This morning only limited activity could be done, due to the issues reported in Entry 51391; during the shift we managed to check and/or recover the main locking sequences:

• relock the cavities on the IR beam and the check the ALS alignment, then its lock with the VCO;
• unlock the cavities from the IR and do a full arms lock acquisition with the ALS system: independent arm lock with the reallocation, move to the CARM/DARM basis, switch to the beatnote signals, scan of the IR peak and selection of the IR resonance peak (Figure 1);
• realign and relock, although briefly, the DRMI in standalone with the PR alignment loop in Drift Control; while doing this activity we could test a fix for the bug of the SUS Automation nodes.

The activity will continue in following shifts.

In Entry 50929 it was reported a bug which affects only the SUS_PR node; we remind:

• all the SUS nodes share the same code base;
• a little bit of code is actually executed only by SUS_PR, and it basically checks if the PR safety tripped; in case, it changes the node state to MISALIGNED (to follow the optic, which already misaligned itself) and notifies it to the user.

This piece of code caused a bug related to ConfigParser (see attached logs), which complains about a non-iterable numpy.float32 and/or a missing section in the configuration file (which obviously exists); this is most probably the same behaviour already observed in the OMC_LOCK node (Entry 45476 and comments) and it is related to the implicit conversion done by ConfigParser between float and string; the first symptom is the bug itself, the second one is a consequence, as the error can make the configuration file to be badly closed (and resulting empty in the process).

Actions done:

• changed all the float(config.get('SECTION','parameter')) calls to config.getfloat('SECTION','parameter');
• forced the string type when using the config.set() call (which is more robust than remembering to explicitly call config.read() every time before any config.get).

SUS_PR can now be asked safely the SAVE_ALIGNED state again.

The PR and SR nodes have already been reloaded with the new code base, while the other nodes can be reloaded whenever there is a small window (they are not affected by the bug anyway).

The first part of the shift was dedicated to recover the problem of WEB - ADC7674 boards: Temperature too high; because of this all the suspension top stage boards received corrupted data at different times.

We decided to open all the suspensions loops and to investigated the problem of WEB ADC; I went at WE building and I found the fans of the crate 66 not working. The faulty crate has been changed with a spare one (see 51389).

After the replacement we could restore the communication from the WE; Paolo started the recovery of the susoensions.

At 9:48 UTC the cavities were locked; at 11:30 UTC we realized the SDB2 was controllable.

I contacted the SBE on call at 11:30 UTC that started the investigation from remote; the problem at SDB2 was finally fixed at 13:00 UTC and the ISC commissioning activity could restart.

DAQ
this morning we experienced troubles with the name server; the problem seems related to the many instances of DataDisplay.

Electricity
At 8:51 UTC the TB UPS1 switched off; I contacted Massimo and the UPS was switched on again. The reason of the switch off could be related to the high temperature in the room produced by a broken fan (see attached plot).

Electricity
(12-04-2021 08:45 - 12-04-2021 09:00) On site
Status: Ended
Description: TB UPS1 off

SBE
(12-04-2021 11:30 - 12-04-2021 13:00) From remote
Status: Ended
Description: SDB2 not controllable

SUSP
(12-04-2021 06:00 - 12-04-2021 09:48) From remote
Status: Ended
Description: recovery of suspensions

The GPS time of all the activities can be found in the attached file.

Fans of the crate 66 were found not working, explaining the high temperature of the crate. This crate is hosting two ADC7674 for ENV and one MuxDemux_V1 for the suspension.

Thanks to Francesco, the crate has been replaced by the crate 37.

Almost all the suspension top stage boards have received corrupted data at different times, preventing the correct working of the control loops.

ITF State:many troubles to understand and to be recovered.
ITF Mode: Troubleshooting
Quick Summary: IMC unlocked, SDB1 loops open, BS loops open, NE loops open, PR loops open, WI loops open, SDB2 open,  WEB - ADC7674 boards: Temperature too high.
Activities ongoing since this morning: No activity is known so far in the control room.

