- measurement of the mismatch with the hot ITF, opening the dampers, putting in down and giving a kick on the end mirrors z corr (starting from 14:09:25 UTC): we found 1,5% for the west and 3,8 % for the north (but with more uncertainty), see plots 1 and 2
- scan on the green, starting from 14:21:42 UTC: 1,7% for the west and 3,2% for the north (see plot 2). For the scan around the 02 mode we were on the edge of the ramp and for the next time we should change the offset of the ramp.
The goal of the shift was to characterize the ITF working point with the following RH powers:
| RH powers | NE | WE |
| Current settings | 6.9 W [V=16.04 V] | 8.4 W [V=17.67 V] |
The actions performed during the shift are listed below:
13.08 UTC: ITF relocked in LN3
14.18 UTC: ITF manually unlocked to perform input beam mode matching measurement with the free swinging technique and green laser scan (performed by Suzanne, Camilla and Matthieu).
See detailes in 64204
15.03 UTC: ITF relocked in CARM NULL 1F
16.08 UTC: OMC scan started (performed by Romain and Andrea)
16.30 UTC: OMC scan ended
16.52 UTC: ITF relocked in LN2
After 1.5 hr in LN2, at 18.20 UTC, Andrea will proceed with the lock acquisition.
The main ITF signals during this afternoon are shown in figure 1.
This afternoon we performed some measurements to verify the mismatch towards the arms. The last measurement done is reported in #63374.
First, we performed the free swinging method, giving a kick on the end mirrors (with the dampers opened). The GPS of the kick (we gave the kick simultaneously for the West and North arms) is 14:17:00 UTC.
However, when analysing the data we could not find reasonable values of mismatch. We need to verify if there is any problem with the Matlab code to debug, or there was something wrong with the measurements. In Fig. 1 we can see the time window of the free swinging mirrors. To be noted that the west arm starts swinging after roughly 15 seconds with respect to the North Arm. On Fig.2 we can see the surposed flashs of B7 and B8.
Then, we perform the measurements with the Green Scan. The GPS in this case 14:30:43 UTC (Fig 3). Figs 4 and 5 show the scans of the TEM 00 mode and TEM 02 mode respectively. In this case, for the north arm we found 3.3% and 1.4% for the west arm, as shown in Fig. 6.
We did not manage to analyze the results of the free swinging mirrors with the kick measurement made yesterday (probably because they were not excited enough).
For this reason this afternoon we performed again this measurement, but we first opened the dampers, we manually put in down the ITF and then we applied the corrections (more than once, to keep the mirror excited).
The first kick was done at 15:51:36 UTC, and the data were analyzed 3 minutes later taking the peaks during 220s (so it cannot be defined "hot" itf measurement).
We obtained 2% mismatch for the north and 1,6% for the west (see plots)
As Piernicola confirmed, the sidebands at 46.7 Hz and 53.3 Hz are the beating between 50 Hz and DIFFp_TX line at 3.3 Hz (purple curve).
Indeed, when the DIFFp_TX line was set to 4.8 Hz, there were two sidebands at 45.2 Hz and 54.8 Hz, respectively (blue curve).
After the earthquake in Japan, we relocked at the first trial and after we arrived LN3_SQZ we started to tune the noise injections for the PSTAB loop (both at LF and HF). After the amplitude of the noise was tuned, we added the corresponding functions into the "injection_SSFS.py" script.
We then proceeded to repeat by hand the two last injections of the SSFS loop, that last shift made the ITF unlock. We decreased slightly their amplitude, while checking that the coherence with Hrec was still high. Now the script launches: CLEAN SSFS_UGF PSTAB_LF PSTAB_HF and then the 4 other SSFS injections at lower frequencies (down to 100 Hz). The operator launched the script and it worked without any problem. Figure 1 shows the injection corresponding to SSFS_UGF, and I think the current UGF is safe enough (I wouldn't push much higher, to ensure robustness, unless the noise injections suggest that it is limiting).
At this point we passed to the DIFFp topic. We started by opening the DIFFp_TX servo loop (which zeros the offset!), and then we started to put back the offset. However this didn't zeroed the new signals and so we continued to increase the offset. During this scan we tried to tune the phase, slightly, even if it was already quite good. Then we repeated the same operation with the TY loop. Figure 2 shows the two scans that we did. For the TY the phases were not clear so we left the deafult ones.
The phases we put are:DIFFp_TX_DCP_HF_mag_B1_phi0 = 0.5; DIFFp_TX_DCP_mag_B1_phi0 =0.5
After the intervention of last Monday, the DIFFp_TX_SET servo was trying to build a signal with a line that did not exist anymore.
We changed the frequency of the demodulation to 3.3 Hz in ASC_pre and the frequency of the high-pass cut-off in DIFFp_alignment to 3.3 Hz as well. We also tuned the demodulation phase of DIFFp_TX_B1p to -1.8.
