For tomorrow, after the complete thermalization: check the DARM gain for DC readout, as the UGF was quite low (55 Hz), compensated by the servo in time.
ITF left in LOW_NOISE_2 with the autorelock failsafe engaged with the same target.
To recover LN2 there was not much to do on the DET side.
The SDB1 control with B5 work without any changes, it is likely that in single bounce the power is now too low on B5, and the threshold to enable the drift control in single bounce in the SDB1_LC process will need to be lowered so that it works in single bounce (in DRMI lock the power is orders of magnitude above the threshold in any case.)
The OMC temperature for locking on the carrier fundamental mode increased by ~60mK due to the change in master laser temperature set point. I have added step to the automatic ramp by hand, and then the automation saved that value in the Metatron ini file as the OMC lock was successful.
The OMC was quite misaligned during DC READOUT, once in LN2 the drift control realigned the bench by ~10urad, mostly in TY. I have adjusted the SDB1 B5 quadrant offset to 450um in H to get closer to the position that the drift control brings the bench to in LN2. This starts to be quite close to the edge of the quadrant, as now a larger fraction of the B5 beam is clipped by the diaphragm.
B1 PD3 has quite a large offset of -0.6uW. I have updated the offset in the SDB2 dbox bench process, but I haven't reloaded it as the interferometer continued the lock to LN2 despite the offset. To be done at the next unlock during normal working hours.
I have started the EDB OMC continous scan with the same parameters as in January. I expect the HOM will be shifted in temperature by ~60mK due to the ML temperature change, so the analysis will need to be adjusted. The EDB OMC alignment is quite bad, with 0.3mW instead of 0.6mW of 56MHz sideband TEM00 mode, and the order 1 mode of the 56MHz sideband has half the power of the TEM00.
7:00 UTC at the beginning of the shift the ITF was found in DOWN UPGRADING with LOCKED ARMS. SDB2 angular and position loops and SWEB position loop were open, I closed them.
7:00 UTC ITF set to MAINTENANCE:
- TCS chillers refill (Menzione)
- Cleaning of the central buildind and FCIM/FCEM (external firm)
- Nitrogen refill and vacuum maintenance (Vac & Mech group)
- Telescreel Left machine changed due to malfunctioning (Di Biase)
- Disconnection of accelerometer on SAS supply duct of NE, accelerometer and old sensors cabling on vac pipe removal in NE (#68650 Fiori, Lunghini, Quemener, Tringali)
- TCS Thermal Cameras acquisitions, TCS CO2 Power Checks (operator)
- WI timing adjustment (Boschi)
- MCH Tango server shut off (#68649 Gkaitatzis, Carbognani, Boschi)
- Restart NNF NNN and NSN after security stop (#68651 Van Hove)
- Icewm config updated to support both CL7 and AL9 graphical environment (#68653 Carbognani, Seder)
| CH (W) | OUTER (W) | INNER (W) | |
|---|---|---|---|
| WI | 0.318 | 0.02 | 1.27 |
| NI | 0.582 | 0.09 | 0.582 |
ITF recovery:
8:00 UTC opening of SR BS and CRYO DET valves (Pasqualetti)
11:00 UTC SR optical lever checks (Ruggi)
11:30 UTC SR GUARD closing (Ruggi)
12:00 UTC ITF in PREPARE SCIENCE
The realignment was successful and the lock acquisition went up to CARM_NULL_1F smoothly, though we noticed values of B1p aroun 0.1 mA. We also noticed that NE_RH was off and the process was stuck, it was quickly fixed by Nardecchia from remote (#68655). ITF had being kept in LOCKED_ARMS waiting for NE_RH thermal gradient.
16:22 UTC ITF in CARM_NULL_1F with B1p_DC value down to 0.01 mA
17:30 UTC at the end of the shift I left the ITF in LN2
To recover LN2 there was not much to do on the DET side.
The SDB1 control with B5 work without any changes, it is likely that in single bounce the power is now too low on B5, and the threshold to enable the drift control in single bounce in the SDB1_LC process will need to be lowered so that it works in single bounce (in DRMI lock the power is orders of magnitude above the threshold in any case.)
The OMC temperature for locking on the carrier fundamental mode increased by ~60mK due to the change in master laser temperature set point. I have added step to the automatic ramp by hand, and then the automation saved that value in the Metatron ini file as the OMC lock was successful.
