We supported the TCS team during the thermal actuators' adjustments.

After the common RH step and the differential DAS step, we noticed that the ITF was jumping between two metastable states (Fig.1). The ~3 minutes period of these jumps made us think of a correlation with the diagonalizaton routine. We waited for the ITF to jump in the "good" state, the one with the darker fringe and the higher sidebands' power, and disabled the servo to verify the hipothesis. The jumps stopped, giving comfirmation that diagonalization was indeed driving them, but it is unclear what exactly the servo was chasing when it made these changes.

After the common step of RH the interferometer's tuning gradually improved (Fig.2), although the superposition of the thermal transients of the first RH step, the DAS step and of this last RH step makes it complicated to pinpoint the exact cause of this improvement.

During this transient the ITF unlocked due to a WI glitch (Fig.3) and we reduced the power manually to allow the INJ_MAIN node to recover the IMC. This was harder than yesterday: between the movement of the half-lambda plate and the alignment of the bench, the reflected power of the IMC is close to the critical value (0.68 W - 0.79 W, Fig.4) and we had to increase the transmitted power a little to clear this threshold. The INJ servo then adjusted the rotator setpoint without issues.

After the next lock we increased the power again to keep observing the thermal transient, but the lock was killed 15 minutes after the acquisition by another WI glitch (Fig.5)

We have began the development of the power increase automation in carm_null_1f, for the moment offline, to be deployed later. For the moment, ITF_LOCK.py is unmodified and executes as usual.

We leave the ITF with the arms locked.

At the end of the activity  today we moved the etalon loops setpoints up one full fringe, with a ramp of two days:

• NI = 20.207 -> 20.464 degC;
• WI = 19.6 -> 19.857 degC.

As stated at the daily meeting, the automation would have received some modifications during the afternoon. and the actual deployment would have been decided by the people in shift onsite at the end of the activity.

Despite that, and despite the fact that ITF_LOCK doesn't need a reload when the .ini file gets modified, this happened, from offsite, without any communication to the control room and the people in shift. Luckily, most of my changes were not saved yet, but this is not the way to go.

All changes have been reverted, and any kind of development has been aborted for the day, on purpose.

At around 14.00 UTC, some jumps started to appear in the main ITF signals (see fig.1). Assuming that the jumps could be related to some higher-order mode, Michal suggested to change the ETM RH settings to move the cavity away from this critical region. 

Before starting the tuning of the ETM RHs, the configuration was:

NE: V = 18.3 V (9.02 W)
RoC_NE = 1669.6 m

WE: V = 19.1 V (9.80 W)
RoC_WE = 1670.5 m

A pure differential change was therefore applied, with the goal of changing the RoCs by:

RoC_NE: -2 m
RoC_WE: +2 m

At 15:09 UTC, the RH settings were changed to:

NE: 9.02 W (18.3 V) → 9.82 W (19.1 V)

WE: 9.00 W (18.3 V)

Later, Piernicola and Mattia realized that the jumps were actually related to the diagonalization. After disabling it, the jumps disappeared).

At 16:25 UTC, the ITF unlocked due to a glitch.

From this point onward, the ISC team worked on the lock acquisition automation.

The effect of the step will need to be evaluated during the next locks.

This afternoon we have updated the following thresholds on the detection system, in accordance with the increase of the input power from 12 to 18 W:

• we updated the thresholds on SDB2_Photodiodes triggering the fast shutter for B1p_PD1 and B1p_PD2 (B1s was already changed yesterday). All thresholds multiplied by a factor 3/2 in the configuration:

ACL_RELAY_CH    B1p_PD1_DC_saturation        au    SAMP_FREQ    0    2    -2    B1p_PD1_DC    0.82 0.58 >=   #0.55 0.39 #0.82 0.58 #1.65 1.166 #3.4    2.4  #2 0.7
ACL_RELAY_CH    B1p_PD2_DC_saturation        au    SAMP_FREQ    0    2    -2    B1p_PD2_DC    1.17 0.29    >= #0.78 0.19 #1.17 0.29 #2.35 0.588 #4 1
ACL_RELAY_CH    B1s_PD1_DC_saturation        au    SAMP_FREQ    0    2    -2    B1s_PD1_DC    45    22    >=   #30 #45    22 #35 20 #45 22 #90    45 #43 12

