The difference is the amplitude of the DARM line. I've tried to put back to 1e-7, see figure 1.

I've put in the automation 1e-7 to try (recalibrating the OS and DCP)

These changes have been reverted to do it in a proper shift

here is the estimated calibration coefficient to have the ASC_DOF_CORR spectra in urad.

These coefficients have been found to overlap the measured OLTFs with the theoretical ones.

ggDpTX=2.4300
ggDpTY=0.3704
ggPRTX=0.8547
ggPRTY=4.5100
ggBSTX=1.9000
ggBSTY=0.4545

here is reported the partial noise budget of the angular loops using the measurements of last friday jan 16th. Specific injections has been made in order to both measure the coupling and obtain information to calibrate the several error signals. full budget needs to be performed.

useful GPS:

gpsDpTY=utc2gps(2026,1,16,11,03,30);durDpTY=300;
gpsDpTX=utc2gps(2026,1,16,11,13,10);durDpTX=300;
gpsPRTX=utc2gps(2026,1,16,12,00,10);durPRTX=300;
gpsPRTY=utc2gps(2026,1,16,12,11,50);durPRTY=300;
gpsBSTX=utc2gps(2026,1,16,13,09,10);durBSTX=300;
gpsBSTY=utc2gps(2026,1,16,13,33,00);durBSTY=300; % bpass1 driving inj for Coupling
gpsBSTY2=utc2gps(2026,1,16,13,42,20);durBSTY2=120; % bpass driving inj Plant OLTF
gpsCL=utc2gps(2026,1,16,12,38,20);durCL=300; % clean data

here is the longitudinal noise budget of the measurements performed last friday jan 16th (fig.1).

given the current situation, maybe MICH subtraction filter needs a refinement (see fig.2), new filter will be computed and loaded.

list of useful GPS:

gpsMICH=1452624826;durMICH=120; %gps = 1452624826 (2026-01-16 18:53:28 UTC)
gpsSRCL=1452625143;durSRCL=120; %gps = 1452625143 (2026-01-16 18:58:45 UTC)
gpsPRCL=1452624984;durPRCL=120; %gps = 1452624984 (2026-01-16 18:56:06 UTC)
gpasDARM=1452625301;durDARM=120;
gpsCL=1452624515;durCL=120;% gps = 1452624515 (2026-01-16 18:48:17 UTC)

injections have been used to retune also the correct UGF setpoinf of SRCL loop, which now is slight overestimated (6.5 Hz instead of 4.5 Hz). The setpoint in arbitrary units will be reduced from 0.28 to 0.24 (fig.3).

The etalon loops have done several fringes since the restart of the loop. It is visible that:

- there is a MICH set and B1p increasing trend not correlated to etalon.

see plot 3, MICH set and B1p in time for locks >LN2. for MICH SET and B1p seems that there is an other very low frequency modulation. to be understood

- the best finesse asymmetry that can be reached is about 2.3e-3 (2*(B7-B8)/(B7+B8))

At 08:35 UTC, in the CITF configuration, we performed a differential step on the WI and NI CH powers.

The WI CH power was increased by 20%, while the NI CH power was decreased by 20% relative to their initial values.

The table below summarizes the current values.

- Starting from LN2 this morning we went in LN3 aligned to try the new 200s timer added to smooth the transition while changing the filters. It worked smoothly.

- Then in CITF the Central Heating has been tuned with a differential step (more details in TCS entry) improving both the B1p DC and sidebands (plot 1). We relocked in Carm null 1f and no bursts were present

- Finally, we locked  in the new LN3 aligned configuration and increased the SR TY gain from 6 to 10 (saved in the ini file) in order to increase the accuracy of the SR alignment control (plot 2). The improvement has to be better assessed during long locks

- From around 11h to 12h Manuel started to inject noise (see dedicated entry)

After the last movement of the DAS, we tested a few lock acquisitions to make sure that it was robust enough. The first thing we did was to reload the old offsets of the SDB1 control, to make sure that the OMC was aligned when the SR mirror was aligned. We tested few lock acquisitions (all of them successful until LN2), and then we focused on consolidating LN3_ALIGNED, which was systematically unlocking.