The OMC was quite misaligned during DC READOUT, once in LN2 the drift control realigned the bench by ~10urad, mostly in TY. I have adjusted the SDB1 B5 quadrant offset to 450um in H to get closer to the position that the drift control brings the bench to in LN2. This starts to be quite close to the edge of the quadrant, as now a larger fraction of the B5 beam is clipped by the diaphragm.
B1 PD3 has quite a large offset of -0.6uW. I have updated the offset in the SDB2 dbox bench process, but I haven't reloaded it as the interferometer continued the lock to LN2 despite the offset. To be done at the next unlock during normal working hours.
I have started the EDB OMC continous scan with the same parameters as in January. I expect the HOM will be shifted in temperature by ~60mK due to the ML temperature change, so the analysis will need to be adjusted. The EDB OMC alignment is quite bad, with 0.3mW instead of 0.6mW of 56MHz sideband TEM00 mode, and the order 1 mode of the 56MHz sideband has half the power of the TEM00.
For tomorrow, after the complete thermalization: check the DARM gain for DC readout, as the UGF was quite low (55 Hz), compensated by the servo in time.
ITF left in LOW_NOISE_2 with the autorelock failsafe engaged with the same target.
Today during the maintenance I fastly checked the SC alignment in the OPA and I bit improved it. The total alignment is higher tha 98%. To further improve we should align the PD in DET lab but I did not manage from remote with only picomotors. The small peak in the second plot is the CC beam so it is not to be taken into account in the computation
In order to check the phase noise issue we tried to measure the HD detector CMRR as was done in 2022.
We inserted HD Mir and we measured the DIFF/SUM gain that is still 2.1. Then we injected 0.0125 V of amplitude of white noise in the fast CC and we measuired about 58dB of CMRR (fig 1 and 2)
In order to try to mitigate the phase noise we rotated the HWP of EQB1 in ASQZ as was donne in 2022 with the aim of removing all the peaks of mirror mounts resonances.
We performed other two CC phase scans (delay line)
- the first at 17:1:00 of 09/02
- the second at 18:54:30 of 09/02
In fig 3 and it is visible that the pahase noise is sligthly decreased from 68 to 62 mrad by rotating the HWP. This variation is inside the measurement error so we are nto completely sure about this improvement.
Fig 5 we also checked the behaviour during night to see if the laser is running in single mode and it seems the case
Fig 6 ASQZ,SQZ and shot noise spectra with delay line Shot 20:03, ASQZ 20:12, SQZ 20:23 UTC
Fig 7 a phase scan with the retroreflector and comparison with delay line. In the same condition. The phase scan was run at 15:45:30 UTC of Tuesday 10 February. The amount of losses increased by 4% between DL and retroreflector and the phase noise remained similar
Today Cecilia called me to report that the NE RH channels (voltage and current) appeared grey.
I checked and noticed that the NE RH had switched off on February 3rd and that, at the same time, the data disappeared from the DD.
I then restarted the process from the VPM and switched the NE RH back on at 14:59 UTC (see Fig. 1).
I do not know the cause of this issue.
In the previous days, I do not recall seeing the NE RH box in the TCS line on the DMS in red, although I am not completely sure (see Fig. 2).
In the frame of Alma Linux 9 OS Desktop Environment finalization the Icewm configuration (the contents of the /virgoData/icewm directory) has been updated in order to be able to support at the same time both CL7 and AL9 environments.
The main mechanism for supporting multiple environments is the elimination of hardcoded paths inside the Icewm config file in favor of using just the name of the application executable (dataDisplay, metamedm, etc.) to be made available via the PATH mechanism in platform specific target directories under /virgo/InstallArea
The most important applications such as dataDisplay or the Automation clients (metamedm) have been ported to AL9 and can now be started from the Icewm menus. Some python GUI applications are still under porting and will progressively be made available.
Anyone is welcome to log into the AL9 user machines (using the ctrln cluster in ThinLinc) and provide feedback.
During today maintenance, NNN, NNF and NSN passed security movement threshold and were stopped.
I restart them (rearm and enable motor) around 12:30 UTC. NSF phase required several close/open/close iteration for phase locking
This morning, during the maintenance, several ENV activities were carried out at NEB and CEB.