• Thresholds in ITF_LOCK.ini for B1p_PD1 have been increased accordingly:
  • in the section SHUTTER: b1p_pd1_thr_lo = 0.3 ; b1p_pd1_thr_hi = 0.5 ; b1p_dc_thr = 0.04
  • in the section ITF_LOCK: b1p_pd1_thr_lo = 1.5 ; b1p_pd1_thr_hi = 3.0

We used the external configuration of SDB2_Photodiodes in order to apply the new threshold on B1p_PD1 (3.0 instead of 2.5) while the ITF was locked in CARM_NULL_1F.

• Thresholds in DET_MAIN.ini have also been increased by the same factor, namely:

max_power_B1p = 0.04, max_power_B1p_PD1 = 0.44, max_power_B1s = 45 (unchanged with respect to yesterday)

Both DET_MAIN and ITF_LOCK were reloaded at 16h34 utc after an ITF unlock.

First, we implemented the lock acquisition procedure performed yesterday evening: it worked reliably again this morning.

At 07:23 UTC, Manuel gradually increased the injected power up to 17.5 W (INJ_IMC_TRA_DC) in CARM_NULL_1F. The ITF unlocked at 07:53 UTC due to a suspension glitch similar to previous events.

We reacquired CARM_NULL_1F at around 08:11 UTC with an injected power of 17.7 W, closer to the nominal 18 W. The ITF unlocked at 08:27 UTC, as MICH and DARM drifted away without a clear trigger.

Finally, at 08:43 UTC, CARM_NULL_1F was acquired again at 17.8 W of injected power.

At this point, we started the thermal tuning; the initial configuration is reported in 69838.

The plan is to explore the effect of the WI DAS IN power and to perform a differential step. We agreed to double the NI DAS IN injected power and evaluate its effect on the ITF.

08:56 UTC STEP 1
WI DAS IN +100%

The SBs responded quite rapidly going down and the dark fringe worsened, indicating that increasing the WI DAS IN power is not the correct direction. So we agreed to close the flip mirror of the WI INner DAS.

09:38 UTC STEP 2
WI DAS IN set to 0 (flip mirror closed)

See Fig. 1 for ITF signals corresponding to these two steps.


Since the DAS differential steps did not seem to have a significant impact, Michal suggested performing a differential ETM RH step. In particular, we decreased the WE RoC by acting on the RH, as the strong PA on the WI is expected to induce additional curvature

10.31 UTC STEP 3
WE RH RoC reduced by 4 m, that is V = 19.1 V → 20.6 V and P = 9.8 W → 11.4 W

This last step rapidly increased the dark fringe brightness (see Fig. 2), so the step was reversed and then continued in the opposite direction of the same amount.

11.03 UTC STEP 4
WE RH RoC increased by 8 m, that is V = 20.6 V → 17.5 V and P = 11.4 W → 8.2 W

The effect of the last step of WE RH seemed good: B1p decreased up to 0.02 mW (around 12.40 UTC), before restarting to rise up (see fig 3).

13:29 UTC STEP 5
NI DAS IN and OUT +10%
WI DAS OUT +10%

The effect of this last step is ongoing.

The configuration of the CO2 actuators, so far, is summarized in the table below:

ITF found in LOCKED_ARMS_IR in COMMISSIONING mode.
At 06:00 UTC I locked ITF up to CARM_NULL_1F at first attempt.

Then the planed activity on "TCS tuning / CARM NULL locking (Pinto, Corubolo, Lumaca)" started. Still in progress...

Following the increase in input beam power, I repeated the analysis of the PAs on the ITMs.

Figures 1 and 2 show the NI maps, while Figures 3 and 4 show the WI maps.

The new analysis confirms the PAs identified in #69827 for both mirrors. The higher input power also allows for a slightly more precise estimation of their thermal signatures compared to the previous analysis.