In the EQB1_HD_AA process we added a channel that will be used to measure the squeezing on the homodyne detector. The channel name is SQZ_EQB1_HD_DIFF_AUDIO_BPF_rms_10Hz and is obtained:

• filtering the homodyne detector audio channel with a band pass filter between 2 and 3 kHz
• computing the rms of this channel each 0.1 second

Fig 1 shows an example of this channel during a phase scan of the coherent control loop

On Thursday and Friday we worked at the optimization of the coherent control loop. The goal of the shift was to minimize the phase noise and improve the long term stability of the coerent control loop.

Structure of the loop:

The coherent control loop is closed with two actuators. A fast phase shifter PZT  (PI S-303.CD ) placed on the mirror LO_M1 and a slower phase shifter (PI S-325.30L) placed on the mirror LO_M4 (HD_CC_COARSE loop). The error signal of the Fast CC loop is the SQZ_EQB1_HD_DIFF_RF_4MHz_I channel (i.e. the RF Difference channel of the homodyne detector demodulated at the Coherent control laser frequency offset), whereas for the HD_CC_COARSE loop the error signal can be selected between the correction signal of the Fast Coherent control (SQZ_EQB1_HD_CC_Corr_50kHz) or the Error signal of the fast Coherent control loop ( SQZ_EQB1_HD_DIFF_RF_4MHz_I). We can select between the two error signal by using the relay SQZ_EQB1_HD_CC_COARSE_ERR_SELECT (0 =  4_MHz_I, 1=CC_Corr). We added also a protection for the COARSE CC loop: if the correction of the Fast CC loop is lower than-2V or higher than 2 V the Coarse CC Loop has to be open. This protection is implemented with the relay HD_CC_COARSE_TRIG.

During these two shifts we optimized both the loop and in particular we use the COARSE CC LOOP to keep the correction of the FAST CC LOOP around zero.

Transfer functions measurement:

We performed a noise injection in both the loops. Both the loops were simultaneously closed. The loop filter used for the fast CC Loop was a pure integrator and a double pole at 20kHz with Q=1. The gain of the loop was 15000 at 1Hz. The loop filter used for the COARSE CC loop was a pure integrator and a double pole at 500 Hz Q=0.7. The gain of the loop was 0.05 at 1 Hz

• 8th April at 11:59:00 UTC we took five minutes of clean data with both the loop closed
• 8th April at 12:14:00 UTC we performed a noise injection on the Fast CC loop with a white noise (amplitude 50uV). Figure 1: OLG of Fast CC Loop
• 8th April at 14:24:00 UTC we performed a noise injection on the Coarse CC loop (amplitude 25mV). Figure 2: OLG of COARSE CC Loop

Proposal of new filters:

We tried to improve both the loops. In particular we added a boost at low frequency to the COARSE CC LOOP, we removed the integrator in the FAST CC LOOP, we increased its bandwith and we added a boost berween 0.011Hz and 50 Hz to the Fast CC Loop.

We saved the new controller for the Fast CC Loop in the filter configuration file with the name "CC_flt_boost". The gain is 150=50x3 @ 1 Hz: where 50 is a comoensation factor and 3 is due to the fact that we have increased the gain by a factor 3 respect to the previous filter.

ACL_FILTER_SET        "CC_flt_boost"    1        150.0        1.0    20
ACL_FILTER_POLES    "CC_flt_boost"    0.1        1.0
ACL_FILTER_ZEROS    "CC_flt_boost"    50.0    0.0
ACL_FILTER_ZEROS    "CC_flt_boost"    5000.0    0.7
ACL_FILTER_POLES    "CC_flt_boost"    20000    1

The new filter for the HD COARSE LOOP is saved with the name  "HD_COARSE_CC_BOOST_FLT" 

ACL_FILTER_SET        "HD_COARSE_CC_BOOST_FLT"     1             1     1        20    
ACL_FILTER_POLES    "HD_COARSE_CC_BOOST_FLT"     0            0
ACL_FILTER_POLES    "HD_COARSE_CC_BOOST_FLT"     0.001        0
ACL_FILTER_ZEROS    "HD_COARSE_CC_BOOST_FLT"     6.0        1.0
ACL_FILTER_POLES    "HD_COARSE_CC_BOOST_FLT"     1000        0.5
 