NEB activities
We reorganized the configuration of the accelerometers previously installed along the tower vacuum chamber for the mechanical modes measurement (#60441). Three sensors had been mounted (ENV_NE_TOP/MIDDLE/BOTTOM_ACC), with the latter two used in past investigations of the so-called “cactus” noise related to the vacuum system (#63540,63545,63647)
The top accelerometer was left untouched and remains installed. The middle accelerometer has been reinstalled, while the bottom accelerometer has been completely removed. In addtion, the cabling of the accelerometer previously installed on the safety structure (ENV_NE_SS_ACC) during the opening of the tower (cupola and virola temporarly dismounted, #60475) has also been removed.
Another planned activity was the removal of the accelerometer installed on the SAS air supply duct. This could not be completed using only a ladder; a mobile scaffold will be required to safely remove the sensor. The same intervention will be needed at WEB as well.
In the meantime, the sensor has been removed from the data acquisition process (ENV_NE_SAS_SUP_ACC). The corresponding action at WEB had already been performed last week (#60748)
CEB activities
Some reference measurements were taken for Gilles's magnetic-coupling simulation model at CEB (coil position and distances from surrounding structures). These measurements will be used to validate the model against the coupling observed during far-field magnetic injection campaigns.
Today we shut off the Tango MCH servers that were no longer in use. No issues have been observed following the shutdown, and all related services continue to operate normally.
ITT Found in UPGRADING Mode and DOWN State with arms locked on the IR.
All times are UTC.
7:09 Closed SWEB_SBE opened during the weekend.
7:20 Shelved Chrocc_SR DMS flags under TCS, SR Chrocc not used, shelve requested by Nardecchia.
9:32 SR RH Turned ON since the vacum level was fine (Nardecchia, Lumaca, and Pasqualetti, see #68645).
9:51 - 11:15 CEB Magnetometers measurements (Tringali, Fiori, and Quéméner).
10:03 Drillings at NEB (Romboli).
10:23 NI Etalon Set Point Change: NI_RH_SET=18.62, rampTime=259200 (Operator, requested by Mantovani).
10:24 WI Etalon Set Point Change: WI_RH_SET=18.73, rampTime=259200 (Operator, requested by Mantovani).
10:31 Guided visit to CEB.
13:34 Recovered both arms on the IR after Injection unable to lock, IB and BPC restored by Spinicelli, arms recovered moving PR X and Y to recover the alingment and then put them back to the old locked position (Operator, and Spinicelli).
ITT Left in UPGRADING Mode and DOWN State with arms locked on the IR.
The mismatch values we were getting with the MATLAB code we usually use to measure the mismatch from the kick were quite off.
The code had several small sources of uncertainty which, when combined, were giving a pretty wrong value with a large uncertainty. In particular, the results were very sensitive to a global offset that was manually applied to shift the whole dataset to zero.
I’ve cleaned up the code to make it more robust, especially the offset subtraction. The offset is now computed and subtracted locally (i.e. per FSR) instead of using a single global offset set by hand.
The updated code is here:
/users/optics/Commissioning/251029_MatchingArms/Mismatch_measurement.m
With this version, the results look much more consistent and the spread of the measured mismatch is clearly smaller. Updated values are shown in the table below, and the corresponding histograms are attached.
| steps from inital position | TZ_corr (V) | estimated meniscus lens from initial position (mm) | W MM % | N MM % | comment |
| 65k | -2.25 | -1 | 3.5 | 3.7 | after CH tuning |
| 65k | -2.25 | -1 | NA | NA | |
| 0 | -0.05 | 0 | 1.5 | 2.0 | |
| - 65k | 2.2 | 1 | 1.5 | 2.0 | 26/01 5pm |
| - 65k | 2.2 | 1 | 1.4 | 1.8 | 28/01 3pm |
| - 162.5k | 5.5 | 2.5 | 3.0 | 4.3 |
At the end of the shift, since the results were not fully conclusive, we decided to move the lens back to its initial position. However giving the new values of mismatch it would be worth exploring the meniscus lens positions between 0 and 1.
Following confirmation from Antonio that the vacuum level was OK, the SR RH was switched ON, at 09:32 UTC, at nominal value: 7 W (23.13 V).
ITF found in UPGRADING Mode and DOWN State with arms locked on the IR. High wind activity reported on site.
All times are UTC.
10:16 Recovered SQB1_SBE Vertical Position with motors and Closed Positional Loops.