For the NI, the high-temperature region observed at the center of the mirror appears to have a smaller impact in terms of ΔT with a value below 0.2 °C compared to approximately 0.3 °C in the previous analysis. The stronger PAs, instead, are consistent with the previous analysis in both position and estimated ΔT.

For the WI, the results are overall consistent with the analysis performed at lower input power. The estimated ΔT values of the two main PAs are slightly higher, at approximately 0.9 °C and 0.46 °C for PA1 and PA2, respectively. In addition, at least four other weaker PAs can be identified, with estimated ΔT values of around 0.2 °C, consistent with the features already observed in the previous analysis.

Fig 1 shows that NI and WI inertial control of IP are going through a bad time. The shown period is just the lates 5 days, but the situation has been like this for a long time. This situation is happening more and more often in general indifference, although it is quite simple to spot: the IP rotational dof is quite insensitive to external actions, so a sudden excess motion (up to 100 times more the normal rms) is almost for sure due to a control problem. The direct reason is known: a fast glitch on the accelerometers; a deeper explanation and a permanent solution has never been proposed.

I made a fast quantification of the impact of this situation on locking stability. In the latest 5 days one can count 34 lock losses from locking index larger than 45 (lock acquisition started). Among them, 7 are due to  NI glitches; 3 are due to WI glitches: 30% of the total unlocks. Fig 2 and fig 3 show the dynamic of the unlocks: the rotational event is obviously associated to a large longitudinal motion, which lead the locking correction to saturation. To be noticed that the large IP rotation is never due to the unlock: if NI moves, ITF unlocks and no effect is visible on WI. Viceversa, in WI type unlocks, NI remains quiet.  

Today, we performed two DAS steps to induce differential RoC variations between the ITMs. Specifically, the DAS powers were adjusted to produce an estimated RoC change of +1.5 m on one ITM and -1.5 m on the other, and vice versa.

The initial configuration of the CO2 actuators remains identical to yesterday evening's and is detailed in the table below:

When we arrived in the Control Room, the ITF had been locked at CARM_NULL_1F since 06:50 UTC. Manuel adjusted the alignment around 07:15 UTC, after which we proceeded with the DAS steps as follows:

07:45 UTC STEP 5
NI DAS IN and OUT -4%
WI DAS OUT +4% (DAS IN OFF) 

Just after this step, Manuel adjusted the lines in 69835.

12:16 UTC STEP 6
NI DAS IN and OUT +8%
WI DAS OUT -8%

Before the Daily Meeting, as about an hour was still available, Paolo proposed re-exploring the effect of the beam position on the ITMs via PR alignment. Paolo, Manuel, and Piernicola used the remaining time for this test (a dedicated log entry will follow). We intentionally chose not to restore the TCS settings immediately, as the system was still reasonably close to optimal conditions and another TCS change would have triggered a long thermal transient. 

None of the RoC differential steps yielded significant positive effects (see Fig. 1), so we agreed to restore the initial configuration of the CO2 actuators, but we performed the step just after a test on the SHUTTER_OPEN had been performed (see ISC entry for details, 69839).

14:06 UTC STEP 7 (equivalent to the resulting configuration after the STEP 4).
NI DAS IN and OUT -4%
WI DAS OUT +4%

At 14.10 UTC, the ITF unlocked due to a glitch on one of the ETMs. The CARM_NULL_1F state was acquired again at 14:42 UTC.


Between 15:25 UTC and 15:40 UTC, we began increasing the injected power in the following steps: 11.2 W → 12.6 W → 16.05 W (a total variation of approximately 4.85 W).

Since the power variation was larger than intended, the stability of the lock rapidly degraded, requiring prompt adjustment of the CO2 compensation.

We roughly estimated the power variation in the arms to be approximately +37% and performed a common DAS step of the same magnitude; furthermore, we opened the WI DAS IN flip mirror (at the minimum power value).

15:55 UTC STEP 8
NI DAS IN and OUT +37%
WI DAS OUT +37%
WI DAS IN open at minimum power

The compensation was good enough in CITF, and the SB levels were high enough to attempt going further. We therefore proceeded with the lock acquisition as reported in the ISC entry (69839) and managed to recover CARM_NULL_1F at the higher injected power around 18:45 UTC.