Figure 3 shows the comparison between the new and the old loop filter for the HD CC Coarse. The new proposal has an UGF around 2 Hz

Figure 4 shows the comparison between the new and the old loop filter for the HD CC Fast. The new proposal has an UGF around 600 Hz

Figure 5 shows the FFT of the Slow CC loop error signal with the old filter and the projection with the new filter

Figure 6 shows the FFT of the FastCC loop error signal with the old filter and the projection with the new filter. We expect a 12% reduction of the pahse noise with the new filter

Test of new filters

All the measurements are taken with the Cpherent control demodulation phase equal to 1.75, i.e. all the CC signal was in the Q Quadrature. Moreover all the measurement are taken after a switch off of the HD AA dither lines.

• 9th April at 11:55:00 UTC 2 mins of data with the old filter (Fast CC Loop G=15000 Coarse CC Loop G=0.05)
• 9th April at 11:57:30 UTC 2 mins of data with the new Fas CC Loop Filter G=15000 and with the old Coarse CC Loop filter G=0.05
• 9th April at 12:07:00 UTC 2 mins of data with the new Fast CC Loop Filter G=15000 and with the new Coarse CC Loop Filter G=3
• 9th April at 12:11:00 UTC 2 mins of data with the new Fast CC Loop Filter G=45000 and with the new Coarse CC Loop Filter G=3
• 9th April at 12:19:00 UTC 2 mins with both the loops open
• 9th April at 12:25:30 UTC 2 mins of data with both the loops open and a Phase scan with freq = 3 Hz and amplitude equal to 1 V with the Fast CC PZT. These data are used for the error signal calibration Figure 7
• 9th April at 14:54:00 UTC 2 mins of data with both the loop open and the IR Shutter closed (we measured the sensing noise)

Figure 8 shows the residual phase noise of the coherent control loop in 4 different configurations. Blue: with the Old filters (16.4mrad of residual phase noise), Red with the new filter only in the Fast CC Loop (13.44 mrad of residual phase noise), Cyan with both the new filters (17.42 mrad of residual phase noise), Black with both the new filters but G-45000 for the fas CC Loop (6.58 mrad of residual phase noise).

Observing the cyan spectrum we noticed that increasing the slow CC loop caused a loose of gain in the fast cc loop, thus we compensated this by increasing the gain of the fast CC Loop of a factor 3 (black curve). We keep this as final configuration for the CC Loop

Figure 9: is the comparison between the phase noise with the Coherent control loop closed with the new filters (black), with the CC Loop Open Blue  (phase noise 139 mrad) and the homodyne sensing noise red (phase noise contribution 0.87 mrad)

From Figure 9 is visible that the high CC Loop gain below 10 Hz is useless because we are limited by the sensing noise. By comparing the Blue and the black curves we can observe that between 500Hz and 5000 Hz we are slightly reintroducing phase noise.

Strategy used to engage the new CC Loop filters

1. Use the SQZ_EQB1_HD_CC_CORR signal as error signal for the COARSE CC LOOP. This means set HD_CC_COARSE_ERR_SELECT = 1
2. Close the Coarse CC Loop with gain = 3. This means set HD_CC_COARSE_GAIN = 3
3. Select the old filter for the fast CC Loop (Filter 0) and reset it
4. Close the Fast Coherent control loop with Gain = 15000 
5. Select the new fitler for the fast CC Loop (Filter 1) before reset it
6. Increase the gain by a factor 3. I.e. Fast CC Loop Gain = 45000

When you want to open the loop open before the phast CC Loop and after the Coarse CC Loop. 

We will write soon a python function to open/close the coherent control loop and we will implement soon the automation for the HD detection

At 11h30UTC the WEB ADC7674 boards temperature started to increase to reach more 110 C .One hour later their timing_error signal started to oscillate (see the plots).