We tried to remove few actions, to try to understand what was the culprit and we realized that commenting the change of filter of MICH and SRCL solved the problem. So we focused on these two actions. We engaged the filter of MICH by hand, which seemed to work, and the ITF unlocked when we engaged the SRCL one, with a low frequency oscillation. Next trial we engaged automatically MICH filter but not SRCL one and it worked. And we engaged then by hand SRCL one with a higher UGF (from 0.24 to 0.28), and it remained stable. However we noticed that the new LOW_NOISE_3_ALIGNED didn't let many time for the UGF servos to work, so we increased the timer to two minutes, to allow some time to the servos to work. We didn't have time to test this, because we had a couple of not-related unlocks (I will take a look tomorrow), so we leave the ITF in LN2 for the night.

Finally we turned on the Etalon loops at 19.39 UTC.

This morning we worked on reducing the power on B1s in order to lock the ITF in LN2 without requiring manual SR misalignment.

When the WI CO₂ laser was replaced in 2024 (64847), a similar issue was observed and mitigated by moving the WI DAS in the horizontal direction ( 64851-64865).

Thus, today we acted in the same way.

We applied 5 steps:
STEP 1: with the ITF in LN2  (locked at 8.48 UTC), at 8:57 UTC we moved WI DAS by 1.5 cm (5556 + 2778 steps) in the forward direction. The ITF unlocked immediately after the DAS movement due to an increase of B1s power above the shutter threshold.

CARM NULL 1F reached again at 9:11 UTC. 

STEP 2: We agreed to come back 1 cm (5556 steps backward) at 09:18 UTC.

After this step, the situation improved:

• 12 MHz: 0.038 mW (with respect to tonight’s CARM NULL value of 0.036 mW)

• 112 MHz: 0.0255 mW (with respect to tonight’s CARM NULL value of 0.024 mW)

• B1p: 0.020 mW (with respect to tonight’s CARM NULL value of 0.018 mW)

We then proceeded with the lock acquisition by misaligning the SR by 0.5 µrad. However, the ITF unlocked at 09:57 UTC during ACQUIRE_LOW_NOISE_2, but not due to B1s saturation (cause unclear).

The subsequent lock acquisition did not show any issues, and the ITF locked again at 10:09 UTC in CARM NULL 1F. Then, we proceeded to LN2 without misaligning the SR; the procedure was successful and the ITF locked in LN2 at 10:29 UTC.

The bump in the right part of the camera of B1s is still present but it was less brighter (see figure 1- STEP 2). Thus, we  performed a new step in the same direction .

STEP 3: 11:01 UTC, we completed a movement 0.25 cm forward (1389 steps).

STEP 4: 11:31 UTC, we completed another movement 0.25 cm forward (1389 steps)

Looking at the B1s camera, we observed that the  spot moved to the left side of the camera (see figure 1- STEP 4). Therefore, we reversed the step by half to improve the left-right asimmetry .

STEP 5: 12:53 UTC, we completed another movement 0.125 cm backward (694 steps).

After these steps, the situation appeared to be worse with respect to STEP 4. Therefore, the last step (STEP 5) was completely reverted at 13:48 UTC.

Thus, overall, we moved the WI DAS by 1 cm with respect to the nominal position, in the right direction on the cardboard (and on the CP).
The ITF was then manually unlocked at 14:27 UTC to check the lock acquisition procedure.
The main ITF signals during the shift are shown in fig. 2.
The ITF relocked in CARM NULL 1F at the first attempt at 15:43 UTC and in LN2 at 15.01 UTC.

Starting from yesterday, the lock acquisition was less reliable, in particular, comparing to last Friday, we got lot of unlocks trying to reach CARM_NULL_1f.

Yesterday, we could somehow survive by misaligning the SR also between CITF and CARM_NULL. The same trick didn't work today. An oscillation at low frequency (~2Hz) was starting at the very end of the CARM_NULL. 

Looking at the loop engaged in that step, it appeared that the main culprit was the engagement of the full bandwidth for the SDB1 alignment. It seems that during the transient to go in carm null, the new filters for the AA were saturating the corrections, causing a general instability of the ITF. Keeping the bench in drift control at first, and postponing the full bandwidth at the end of the 1f waiting time (~10') seems to solve the issue. In addition, Michal reduced the gain of the full bandwith by half (in ACL directly).