10:51 Recovered SPRB_SBE Vertical Position with motors due to SBE_SPRB_ACT_F0V_raw_50Hz close to saturation (DMS flag red, see Fig 1.). Operation done without open Position Loops.
11:26 Recentered Sa_PR_F7_LVDT_V_50Hz DMS fag under PR_TestMass after "PR MAR actuation balancing" activity (see elog #68592). SUSP expert informed (Boschi).
13:17 - 13:26 Paoli L. went to CEB to bring material.
ITF left in UPGRADING Mode and DOWN State with arms locked on the IR. Wind calm.
We noticed, first in VIM spectrograms, that the infamous low frequency magnetic glitches seen by all magnetometers site-wide suddenly disappeared during last week-end.
Searching information we found out that railway traffic was suspended during those two days between Pisa and Empoli stations because of works along the line (also in the attached pdf):
Few observations:
We altready knew that these glitches are related to trains but...
- This finally proves that they are generated within the Pisa - Empoli railway section
- These glitches disappeared completely from CEB and NEB magnetometers, instead some are still seen in WEB magnetometers (Figure 4). This probably tell us that the WEB sees also glitches from a different railway line, the freight line that runs approx. North-South: see the map in Figure 5.
Spectrograms in Figures 6 to 9 show an insight of the time of the disappearing/reappearing:
- Figure 6, 7 and 8 show the NEB magnetometer (the closest to the Pisa-Empoli railway): the interpretation (Federico) is that first the train transit stops, than approx one hour later (/ one hour before) the electric line is de-electrified (/ re-electrified). A large glitch can be seen at these times: in Figures 7 and 8, respectively.
- Figure 9 is the WEB magnetometer, on the same color scale and time window as Figure 6: note that in this case the glitch density reduces just a bit.
Yesterday (February 5th, 16:00–18:30 LT) the NI PA heater was removed at the NI base tower.
To access the rear of the actuator, the actuator was rotated and the Optris/NI_cam support on the HR face was temporarily moved, taking references for proper repositioning (see Fig. 1).
Power supply and thermocouple connections were disconnected, and the heater was removed from the rear side of the setup and prepared to go to Roma (Fig. 2).
The PA configuration and camera support were restored (Fig. 3). Note: the NI_cam may require refocusing.
Motor connections were reattached (NI-ACT03a.001.25m, NI-ACT04a.001.25m, NI-ACT03a.002.25m, NI-ACT04a.002.25m) and the motor drivers were switched ON in the TCS room (Fig. 4).
At North End bld.new pump is running starting from 05/02 at 12.15 (local time) and at West End bld. starting from 06/02 at 15.00 (local time)
Looking at recent data when the INJ air conditioning was off (https://logbook.virgo-gw.eu/virgo/?r=68269) we noticed that the microphone on the laser bench measures some, not large but significant, excess acoustic noise with respect to the microphone in the room (_LLR_) and the microphone on EIB.
This noise shows up mostly between about 200Hz and 2 kHz. Most evident peaks are around 370Hz and 635Hz, which are very coherent with LB vibrations. There seems to be an excess noise also around 30-80Hz, but this is less coherent with the LB accelerometer.
It seems to us an indication for some acoustic-vibration source positioned on the LB. Since the HVAC is off, the source is unlikely related to air turbulence underneath the LB cover. One suspect might be the chiller water flow. To be investigated.
When the HVAC is on -- second figure -- this noise is less visible because partially covered, but two coherent bumps aroud 370Hz and 635Hz are still evident.
To be noted other two bumps present in all microphones, most evident when HVAC is off. They are at about 12 Hz and 18.5 Hz. These might be acoustic modes of the room.
4/02 Realignment of the squeezer
In the Morning we tried to align the PD photodiodes but we lost a bit the CMRR of the HD, and we also tuned a bit the EQB1 FI HWP1
Martina went in Det lab to exchange the discs for the dust monitoring so she also realigned the MZ and the OPA. We did not gain much in terms of MZ visibility Max ~1.6V. But we improved a bit the level of the coupling of the green beam into the OPA cavity.