It s probably due to the VME crate fan who stopped running . This VME crate host the MxDX_v1 used to collect the WE suspension data and for the GIPC

The shift was dedicated to the planned OptChar Activity, FP cavities characterization, carried out by Guo, Capocasa, and Degallaix remotely.
To allow an initial measurement, I manually unlocked the Mode Cleaner from 8:30 UTC to 8:33 UTC, the system went back online without further adjustment. Also, as preliminary action, the Flip Mirrors dedicated to ALS at the Terminal Buildings were moved Up for the whole shift.
The following actions were focused on the lock of the infrared cavities. With only one set of mirrors aligned at times, we relock the arms, unlock them manually and misalign the End Mirrors, waiting after every action a specific time. Also, from around 12:00 UTC to 12:30 UTC, we set the ITF with only the NI aligned and moved the mirror to maximize the power on B4 as a reference.
The Activity was concluded at 13:40 UTC, infrared cavities left locked.

Some additional information about the possible reconstruction of BS angular signals from B1p quadrants:

- Signals are present both on QD1 and QD2, but QD2 looks a bit better;

- the recontruction is possible using both 6MHz and 56 MHz; in fig 1 the reconstruction of BS_TY is done multiplying B1p_QD2_6MHz by 100, and B1p_QD2_56MHz by 350 (the demodulation phase is far from being optimal)

- there is also some signal for SR alignment; in fig 2 we can see a reconstruction of SR_TX obtained combining 6MHz and 56MHz --- [SR_TX=(Q6-Q56*2)*4000].

ITF State: Cavities locked on the infrared.
ITF Mode: Commissioning.
Quick Summary: all SBE loops closed, all Suspensions Loops closed, DET UTA in "portata nominale", Central Building Clean Rooms AHU is OFF for testing purpose.
Activities ongoing since this morning: No activity is known so far in the control room.

After the ISC shift, I finished the recovery of the green lock on the North arm, since this morning there was some DAQ problem.

I re-aligned the green beam using the galvos in order to recover good alignment in reflection, and so I was able to close the BPC in this good position. The new values are:

• TILT X SET: -0.165
• TILT Y SET: -0.3
• SHIFT X SET: -0.485
• SHIFT Y SET: -1.37

The demodulation phase seemed fine so I tried to lock, and it worked without problems. I left the ITF with the two arms locked in the IR.

Today's shift was dedicated on working on the DRMI configuration and 3f new characterization.

At the beginning of the shift an alignment was performed, setting the new setpoints for SR and PR:

16.41.00 UTC new set for PR:
TX = -44
TY = 0.7

16.45.00 UTC new set for SR:
Tx=63.7
Ty= -239.09

16.51.00 UTC: we reached a good alignment and we were able to lock the DRMI on 1f.

After several unlocks, we managed to obtain a quite stable lock configuration, which allowed us to perform some noise line injection in order to characterize the DRMI loops: 

17.22.30 UTC: MICH inj line @31.1 Hz, ampl. 0.02 --> measured MICH UGF = 17.5 Hz
17.26.00 UTC: PRCL inj line @67.1 Hz, ampl. 4e-4 --> measured PRCL UGF = 10.5 Hz
                         SRCL inj line @33.3 Hz, ampl. 0.07 --> measured SRCL UGF = 1.5 Hz
              
Since the low values on the expected monitored UGF values, we worked on the tuning of the gain of the three loops in order to increase their values:

17.36.00 UTC: MICH GAIN moved to 4.5 -> measured UGF = 21 Hz
17.38.50 UTC: PRCL GAIN moved to 0.28 -> measured UGF = 22 Hz
17.30.00 UTC: SRCL GAIN was set to 0.3 from the beginning-> measured UGF = 1.5 Hz

Note that here we can see a lot of variation of the UGF for MICH and PRCL for fixed loop gains. See figure 1 ( maybe the OLTF  need to be remeasured again for this TCS new working point  ).

18.33.00 UTC WE  locked on 1f but we observed a lot of oscillations on MICH,SRCL, PRCL and corr signals.
We tried to fine adjust the gains of MICH and PRCL, but we kept getting loop oscillations.

Then at 18.45.39 UTC we profit of a more stable lock and we measure the ratio of 1f/3f for PRCL, confirming also that the phase of B2 demodulated at 18MHz was right. Therefore we kept its demodulation phase on 5.28 rad and the corresponding ratio is 6.8e-3 (positive sign). See figure 2.