The final TCS power configuration is reported in the following table:

In order to support the TCS tuning of DAS for the final injection power of 18 W we changed online the transmission of the IMC from 11 W to 18 W while keeping a lock in CARM_NULL_1F.

The commands used are:

• to increase the input power:
  • cm_send('EIBAgilisRot', 'MOVEREL', 1, 1, -2000, 50)
• to decrease the input power:
  • cm_send('EIBAgilisRot', 'MOVEREL', 1, 1, 2000, 50)
     

Where the '2000' is the step size not a setpoint. Note that the steps have some hysteresis, and coming back usually requires more step than going up in power.

Overall the DAS tuning is positive, with a well tuned common component that grants adequate sideband power but a less than optimal differential tuning that leaves a bright fringe of ~0.035 mW, against the 0.011 mW observed during O4.

The process of changing the power midlock was surprisingly easy, which gave us the idea of attempting a lock acquisition at low power. After tuning the CH to compensate for the new setpoint of the DAS, and increase again the power at CARM_NULL_1F, we modified the automation to close the flip mirrors of the CH at CARM_NULL_1F (before was at carm step 2/3), re-engaged the final working point with 11 W and then increased the power in steps to 17.3 W, starting when the flip mirrors close to roughly replace the CH with the YAG at the same time.

Against all expecations this worked very well, with the very appreciated advantage of avoiding the recovery of the whole lock acquisition and to ease the further TCS tuning.

This procedure can be stably adopted and be automated at least for the interferometer tuning phase. 

Tomorrow this process will be automated, substituting the manual steps of the rotator on SIB1 with a servo loop.

We leave the ITF with arms locked, after restoring the transmission of the IMC to 11 W.

The origin of the uncorrelated trend observed yesterday in the arm cavity powers has been found to be the consequence of the input power trend.

The figure 1 shows the comparison of B8 DC and the PSTAB_power * a + o, showing a good superposition (for B8>300mW).

We increased the B1s threshold triggering the fast shutter to 45 mW, which is slithgly below the safety threshold of the phododiode (~50 mW).

The change was applied to DET_MAIN.ini and to SDB2_Photodiodes. The external configuration of SDB2_Photodiodes was reloaded at 15h16m18-UTC.

ITF found in COMMISSIONING Mode and LOCKED_ARMS_IR State.
All times are UTC.
06:52 ITF in CARM_NULL_1F.
07:00 Begin of TCS tuning / CARM NULL locking activity (Pinto, Corubolo, Lumaca).
07:13 NI Y offset added (Pinto).
13:59 SDB1 slow shutter has been open.

ITF left in COMMISSIONING Mode and SHUTTER_OPEN State.
 

These modulation are caused by the high amplitude of the DIFFp tx and ty lines used for the UGF servos. The choice of these amplitudes is due to the 'necessity' seen in the past of surviving the initial transient of CARM null in which diffp gains change a lot. Reducing of a factor 2 the amplitudes improves the fluctations on the fringe, keeping enough coherence for the servos, fig.1.

We can simply decide to reduce the amplitude in the automation once we reach the steady state (or checking if we still need such high values of the lines amplitude in the initial transient).

Figure 1 and 2 zoom on the lines at 6.6Hz and 10.4Hz. They are extremely narrow (less than 1mHz), and at a round number of decihertz, so I don't believe to be a mechanical resonance, and they must be some kind of intentional dither or modulation.

ITF found locked at CARM_NULL_1F in COMMISSIONING mode.
Activity on TCS tuning / CARM NULL locking (Bersanetti Mantovani, Corubolo, Lumaca) in progress.

It went on till 19:40 UTC.

ITF left locked in LOCKED_ARMS_IR.

