EnvMoni
At 11:24 UTC we lost the connection with the imms-13 box (see "Network" section). After the switch was restored I unplugged and plugged back the Ethernat cable of the device: this action restore its connection and the ENV signals were acquired again.

Network & GSS
At 11:24 UTC signals related to the NE started to be corrupted (ENV signals related to Hall temperature and lights) or missing (Sa_NE Loop and Guardians).
At 11:50 UTC I went to investigate the issue at the building and found that one of the Network Switch was off: after pressing its electrical outlet lightly in the socket it turned on again. After this action the connected devices restored their network connections. Expert informed via phone of these events.

SBE
SNEB SBE opened multiple times during the shift due to human presence in the Building. Loop closed from VPM.

SUSP
Due to the issue at NE (see "Network" section) the signals related to NE_IP Loops status and Guardians weren't available anymore from 11:24 UTC. After restoring the Network switch the signals were still missing, SUS_NE Metatron was in Error state and unable to change state. From 14:15 UTC Boschi worked to restore the connection of Sa_NE, activity still in progress.

The FbTrend_200_{,2} servers were found using more than 80% of one cpu . In order to improve the situation a third FbTrend has been setup .

The running setup is now :

• FbTrend_200: using the FDIN_TAG "* -V1:FDS_* -V1:LFC_* -V1:AFC_* -V1:SQZ_* -V1:SBE_SQB1_* -V1:SQB1_* -V1:SBE_SQB2_* -V1:SQB2_* -V1:SBE_FCIM_* -V1:SBE_FCEM_* -V1:FCIM_* -V1:FCEM_* -V1:FCIB_* -V1:FCEB_* -V1:VAC* -*_FS"
• FbTrend_200_2 using the FDIN_TAG "V1:FDS_* V1:LFC_* V1:AFC_* V1:SQZ_* V1:SBE_SQB1_* V1:SQB1_* V1:SBE_SQB2_* V1:SQB2_* V1:SBE_FCIM_* V1:SBE_FCEM_* V1:FCIM_* V1:FCEM_* V1:FCIB_* V1:FCEB_* V1:VAC* -*_FS"
• FbTrend_200_3 using the  FDIN_TAG "*_FS"

The FbTrend and the FbTrend_10800 server configurations have been updated to as input the FbTrend_200{,2,3}  ouput

The operations were performed between 12h45UTC and 13h45UTC . During this period some trend data are missing

This morning  the demodulation of the SDB2_B1p_QD{1,2} signals at 50MHz has been setup .

To allow this :

• In agreement with the ISC team the SDB2_B1p_QD_{1,2}  demodulation at 18MHz has been removed
• The fpga demodulation frequency has been set  to exactly 8 times the one use to demodulate the 6MHz  to allow the DAQ-LNFS phase correction with the  8*6 MHz phase extracted at the LNFS level .

The channels SDB2_B1p_50MHz_{H,V}_{I,Q} are sent to the ASC_Acl server  . The ASC_Acl server has been upgraped to acquire these new channels , of course the SDB2_B1p_18MHz_{H,V}_{I,Q}  channels have been removed.

Only the demodulation mezzanines acquiring the SDB2_B1p_QD{1,2}  channels have been reconfigured : as consequence only the  SDB2_B1p_QD{1,2}_XMHz phases have to be adjusted.

ITF State: Infrared Cavities locked.
Quick Summary: IMC and RFC locked; all Suspensions loops closed, all SBE loops closed; Sa_MC_F0_Z out of threshold.
Activities ongoing since this morning: Cleaning of Central Building.

Figure 1 shows the long (many hours) lock from yesterday during which alignment of SR and BS was tuned in steps over several hours

Figure 2 is the same zoomed into the first hour. During the SR TY adjustment, SDB1 moves by ~10urad to follow the beam and the B5 power changes by up-to a factor 2. What is concerning is that it also change in shape.

Figure 3 shows that a beginning there is mostly a single bright spot on B5.

Figuer 4 shows that when SR is moved and B5 becomes brighter then several distinct spot are visible on B5.

Figure 5 shows that in the final working point the B5 beam shape is better, but still not a single round spot.