We tried to continue with lines injected either in MICH or SRCL, to characterize 3f signals for these DoF, but the DRMI kept unlocking quite often, showing the presence of high frequency noise and low frequency noise wich made it a lot unstable. This issue spoiled the rest of the shift since we were not able to obtain a stable lock anymore.

At the end of the shift we put the DRMI automation in DOWN state, we parked PR & SR, and we locked the arms on IR.

After the short intervention to solve the problem whit the B2 PD1 signals (#51372) the scheduled activity on DRMI lock/alignment started; the work went on without any major problem and it is still in progress...

Air Conditioning
central building clean rooms AHU machine switched OFF at 14:25UTC for testing purpose (Dattilo).

The ISC team activity focused on supporting the Central Heating tuning activity.

The central interferometer was aligned and locked at 7.40 after tuning of the SRCL gain. The UGF monitor lines have been turned on with the following frequencies and amplitudes:

• MICH LINE: Freq 31.1 Hz   Amplitude : 0.02 
• PRCL LINE: Freq 67.1 Hz   Amplitude: 2e-4
• SRCL LINE: Freq 33.3 Hz   Amplitude: 0.07

At the beggining of the shift, the 11.6 Hz PR vertical oscillation was excited, but an intervention was able to damp it with success (see logentry  51368)

At 7.42 UTC the TCS tuning activity started. Due to the change in the recycling gain of the sidebands during the TCS transients (See figure 1), the GAIN og the PRCL, MICH, and of the PR_TY AA loop required to be reduced several times in order to keep the loops stable. The SRCL UGF remained instead stable (except for a change at 9.16 UTC due to a realignment of the BS) and did not require any gain tuning during the NI CH transients. The gains were changed as follows :

Additionally, the B1p quadrants galvo loops have been enabled during the shift and the quadrant signals have been checked to be good error signals for the alignment of the BS (see figure 2)

At the beginning of the shift, the CH settings were:

• P_WI = 41 mW

• P_NI = 15 mW

Identified by the pink star labelled “Start” in figure 1. Figure 1 shows the gain of the 56 MHz sideband gain for the CH powers. Thus, according to the simulation, increasing the NI power, while keeping the WI constant, will improve the gain.

Reminder: when the steady state has been reached after each step in the CH power, ISC team checked the alignment.

After step 8, we observed a decrease of the sidebands’ absolute values and hence,  at 13.40:50 UTC, we came back to the value set at step 6 (see figure 2).

We left the CH in the following configuration:

• P_WI = 41 mW

• P_NI = 56 mW

To be sure to be sitting on the maximum of the sidebands power, the equivalent scan for the WI CH should be done.

The shift was dedicated to the planned ISC activity, CITF TCS tuning, carried out by Casanueva, Valentini, and the TCS Team from remote. 
During the shift in parallel, the Squeezing team worked from remote on Homodyne Coherent Control Improvement & Auto-Alignment.
ISC activity still in progress.

DAQ
This morning the rtpc22, related to ALS NEB, was not receiving anymore the related Tolm packets (see entry #51363). Following Masserot and Letendre instructions at around 6:10 UTC I went first in DAQ Room to unplug and replug the optical receiver of the Mux/Demux, and then later in Computing Room to repeat the same operation on the Transceiver of the rtpc22 (see entry #51365).

DET
From 7:24 UTC to 7:28 UTC - Moved SDB1 in safe position.

This morning the instability at 11.6 Hz decided to come out again, so we had the opportunity to test the solution. A small portion of PR MIR correction has been sent to the marionette vertical actuation, passing through a resonant filter centerend at 11.6 Hz. Some coefficients, from -0.1 to -0.2 have been tested. A slow decreasing trend has been observed soon, but the higher value did not result to be better than the lower, because we started to see some 'beating', which was considered a bad thing. The final coefficient has been set at -0.17.

As we can see in the attached plot, the slow decreasing of the resonance went on constantly, up to the total disappearing. The engagement of the TX automatic alignment full bandwidth did not change the trend. Even if the tuning of the driving could be not perfect, the unstable parasitic loop should have been removed.