5/02 Fine tuning of the alignment
As last work on the DL we aligned better its mirror, recovered the HD PD alignment and tuned again the HWP of EQB1 FI and the one in front of the HD Detector
We also restored the lock of all the PLLs
We investigated a bit a possible effect of phase noise
For all these operation we did a CC phase scan here a summary table
| Time | Param Gain | Efficiency | Phase Noise | SQZ Level | ASQZ Level | Notes | Fig |
| 02/02 18:47:44 | 0.3826 = 8.34 dB | 94.8(2)% | 92(2) mrad | 6.12 dB | 8.11 dB | Laser Multimode (Stand Alone) - After DL alignment | Fig 1 |
| 03/02 14:35:42 | 0.3788 = 8.4307 dB | 92.8(2)% | 77(2) mrad | 6.15 dB | 8.13 dB | Main PLL offset moved (no more multimode) | Fig 2 |
| 04/02 09:56:12 | 0.3601 = 8.87 dB | 91.9(2)% | 65(2) mrad | 6.40 dB | 8.54 dB | DL alignmen - t Stand alone - No multimode | Fig 3 |
| 04/02 18:04:36 | 0.3474 = 9.18 dB | 88.4(2)% | 20(2) mrad | 6.47 dB | 8.71 dB | Lost a bit HD alignment, OPA alignment No multimode | Fig 4 |
| 05/02 10:27:27 | 0.3523 = 9.06 dB | 92.1(2)% | 66(2) mrad | 6.48 dB | 8.73 dB | Improved HD alignment | Fig 5 |
| 05/02 13:56:06 | 0.3587 = 8.9 dB | 94.3(2) % | 85(1) mrad | 6.38 dB | 8.65 dB | No alignment, PSL laser freq Shift all PLL locked | Fig 6 |
| 05/02 15:48:44 | 0.402 = 7.91 dB | 94.1(2) % | 110(1) mrad | 5.53 dB | 7.34 dB | Less generated squeezing/ no alignment | Fig 7 |
Phase scans in similar configurations gives a different level of losses and phase noise. In general the losses are under control between 6 and 10 % whereas it seems we have an excess of phase noise. To be investigated.
The amount of losses is quite well explained (4% from visibility) 1% OPA escape efficiency 1% HD PD Quantum efficiency 1% EQB1 FI 1%HD clearence. Known losses 8%
If we do an average of the results above the level of losses is 7.4±0.8% and the phase noise is 74±11 mrad. In September 2022 (with the Pump power scan) we obtained 11.0 ± 0.4% of losses, 14±12 mrad of phase noise.
Next step is to remove the delay line
In order to check the phase noise issue we tried to measure the HD detector CMRR as was done in 2022.
We inserted HD Mir and we measured the DIFF/SUM gain that is still 2.1. Then we injected 0.0125 V of amplitude of white noise in the fast CC and we measuired about 58dB of CMRR (fig 1 and 2)
In order to try to mitigate the phase noise we rotated the HWP of EQB1 in ASQZ as was donne in 2022 with the aim of removing all the peaks of mirror mounts resonances.
We performed other two CC phase scans (delay line)
- the first at 17:1:00 of 09/02
- the second at 18:54:30 of 09/02
In fig 3 and it is visible that the pahase noise is sligthly decreased from 68 to 62 mrad by rotating the HWP. This variation is inside the measurement error so we are nto completely sure about this improvement.
Fig 5 we also checked the behaviour during night to see if the laser is running in single mode and it seems the case
Fig 6 ASQZ,SQZ and shot noise spectra with delay line Shot 20:03, ASQZ 20:12, SQZ 20:23 UTC
Fig 7 a phase scan with the retroreflector and comparison with delay line. In the same condition. The phase scan was run at 15:45:30 UTC of Tuesday 10 February. The amount of losses increased by 4% between DL and retroreflector and the phase noise remained similar
We removed the DL and we started to reflect sqz with SQB1 retroreflector. To optimize we did the following steps
- remove DL
- Close CC (Fast and corse loop)
- Close HD AA in ASQZ
- Move Retroreflector X and Y to remove clipping
- Move EQB1 HWP2 (HD) until we maximize the mag in SQZ
- Move SQB1 HWP2 (FI2) until we maximized the mag
- Move SQB1 (HWP1) (FI1) by reducing stray light bump on HD DIff
We performed a phase scan that should be analuzed at GPS: 145354929
Note that before aligning the retroreflector we tried to aligne the PD for SC into OPA AA but without managing. We centered as better es possible the PD by looking its aperture. Further investigations will be done
Here the new set point of Squeezing PLL
PLL MAIN: 29668
PLL CC: 29640
PLL SC: 36026
ITF found with the two cavities locked.