Some notes:

• we slightly changed the gain of MICH and SRCL in CARM_NULL 
• after disabling the SDB1 full bw, we got few unlocks 
  • two at CARM_NULL_3f due to  low freq DARM oscillation (didn't happen again)
  • two consecutives at the engagement of the SSFS. Restarting the SSFS_CTRL server did help to lock correctly right after    
• the SDB1 full bw AA has been moved from line 4512 to line 4615 in ITF_LOCK
• we put BS_TX_SET and _RAMP to 0 in LN3_ALIGNED 
• since the recovery after the installation of the SR baffle,  the reconstructed BSN RANGE is sistematically below the RAW data

To reach the final state, we misaligned the SR_TY before locking the OMC and waited in LN2 till the SR was almost realigned.  Because of this, the automatic relock in LN3_ALIGNED will most probably fail.  

We leave the ITF in LN3_ALIGNED for the night. 

The aim of the shift was to find a good working point of BS TX to reduce CMRF in order to be able to engage the servo for TY.

In LN2, we engaged the BS TX full bandwidth and we scanned its setpoint to minimize the LSC DARM SSFS LF Q. We set the TX offset to be -4.5e-4 and then we engaged the servo on BS TY. In the first trial the ITF global alignment was already ruined by the large scans- For this reason we re-tried setting the final offset and closing the BSty servo just after reached LN2.  Unfortunately the ITF unlocked few minutes later (14:41:01 UTC), due to a drift of the SDB2 bench (plot 1).

We repeated the procedure with the following lock and it worked (plot 3).

I ve compared the main signals in three configurations;

- October 12 (before the "change of wp") blue,

- Dec 11 (good sensitivity before the baffle installation) orange

- Jan 8 (after baffle installation) purple

It can be noticed that:

- B7 is slightly different (Etalon NI loop actually open)

- B1p DC is back to the October level

- Sidebands lower than October but higher than December

- Higer CMRF (similar to October)

- SRCL set lower than October but higher than December

- DARM OG same as December

- MICH set much lower than december

The working point seems that has been recovered with respect to December (indeed no MICH offset is needed in the lock acquisition)

The SR has been tralated horizontal of about 0.5mm to check effect on the sensitivity and on the WP.

see attached plot

no clear effects are visible

Today after having reached LN3 (56MPc) we have aligned the SR (no offset on tx and ty). The transient is visible in the attached plot.

The sensitivity was quite low and worsened by the high CMRF. We re-engaged the BS ty loop but the required offset was too high to keep the lock.

data to be analyzed to evaluate the level of noise.

to be noted:

DC readout gain tuning (ITF_LOCK.ini)

#371 darm_gain_pd3 from 0.25 to 0.3

#383 darm_gain_b1_dc from 0.4 to 0.35

Today we reached the lock up to LN

We had several unlocks:

- many ALS unlocks

- many CARM2MC (for this reason the gain has been decreased from 1.3 to 1.1)

- two DIFFp ty high frequency oscillation @ CARM 0, gain decreased from 12 to 9

- 2 unlocks at the B1 PD3 handoff (gain decreased from 0.36 to 0.25)

- 1 unlock for B1s too high (threshold increased fro 44 to 48)

In order to lock the OMC we have misaligned the SR ty +2urad at the end of carm0 1f

In agreement with Antonio, I ve switched off the NI etalon loop

I ve repeated the analysis shown in logentry # 68214, having the starting time (the 0) when I ve equalized the controls of the NI and WI to spot if this could have an effect on the response to the YAG.

As it was expected nothing changed, since the effect of the YAG on the mirror temperature is too fast to be corrected with the etalon loop. It is visible that the effect of the YAG is 2.5 larger on the WI wrt the NI.

Moreover if I compute the difference between the delta temperature, to remove common temperature variation due to external disturbances, it is visible from the bottom plot that the WI temperature decrease of ~0.01 [C] more than the NI after 1h of missing YAG.

Analysis :

I' ve made a simple analysis. I ve taken since the start of the run, the 0 correspond to when I ve equalized the controllers of WI and NI, all the data set for which the ITF was locked in steps >carm 0 for at least 1h and unlocked for 1h later and compared the delta in temperature from the unlock (when the YAG just disappears) and 1h later.