After that the offset manually added during debugging (63916) was removed and the servo changed it back to a value that minimizes B1p, as it is supposed to.
We reloaded the filters in ASC_pre to make this solution permanent. The new value of the demodulation phase is saved in ASC_pre but the config file has not been reloaded to avoid the risk of an unlock. Since Metatron does not change it, the value we set online should remain until there is a chance of reloading ASC_pre safely.
It should work as intended now (the servo brings the setpoint to a value that corresponds to the minimum of the "ball" that appear on the istogram with SDB2_B1p_DC and ASC_DIFFp_TX_INPUT, Figs.1,2), but we will keep monitoring it over the next few days.
The DIFFp_TX_SET servo is indeed not working.
For the moment, we disengaged the servo and set a better sepoint by hand, since the servo was slowly but constantly drifting away from the minimum of the fringe.
The state of the ITF was changed to ADJUSTING for this intervention and it's now back to SCIENCE.
During the recovery of last Saturday, one of the unlocks was caused by DIFFp_TX's gain becoming too low overtime. The coherence of the UGF monitor line was too low and the servo was not functional. As a patch, the ISC crew increased considerably the lines' amplitude.
As a result, the lines became dominating on B1_PD1/2_Audio rms (63876) and caused them to occasionally saturate introducing loud glitches in the sensitivity. The issue concerns DIFFp_TX line.
Today we tried to find an amplitude that allowed the UGF servo to work while avoiding the saturation on B1 audio channels, but we observed that there was no possible amplitude that satified both conditions.
Instead, we elected to move the line at lower frequency (4.8 -> 3.3 Hz) and recalibrate the target UGF (2.7 -> 3.6). The result is positive: audio channels remain within 0.025 mW instead of
0.04 mW and the line coherence is good (Fig.1).
Moving the UGF servo line allowed Michal to lower SDB1's dithering line as well, since it is no longer "competing" with DIFFp's line.
During our tests, we noticed that the noise floor on the audio channels spike when the filters of MICH and SRCL are swapped between LN2 and LN3. We tested a lock acquisition without this filter swap (we changed the UGF servos' target as well to remain consistent with the control filters), and then introduced the new filters manually one at a time. The noise level on B1 audio channels did not increase as a result, what we observed must have been due to something else.
After this test, we leave the ITF in the final state.
After solving the issue of the marionetta of the NI, the ITF started unlocking at acquire LN3. We noticed very fast unlocks, due to the fast shutter closing very soon. No clear loop seemed guilty, but we noticed that the DIFFp TX UGF servo was not working because its line didn't have coherence. However the DIFFp TY gain was increasing quite strongly during the transition. So we increased the gain of the loop to keep the UGF stable, and it seemed to work. We also noticed that the UGF servos were off once in science, even if the lines were ON, so we commented the commands that turn them off.
We left the ITFlocked in LN3.
The locking troubles were due to an instability at 1.8 Hz of NI marionette reallocation. The problem appears when the correction on the mirror is moved from NI to NE, and at the same time the splitting filters with higher crossover are engaged. Such a problem has never been observed before, but in principle an instability in that region cannot be excluded. There are resonances of the two suspensions which can have slightly different frequency and can create unstable crossing points. Why this problem should appear now, after years of correct work, I don't know. Some measurement is required in order to see if the problem can be explained by a loop model, and a more robust strategy can be found. For the moment, the gain on the filter on MA branch has been changed from the nominal one (0.416) to 0.38. The instability is no more there, but clearly is not very far. Changing more the gain could create different problems, so let's leave the things like that untill a good model will be available.
To be noted that we managed to improve the alignment only using the PZT mounted on M19.
The PZT mounted on M20 had a strange behavior (red circle on figure attached). We were doing steps of 5V that were slightly changing the transmission and but then this transmission was going back to its original value by its own. Nothing was visible on the QPD.
On the March 12th, a shaker has been installed on the Laser Bench to perform sismic noise injection. After those studies done on the 14th, the electronics that controls the shaker (deployed on the ground) has been left ON (with 0V of input voltage) to have the possibility to further perform other remote injections.
However, entering in the laser lab few days later, we noticed a slightly higher acoustic noise than usual. We decided then to switch OFF that electronics, and indeed the noise reduced.
Looking at the environmental sensors present in laser lab, we found this noise visible as both high frequency broad noise (>200Hz) and with some lines a low frequency (~45Hz and harmonics).
Moreover, we see it on acoustic (see fig. 1-3, 5) and sismic (fig. 4) sensors. While the noise is visible on both benches, the effect (seismic and acoustic) on EIB is really faint (see fig 4-5).
Nevertheless, we didn't find any noticeble effect on Hrec.
We checked some signals of interest after the realignment of the RFC and could not see any clear improvment neither in LN3 (figure1) nor in INJ stand alone (figure 2).