The SR tower evacuation is in progress, nothing to report on this side.
Other activities carried out during the day:
- Crystal temperature change of the ML;
- in the morning the HVAC team worked on the WE ACS system;
- Squeezing recovery, still in progress;
- at around 14:30 UTC Diana (TCS) started an activity in CB, still in progress.
After the change of the crystal temperature of the last summer, the main laser of the squeezer was too far from the Virgo ML (see here and there).
In agreement with Marco, we lowered the Virgo ML temperature by ~0.45⁰ to try improving squeezing system stability, which seemed to be stable in July. PSL/INJ system relocked quickly after the operation (a check of the OLTF of the IMC didn't spot any problem).
The locking of the squeezer laser is ongoing.
We will check in the following days the stability of the PSL/INJ to be sure the new T setpoint is stable enough.
Here the new set point of Squeezing PLL
PLL MAIN: 29668
PLL CC: 29640
PLL SC: 36026
Through analysis from geophones and LVDTs signals around SWEB, SNEB, SIB2, and SDB2, resonance frequencies lower than expected were identified.
In the case of SWEB, a first resonance mode was theoretically calculated at about 45,8 Hz. However, modes around 22 Hz and 26 Hz were identified by plotting transfert function between geophones and suspension ‘s LVDTs.
The reason for this discrepancy between theoretical calculations and experimental results is therefore questioned. The objective is, through an experimental vibration test using an impact hammer, to identify the resonance modes and understand their origin.
Test Principle:
The experimental vibration test using an impact hammer is a method used to characterize the dynamic behavior of a structure. It consists of exciting the structure using an instrumented hammer equipped with a force sensor. The vibratory response of the structure is measured using accelerometers placed on the structure under investigation.
The combined analysis of the impact force signal and the measured responses permit to determine the Frequency Response Functions (FRFs), from which modal parameters such as natural resonance frequencies, mode shapes, and damping coefficients are identified.
To increase the number of measurement points, it is possible to:
• always strike the same location and move the accelerometers
• fix the accelerometers in place and move the impact point (option chosen)
Experiment:
We have one impact hammer and three accelerometers available.
Installation of the equipment began on Monday, 02/02, at around 2:30 p.m.
The accelerometers were positioned respectively at the centre of each door and at the top of the dome.
A first series of tests was carried out in the afternoon to validate the setup and verify the consistency of the measurements.
Although this was not sufficient to obtain a good representation of the modal shapes, it was already possible to observe resonances around 22 Hz and 26 Hz, and then at 45 Hz.
The following day, Tuesday 03/02, we refined the model using parameters allowing better resolution at low frequency. Experimental set up had been configured to measure transfer function between accelerometers and instrumented hammer in 0 to 100 Hz range. (see attached parameters)
From 2:00 p.m., we started a new measurement campaign consisting of 110 impact points distributed as evenly as possible around the minitower. As some areas were inaccessible, certain desired impact points had to be removed or relocated.
The frequency response was estimated from the average of three impacts (at the same point and in the same direction).
This measurement campaign allowed us to confirm the resonance of SWEB at 22.6 Hz, 26 Hz, and then 45,2 Hz.
Thanks to the large number of impact points, we obtained a relatively good estimation of the modal shapes. (see attached charts)
At 22.6 Hz, the minitower appears to oscillate in an almost rigid-body manner along the x-axis.
At 26 Hz, the minitower oscillates this time along the z-axis (y-axis on our model).
At first glance, these look like pendulum modes (almost rigid-body OR hinge movement around fixation plates). Without y-axis measurement, it is not possible to define more precisely.
Observations and perspectives:
Clearances were observed between the feet of the minitower and the shims/ground. At the location of some bolts, this clearance reaches up to 1.8 mm.
This could support the hypothesis of inadequate anchoring to the ground.
However, on SDB2, clearances between the feet and the ground are also observed, while the resonance frequency remains higher (around 31 Hz). Two main differences between SWEB and SDB2 can be identified: the presence of vacuum pipes instead of two oblong flanges and the way the fixation plates are welded to the feet (discontinuous on one side compared to continuous on both sides). (see attached pictures)
To obtain a more precise understanding, adapting the numerical modal analysis is the first objective. If this is not sufficient, carrying out a new measurement campaign on SDB2 could be valuable. Finally, correctly tightening every SWEB screws before re-measuring the resonance frequencies appears to be enriching.