I ve changed the WI controller, which was oscillating, to be equal to the NI to better understand also the difference between the two plants

Due to the external temperature decrease the Etalon loop corrections were saturated since the 22 of November

we decreased the set point of one fringe (for NI and WI) in 3 days NI to 19.543 and WI 20

I' ve made a simple analysis (too simple to be bullet proof since it doesn t take into account the loop). I ve taken since the start of the run all the data set for which the ITF was locked in steps >carm 0 for at least 2h and unlocked for 2h later and compared the delta in temperature from the unlock (when the YAG just disappears) and 1h later.

It is visible that the response of the WI has been always higher. There is an increment of the delta (higher delta means higher absorption) in the last 30 days, but it is not proven to be a stable behavior and a better analysis has to be performed in order to take into account the loop.

This morning we carried on with the recovery of LN3.

We observed another example of the complex interaction between MICH offset and SR alignment. This morning we could not recover neither the fringe nor the sidebands balance introducing a MICH_SET, but when we removed it both of these improved considerably (Fig.1). Overall, the configuration in which we attempted the OMC lock was MICH_SET=0 and +2 urad on SR_TX.

During the OMC lock, we had to keep fighting against the SR alignment loop, which produced the sawtooth shape visible in the figure.

While the ITF did reach LN3, the recovery is still not complete since we still did not restore an automated procedure to arrive in this state reliably.

The control of the WI etalon was misbehaving since the 3rd of November. This could be due to a too low gain of the normal control (the one that is activated once the error signals exceed the treshold of 0.05).

We have changed the gain from 2700 to 5500

to be observed in the next days

This morning, starting from the standard TCS configuration used throughout the run, we began exploring a possible DAS tuning aimed at reducing the MICH offset, which is currently around 70 (compared to ~20 measured before). A total of three steps were performed, all involving an increase of the inner ring power on the NI.

STEP 1
NI inner ring power increased by 20%.
NI PICKOFF value: 0.578 W → 0.583 W.
Action completed at 08:44 UTC.

STEP 2
NI inner ring power further increased by 20%.
NI PICKOFF value: 0.583 W → 0.589 W.
Action completed at 09:55 UTC.

At this point, we manually unlocked the ITF to test the lock acquisition.
The ITF relocked at the first attempt at CARM NULL 1F at 11:13 UTC, and, after imposing a MICH offset of about 70, it relocked at LN3 at 12:02 UTC.

STEP 3
NI inner ring power further increased by 20%.
NI PICKOFF value: 0.589 W → 0.603 W.
Action completed at 12:54 UTC.

After this step, some glitches were observed in several signals, including the BNS range (see the attached figure), suggesting a possible degradation of ITF stability. For this reason, we decided to revert the third step.
Consequently, at 14:36 UTC, I reduced the NI inner ring power back to the value of STEP 2.

Then, the ITF unlocked spontaneously at 14:50 UTC and relocked at CARM NULL 1F on the first attempt at 15:09 UTC.
The ITF reached LN3 at 15.57 UTC. 

This morning I ve changed in LN3 the setpoint for NI Y. The overall stablity of the lock has been worsened by the decentering and the behavior was not showing a good direction.

Y positive:

- MICH offset reduced

- B1p increased

- B7 & B8 increased

- DIFFp TX set increased

Y negative opposed behavior but in both directions the ITF stability was worsened

As a part of the global check of the status of any part of the INJ subsystem that could be related to that recent issue we performed a mode matching measurement with the kick method (i.e. applying corrections on the end mir z and analyzing the 02 mode during the free swinging). The last measurement was done in June 2024 #64450.

We had to repeat twice the procedure because in the first attempt only the north arm data were good. At the end, the gps used for the analysis are:

- 27/10/2025 18:44:24 UTC, 102 s (North arm) 

- 29/10/2025 19:10:00 UTC, 120 s (West arm)

The analysis started 25 and 40 sec after the unlocks for the North and West respectively (so that the ITF should be still "hot"), see plots 1 and 2.

We found high values for the mismatch: around 4% for the north and 5% for the west. 

We could try to improve it by tuning the mode matching telescope.