RFC alignment in LN3
We moved the setpoints of the piezo from around 17h45 UTC to 18h10. We improved the alignment as shown in fig.1
(We also updated the thresholds in the DMS).
SBE EIB noise injections in LN3 (sine @ 0.5Hz)
| DOF | gps start UTC | duration | amplitude pp |
| tx | 18:22:45 | 3 min | 10 urad |
| ty | 18:26:50 | 3 min | 10 urad |
| tz | 18:31:10 | 3 min | 10 urad |
| x | 18:34:40 | 3 min | 10 um |
| y | 18:38:08 | 3 min | 10 um |
| z | 18:41:39 | 3 min | 10 um |
Analysis of the noise injections will follow.
We checked some signals of interest after the realignment of the RFC and could not see any clear improvment neither in LN3 (figure1) nor in INJ stand alone (figure 2).
To be noted that we managed to improve the alignment only using the PZT mounted on M19.
The PZT mounted on M20 had a strange behavior (red circle on figure attached). We were doing steps of 5V that were slightly changing the transmission and but then this transmission was going back to its original value by its own. Nothing was visible on the QPD.
After the maintenance, we have had mainly two types of unlocks: some at CARM_MC_IR, which we didn't pay much attention since the seismic and wind activity are high. However, every time that we rach CARM NULL 1f we would unlock. We could see many "oscillations" on all the longitudinal loops, the PR angular loops and then saturation of B2 photodiode.
It took some time to understand the reason for these unlocks, but since they were all of them at the same moment Diego remembered that it was the moment when we enagge the CARM slow loop. Indeed it looked like it was oscillating at veyr low frequency, so we increased its gain and it worked (1/1 times, we need to see if this works in the long time). Figure 1 shows the clearest example.
WHile debugging I profited to tune the demodulation phase of MICH and SRCL on DRMI 3F, which was off by 0.2 only, then I checked it at STEP 3, and CARM NULL 3f and it looked well tuned. I also fine tuned the DARM phase in STEP3, which ws off by 0.3.
This doesn't explain the unlocks of yesteday on STEP 2 and STEP 3. We keep on investigating.
After the weekly check of the input beam position, the operator noticed that the we had an important vertical shift of the input beam.
Thus, I moved the PR X and Y by 60um in order to both recover the centered position and lower the BPC correction (see fig. 1).
This afternoon, at 15.24UTC we started the 2D etalon scan which will take 1 week long.
As previously indicated, we had to initially tune the phase for the WI scan.
The final values used are the following:
amplitude 0.4
duration 604800 s
sin phase -pi/4
All the INJ_rtpc (rtpc19) Acl servers have been started with the Acl v1r23p17 between 2024-03-27-17h24m10-UTC and 2024-03-27-17h25m06-UTC
Today we blocked the beam on the LB to allow HJ. Bulten to work on the control of EIB (details in another entry)
We took advantage of having INJ in down to remove the set up that we added few weeks ago with Walid to be able to inject laser phase noise. We were indeed suspecting that it could be the responsible for the loss of phase margin of the IMC loop that has been seen these last days.
When HJ finished, we try to relock the injection but it was not relocking. After multiple tests we eventually noticed that the BPC was locked on the reference values which were quite far from the ones it had before putting INJ in down.
The alignment of the whole system had indeed drifted away during the long lock of the night+day.to follow the ITF.
After having relocked we blocked the beam again to allow Alain to do the restart of rtpc19.
Then everyting relocked correctly.
We measured the OLTF of the IMC with the attenuation of 20 dB (10+10) we had an UGF of 100 kHz and 14 degrees of phase margin
We added few dB of attenuation 22 (20+2) and we had an UGF of 93 kHz with 16.5 degrees of phase margin (see figure).
It is better than what we had few days ago but still not at the level of what we add when we measured it on the 04/03.
All the Acl servers (AclISC, AclLC, AclSBE, AclFCoplev, Acl) are running with the v1r23p17 version since the 2024-03-26-15h20mn-UTC .
The main new features are
The first trial was done the 2024-03-25 starting at 14H-LT
All the INJ_rtpc (rtpc19) Acl servers have been started with the Acl v1r23p17 between 2024-03-27-17h24m10-UTC and 2024-03-27-17h25m06-UTC
PSTAB0 line (43Hz) has been switched off (amp=0) at 12.52.40UTC.
We could then relock in LN3: the bump at 43 disappeared while the broadband noise was still there.
To be noted that PSTAB1 line (236Hz) is still injected while, as far as I know, not used by any loop.
In order to understand why the IMC OLTF has changed since the 4th of march (UGF at 107kHz with a phase margin of 16deg, #63487), this morning we tuned the angular and longitudinal working points of the IMC loop (plot 1).
We remeasured the OLTF but it was the same as yesterday (84kHz UGF and 10deg phase margin, plot 2).