So the B5 beam may not beam a good pointing error signal for the SDB1/OMC as it was in O2 and O3. Most likely because it is partially recycled by the SR mirror. Checking for potential OMC alignment offsets due to this B5 beam shape will be the first thing to do once DARM DC read-out will work.

This afternoon we continued the work on the robustness of the lock, testing the new signals which have been put online (see entries #54585, #54590 and #54592). For this reason we did not test new features or additional loops, but we focused on a long lock and tried to understand the behaviour and reliability of such signals.

Here follows the summary of the activity; we had three short locks and a longer one, which lasted up to the end of the shift. The description of the four locks will follow; the other causes of unlocks during the shift were:

• 17:07:07 UTC: unlock at the handoff of CARM to the MC;
• 18:10:17 UTC and 18:14:29 UTC: unlock during the DRMI lock acquisition; to be checked offline, but they looked related to the ALS;
• 18:46:01 UTC: unlock during the engagement of the SSFS.

About the four locks at CARM_NULL_1F:

• the first lock at CARM_NULL_1F started at 16:18:40 UTC;
• we started to optimize the working point with the usual methodology: offset on PR_TY, injection of DARM noise and tuning of the high frequency DARM pole, then the low frequency pole for the optical spring;
• the dithering lines for the BS were turned on;
• we wanted then to look the new signal LSC_NArm_B7_OS_I; we added the existing DARM_line_phi0 demodulation phase to the demodulation computation in LSC_Acl_Moni replacing the constant one;
• we unlocked at 16:50:43 UTC when we restarted such process;

• the second lock started at 17:18:02 UTC;
• at 17:22:56 UTC we started to inject DARM noise and perform the DARM TF tuning; the high frequency pole was already in the proper place (DARM TF phase ~= 0.8 at 100 Hz);
• we unlocked 17:38:57 UTC after a 0.2 urad movement on BS_TY destabilized the ITF;

• the third lock started at 17:50:31 UTC;
• we unlocked again at 18:01:53 UTC during the DARM TF tuning procedure, due to a strong ~13 Hz oscillation visible on all signals (Figure 1);

• the fourth, and last, lock of the shift started at 19:02:20 UTC;
• the overall trend of all figures of merit is in Figure 2;
• this time we closed also the dithering loops, with smaller offsets than usual on X, while the Y DOFs were closed at zero; we had an increasing trend of the cavity powers during the lock;
• we followed the usual procedure, but trying to use already the newly generated signals, and relying on the DARM noise injection only to cross-check the high frequency pole:
  • we used the LSC_DARM_HF_pole_mag signal to tune the high frequency pole; as reported in #54590 such signal represents the calibrated frequency once multiplied by 888; so we tried to tune it to be around 0.225 and cross-checking with the DARM noise injection wheter the DARM TF phase was approximately 0.8 at 100 Hz, by moving SR_TX and compensating with BS_TX for the dark fringe;
  • we used the LSC_OS_LF_mag_Hz signal to tune the optical spring pole down to 14 Hz, and looking at the behaviour of the new LSC_NArm_B7_OS_{I,Q} signals;
• the entirety of the shift was devoted to keep the ITF locked and correct the alignment drifts of the BS and SR, in order to collect data and better characterize the new signals;
• the demodulation phase of the new LSC_NArm_B7_OS_{I,Q} signals, i.e. LSC_DARM_line_phi0, was not optimized as it was not straightforward to remove any meaningful information from the _Q phase; we did a couple of attempts later towards the end of the shift, when we took some data with different demodulation phases; when we used 0.8 we found that the _Q signal was quite following the LSC_OS_LF_mag_Hz, so a change of pi/2 should have sufficed to bring the signal to _I; unfortunately for low pole frequencies the LSC_OS_LF_mag_Hz signal is not precise, and moreover we found at the end of the shift that the demodulation phase is not stored in the DAQ yet, so the tuning should be performed again; as a starting point one could set -0.77, which what we found during our tests (overall sign/pi to be checked); we added LSC_DARM_line_phi0 to the DAQ as a 1 Hz channel, but we did not restart LSC_Acl_Moni yet, as we did not want to unlock again as at the beginning of the shift; the process was restarted at the end of the activity;
• during this last, long lock we witnessed four glitches on DARM (20:19:37 UTC, 20:43:47 UTC, 21:17:33 UTC and 21:37:23 UTC); they were relatively harmless as they did not unlock us, I wonder if they are of the same nature of the ones witnessed on Saturday when they were instead too much to handle for the LowNoise1 actuators.