Today we performed various DAS common steps in terms of variation of ITM RoC (changing the ITM RoCs of equal amounts) to increase the SB power in CARM_NULL_1F. The initial configuration of CO2 actuators is described by the following table:

08:24 UTC STEP 1 

NI DAS IN and OUT +12%

WI DAS OUT +6% (DAS IN OFF)

10:30 UTC STEP 2 

NI DAS IN and OUT +12%

WI DAS OUT +8%

Both steps led to a clear increase in SB power, whereas other signals (such as arm power or B1p) degraded linearly, seemingly unrelated to the DAS steps (see Fi.g. 1).

Following these two steps, at approximately 12:10 UTC, after more than 5 hours in CARM NULL, the ITF unlocked due to a glitch on WI. 

The subsequent lock was achieved relatively quickly. By around 12:25 UTC, the ITF was back in CARM_NULL_1F.

Another unlock occurred around 14:12 UTC due to a second glitch, this time on NI, although it was less clear than the previous one. 

We agreed to perform an automatic lock acquisition, letting the ITF realign by itself; however, this took too long (entry 69831). On the second attempt, manual realignment was performed, but the SB power in CITF were low. We then agreed to also compensate for the two morning DAS steps on the CHs.

15:13 UTC STEP 3

NI CH +17%

WI CH +25%

This step increased the SB powers in CITF and facilitated the CARM_NULL_1F acquisition (CARM_NULL_1F around 15:53 UTC).

Then, we noticed that the steps 1 and 2 didn’t induce a pure common RoC variation, as our OPL simulation was using outdated intensity maps. However, the overall RoC differential component was small compared to the RoC common one.

After implementing the updated intensity maps, we performed an additional DAS step, restoring a pure common RoC variation (around 8 m of total variation of RoC, considering all the three DAS steps). 

16:32 UTC STEP 4 

NI DAS IN and OUT +4%

WI DAS OUT +6%

This final step further increased the SB power, though by a smaller amount. 

Figure 2 and 3 illustrates the overall behavior of the signals throughout the day:

• SB power increased: 12 MHz increased from 0.03 mW to 0.041 mW; 112 MHz increased from 0.02 mW to 0.026 mW.

• Arm power decreased: following a trend not directly correlated with the DAS steps.

• Dark fringe power increased: becoming brighter in a manner not directly correlated with the DAS steps.

• U-LSB unbalance: the unbalance at the start of CARM NULL progressively grew over time (before slowly recovering balance during CARM NULL).

• DCP increased: the DCP initially followed the first two steps of DAS power, improving its value.

The origin of the overall trend (decreasing arm power and increasing dark fringe power) remains unclear.

Below, the table summarizes the final configuration of the CO2 actuators.

A consequence of working with the decentered beam in CARM Null (3mm in NI_Y direction, for now done by hand) is that most of the alignment needs to follow and re-adjust.

One smaller consequence is the fact that arm cavities and injection bench are slightly more misaligned once we unlock and try the next acquisition. This is not critical in terms of statistical success of the relock of the arms, but the dithering control at that stage may need more time than usual to converge, so we increased the timer in LOCKING_ARMS_IR from 60 s to 120 s.

One larger consequence is the PR mirror resulting misaligned in TX when the next DRMI lock acquisition happens (Figure): we made one test to leave the acquisition alone and see how much time the automatic macro needs before we can see triggers again, and this is around 13 minutes, clearly too long for an unattended lock acquistion.

We are devising some logic that acknowledges this drift due to the beam misalignment and remembers it at the next relock so that we can compensate it before we acquire the lock of the DRMI again.

B1p is currently flucutation between 200mW and 300mW with fast fluctuations (timescales much shorter than 1 second).

Figure 1. Looking at the spectrum there are 3 lines which are dominating it: at 6.6Hz, 8.5Hz and 10.4Hz. The high line at 8.5Hz makes sense, it is the NI TX line, and we are using the NI vertical drift control to mis-center the beam on the input mirrors. However for the other two I have no explanation.  The lines look quite narrow, so could be some kind of dither lines, or some resonances.

These lines are close to limiting to reach safety threshold that prevent opening the OMC slow shutter.

This morning I performed again the analysis after the shift of the input beam of the both WI and NI mirror images from the thermocamera (previous analysis can be found in #69785 and #69808, respectively). Here below are summarized the results:

WEST INPUT

Fig. 1 shows the thermocamera image of the WI. 