At the end of the shift, to conclude the lock, we tested again the LowNoise1 configuration:

• at 23:34:35 UTC we engaged the boost filters for both DARM and SRCL;
• at 23:37:45 UTC we unlocked during the ramp of the LOWN_ENBL commands, due to a ~ 15 Hz oscillation on DARM (FIgure 3).

We left the arms locked on the IR.

This afternoon was dedicated to ISC, Diego and Piernicola worked the whole shift without any major problem; all the activities are still in progress...


other activities:

-SQZ: alignment recovery, work carried out inside the detection lab from from 17:00UTC;
-ENV: check of TCS CO2 bench environmental sensors from 15:10UTC to 15:30UTC;

DAQ
FDS_Img stopped and restarted at 21:00UTC because of missing image from SQB1 cameras.

DET
DET_MAIN node in the SHUTTER DEAD state because of missing data, restored in SHUTTER_CLOSED at 18:15UTC.

Task of the shift was to further study the robustness of the lock at CARM zero. Two main tasks were performed:

1 - test the procedure developed in the past shift in order to minimize the optical spring and to reach a good alignment working point;
2 - implement an improved signal to monitor the behaviour of the optical spring wrt to the alignment conditions.

Despite a couple of locks during which we unlocked during the engage of the SSFS (8.29.02 UTC, 15.02.29 UTC) and one during the handoff of CARM to MC (9.49 UTC), we obtain the first relevant lock at CARM null 3f at 10.07.10 UTC. The action performed during this lock are the following:

10.08.14 UTC: engage of the SSFS and handoff of the longitudinal loops to the 1f signals;

This time we didn't close immediately the dither loops, but we first performed the alignment procedure in order to minimize the fringe, with the final aim to engage them with zero offsets, once we obtained an optical spring close to zero.
Thus, we started to inject noise on DARM at 10.10 UTC to monitor the DARM TF with respect to the usual alignment procedure to reach the target values (DARM phase at 100 HZ = 0.8 rad, pole of TF at 14 Hz). Such procedure produced the expected effect of reducing the optical spring. In details, between 10.12 UTC and 11.10 UTC we moved the SR in the TX (+) and TY (-) directions of an overall misalignment of respectively 2 urad and 0.6 rad. Moreover, we adjusted the longitudinal offset of SRCL up to a final value of 6.5. During these movements, we were iteratively adjusting the BS in TX (-) in order to minimize the fringe.

N.B. Since at 10.34 we seemed to have reached a good alignment working point (in terms of optical spring), we closed the dither loops with 0 offsets, but after a while (11.39) we added small offsets and proceeded with the aligment.

At 11.36 we performed a first test of closing the BS AA loop, exploiting the B8_DC signal demodulated with the BS dithering signals. The loop stayed closed for few minutes but at 11.39.48 UTC the ITF unlocked due to big flashes on B1p_DC.

Other than monitoring the DARM TF, we were looking, as a figure of merit, at the OS monitoring signal calibrated in Hz(LSC_OS_LF_mag_Hz) which showed as a global trend, the expected behaviour of going towards 0, as we were minimizing the spring. However such signal, since it is a magnitude of the DARM signal, has no sign and also, below 6 Hz doesn't give any meaningfull information (see 54585).

For these reasons, for the rest of the shift we worked on a new signal, which is created from the demodulation of B7_DC with the DARM line at 74.4 Hz. The channel name is LSC_NArm_B7_OS_I and has been implemented on Acl. This new signal has not been tested yet, but it will during the afternoon shift.

We left the interferometer ready for the enxt ISC activities.

In the tuning of the ITF working point in CARM null we found out that moving the SR_MAR_TX set point affects the position of the DARM pole at high frequency  (see elog 54460) . The adjustment of the ITF set point is done injecting noise in DARM. Thus we are studing how to extract information on the DARM high frequency pole frequency using two lines injected in DARM.

The first line is the same line used to measure the DARM UGF at 74.4 Hz. An additional line was added at 304.4 Hz the amplitude of this line is 4e-6.

We cmputed the magnitude of each line in the DARM oltf and we did the ratio of them. We changed the frequency of the DARM high frequency pole from 100Hz to 400 Hz and we plot:

• the magnitude of the 74.4Hz line (up left plot)
• the magnitude of the 304.4Hz line (up right plot)
• the ratio between the magnitude of the two lines (bottom left plot)
• the square of the ratio between the magnitude of the two lines multuplied by a factor (bottom right plot)

The signal in the bottom right plot seems to scale linearly with the pole frequency thus is a good candidate as monitor channel for DARM high frequency pole. The channel was implemented in Acl Last week (elog 54516). The name of this signal is LSC_DARM_HF_pole_mag_sqrt (ratio between the two lines)    DARM_HF_pole_mag (square of the ratio between the two lines).

In the shift of last Monday a channel for the monitor of the optical spring pole is added. The channel compute the transfer function between LSC_DARM and SNEB_B7_DC at the frequency of the DARM UGF Line i.e. 74.4 Hz. The name of the channel is LSC_OS_LF_mag.

After that this channel was added in the data flow we observed the following things:

• The LSC_OS_LF_mag as expected senses different kind of adjustment of the ITF working point (see fig 1)
• During the scan of SR_TX angular alignment this signal remained constant as expected (see fig 2.) (see elog 54460) 
• During the scan of SR_TY this channels changed as expected  (see fig 3.) (see elog 54460)
• At the end of the working point tuning we did a longitudinal scan of the SR longitudinal loop in order to have the oprtical spring pole around zero and we profit of this scan for the calibration of the channel

Figure 4 shows:

• the LSC_OS_LF_mag channel during the SRCL_SET scan that decreased from 40 to 6 (up left polt),
• the LSC_OS_LF_mag channel afrer the calibration (up right plot)
• the LSC_OS_LF_phi is the phase monitor of the channel but it is too noisy (bottom left plot)
• the SRCL_SET channel (bottom right plot)

With each points of SRCL set we measured the transfer function of DARM (fig 5) and we fitted the OS pole frequency:

• SRCL SET = 0.0       pole freq =   14.6 Hz
• SRCL SET = 1.0       pole freq =  13.7 Hz
• SRCL SET = 2.0       pole freq =  12.5 Hz
• SRCL SET = 4.0       pole freq =  10.1 Hz
• SRCL SET = 6.0       pole freq =   7.8  Hz
• SRCL SET = 6.5       pole freq =    6.3 Hz

we had some difficulties to fit the last two poles. We checked also the transfer function between LSC_DARM_ERR and SNEB_B7_DC and we saw that it scales with the set point of SRCL and that in the last two points the coherence is lower than with the other poitns (see fig 6).

For each SRCL_SET we averaged the LSC_OS_LF_mag and LSC_OS_LF_phi channels and we saw (figure 7)

• The LSC_OS_LF_phi behaves randomly with the SRCL set thus it is not trustable (secnod row)
• The LSC_OS_LF_mag scales linearly with the SRCL_SET except the first point (first row)
• The sqrt(LSC_OS_LF_mag) scale linearly with the frequency of the Optical spring pole with the following law sqrt(LSC_OS_LF_mag) = m*freq_pole with m=0.40+/-0.03

We did the same exercise using the SNEB_B7_DC demodulated at the frequency of the DARM line i.e. 74.4 Hz. The results are shown in fig 8 and 9. The name of this channel is LSC_NArm_OS_mag. The linear coefficent is 0.00120+/-0.00005

We did a comparison plot (fig 10) between the LSC_NArm_OS_mag and LSC_OS_LF_mag converted in Hz and we realized that the signal extracted without passing to the TF with DARM is cleaner with higher OS pole frequency.

Conslusion:

We added in the data the channel LSC_OS_LF_mag_Hz that is the LSC_OS_LF_mag signal calinrated in Hz (frequency of the pole of the optical spring).  This signal has 2 issues:

1. it is signless because it is the magnitude of the DARM line in SNEB_B7_DC signal and the phi channel is not trustable
2. it is can not sense optical springs below 6 Hz

We are working on another way to extract this signal. Further details will follow