Fig. 2 shows the output of the matlab code showing a new PA (or at least more visible now) w.r.t the last analysis after the beam shift.

Here there is a table summarizing the positions and information of the PAs:

NORTH INPUT

Fig. 3 shows the thermocamera image of the NI. 

Fig. 4 shows the output of the MATLAB analysis, highlighting a PA that was not present in the images analyzed previously. The figure also shows a concentration of higher-temperature area near the center, with some peaks reaching approximately 0.3 °C.

Here there is a table summarizing the positions and information of the PAs:

Note: 

The thermal contrast associated with a Point Absorber is obtained from the differential temperature map between the interferometer Lockedn (in CARM_NULL) and DOWN states. Measurements performed on different days can therefore exhibit some variability, even when the same Point Absorber is analysed.

This variability can arise from several contributions. The thermal background observed on the mirror is not completely uniform and can contain spatial structures related to thermal radiation reflected by the HR surface. These structures can change with the environmental and thermal conditions of the interferometer. In addition, the absolute temperature of the mirror and its surroundings, the thermal state and stability of the infrared camera, the transmission and optical properties of the ZnSe viewport, and small changes in the camera–mirror geometry can affect the measured temperature distribution.

The differential analysis reduces the effect of global temperature variations by aligning the median temperature of the mirror in the LOCKED and DOWN states. However, it does not completely remove changes in the spatial structure of the thermal background. Furthermore, the Point Absorber temperature is currently estimated from the maximum value within a local search region, making the result sensitive to local fluctuations and residual registration errors between the two images.

Consequently, variations of the measured Point Absorber thermal contrast, for example from approximately 0.65 °C to 0.8 °C (in the WI PA 1 case) in measurements performed on different days, should be interpreted primarily as a combination of the intrinsic thermal signal and the reproducibility/uncertainty of the measurement procedure.

ITF found in COMMISSIONING Mode and CARM_NULL_1F State with Input beam mode matching activity ongoing.
All times are UTC.
16:10 - 16:46 INJ: realignment of the beam with the BPC FF quadrant (Gosselin, Pagabbe, #69825).
17:06 ITF relocked to CARM_NULL_1F at first attempt with no actions needed by the operator.
17:55 North Arm put back in the alignment position of yesterday (#69820).
17:57 INJ: End of activity on INJ to ITF mismatch measurement (#69824).
ITF kept CARM_NULL_1F up to the end of the shift. At 20:55 I manually unlocked the ITF and set it to LOCKED_ARMS_IR.
ITF left in COMMISSIONING Mode and LOCKED_ARMS_IR State.

Guard tours:
- 20:35 

This afternoon we performed a measurement of the current input beam mode mismatch toward the arms. 
Following the usual procedure of the kick of the end mirrors and using the matlab code detailed in #68646

An additional idea was to do a measurement with CH ON and OFF to see if there is any difference the difference. 

First measurement : 13:29:13 UTC CH switch off. 
Unfortunately injection unlocked when we put the ITF in down. 
Analyze could start only at 13:31:00 UTC 

Measured mismatch  for WA : 2.5 %

The data from NA where not usable, there was quite a lot of misalignment.
We realized that it was most probably due to the work done yesterday evening.  
Paolo brought back North arm to its nominal condition. 

Second meausrement : Carm_null lasted only 6 min. 

Third measurement : CH ON, ITF unlocked after 18 min of carm null but we managed to do the kicks anyway. 
Analyze started at 14:57:10 UTC over 200 s

Measured mismatch WA : 1.9 % not that clean.... maybe due to some misalignment when ITF got unlocked.
Measured mismatch NA : 2.8 %

Fourth measurement : CH ON, best measurement so far (figures) 
Analyze starts at 15:45:30 UTC over 200 s

Measured mismatch WA : 2.1 %
Measured mismatch NA : 2.8 % 

Giving the relatively low mismatch and the work to come on TCS we decided not to tune it. 

At the end of the shift (17:55 UTC), Manuel brought the NA back to the alignment found yesterday evening.