Oli, Georgia, Elenna

While we were taking care of several LSC items, Georgia pointed out that we are seeing the 0.8 Hz buzzing in INP1 yaw again, and to a lesser degree CHARD Y. While we did several useless things to debug, I noticed that the DC centering loops were also buzzing with the same frequency oscillation. I decided to try changing the DC centering loop gain (since I had raised the loop gain significantly last week), and noticed that NONE of the DC6 pitch or yaw filters were engaged, but a gain of 1 was engaged, essentially feeding the input signal right back to PM1. Bad!!

Searchingh DC6 in the guardian shows that in PREP_ASC_FOR_FULL_IFO, I found these lines of code:

Not good! Since the gain of DC6 is set on, this effectively keeps the gain on but turns off all the filters and just feeds back the pure error signal to the suspension.

I commented those lines out of the guardian and loaded, and then reengaged all the proper filters. A large 28 Hz line that had been present in DARM disappeared at this point, likely because the DC centering loops require a 28 Hz notch to avoid some suspension mode.

After this, I then tried turning down the DC1 and DC2 pitch and yaw gains to -1. This seemed to help the buzzing in CHARD and INP1. Georgia and I were able to engage all the low noise ASC settings without any issues (CHARD P had been held in high bandwidth to avoid a ring up and the soft loops had high gain for the same reason).

We need to monitor for a while longer and also try relocking, but these changes may fix many of our ASC issues.

I have SDFed these gain changes.

I initially tried rephasing POP 9 by hand, but moving the POP A phase by 5 degrees while we're locked using it was way too much and immediately caused a lockloss.
The next lock, we found Gabriele's old LSC sensing matrix measuring software in 
/ligo/gitcommon/labutils/lsc_sensing_matrix/lsc_sensing_matrix.py
It works great still.
The 250 Hz line is injected using awg, and the FM5 EBS250 notch filters are still in the right location. 
The excitation amplitudes were 10x too strong, so we lowered them and it stopped the saturations.

The resulting phases found initially are posted as the first PDF.
We were pretty far off in POP 9, by 20 degrees, and by -11 degrees in POP 45.

To rephase, I wrote the script to rotate the phase of POP very slowly while in lock:

step_size = 11
steps = 100000
total_time = 110
init_value = ezca["LSC-POP_A_RF45_PHASE_R"]
for ii in range(1,steps):
    small_step_size = step_size * ii / steps
    small_time = total_time / steps
    print(f"{ii}: {small_step_size}")
    time.sleep(small_time)
    ezca["LSC-POP_A_RF45_PHASE_R"] = init_value + small_step_size

This one worked, with only very small glitches visible in DARM.
PDF 2 shows the resulting fixed LSC matrix measurement, at 60 W input power.

Next, we adjusted the LSC sensing matrix according to Jenne's old instructions: alog 68547.
We turned on SRCL2 and MICH2 FM5 EBS250 notch filters, and started a 250 Hz line in PRCL again.

We then plotted SRCL1 and PRCL1 error signals while adjusting the POP 9 I -> SRCL intrix value by 0.005 steps, looking to minimize the appearance of the 250 Hz PRCL line sensed in SRCL.
We didn't move LSC-PD_DOF_MTRX_SETTING_5_1 much: only by 0.02 from 0.10408 to 0.085, but got significant improvement, maybe a factor of 3 better.
Makes some sense we'd have to adjust this after rephasing the POP diodes.
The resulting LSC sensing matrix is posted as the PNG.

M. Todd, C. Cahillane

I've scheduled a suite of injection measurements (low-risk, I don't forsee it unlocking the IFO), to start at 10:30 tonight (2025-0603 22:30:00 PDT), as we are interested in figuring out the loop gains for the IMC DOFs Pitch and Yaw.

If it needs to be stopped, you can Ctrl+C my work station or xkill 248480.

ISC_LOCK LOWNOISE_ASC

• ADS clock gains were set to 30, we raised this to 300
• ADS loop gains were increased to 10 but now we leave them at 1
• Some consolidation of switching on filters one-by-one vs in one line
• Removed some ancient TRAMP setting of ASC_AS_A

ISC_DRMI

• We've been manualling over PRMI ASC since there was a convergence checker on PRC1, Ryan, Oli and I added a use_DRMI_ASC and use_PRMI_ASC dictionary to lscparams, replacing the self.usePRC1 flags that were set in individual states of ISC_DRMI. If we did this right, it will only wait for convergence of loops that are engaged :)

CAMERA_SERVO

• We've been switching on the camera servos on earlier using the camera_servo guardian. It has a few points where it checks the ISC_LOCK guardian state. I lowered the number to trick it into letting me turn it on earlier...

I updated the beam mode model prevously generated in alogs 30126-30130 (20160930) and alog 39706 (20171208).
It includes arm optics changes, as well as and output path model based on alog 84089 (20250425). With Kevin's help, I checked it aganst the default FInesse model. The models are reasonably close, although this mo9del does not include a) astigmatism, and b) thick otpics.
The model is available here:

/ligo/home/controls/sballmer/20250603/beamCalc.m

Some conclusions:

- In the cold state the SRC eigenmode has a beam size on the ITM of about 6.3cm (the arm mode is 5.3cm).
- To get to the matched "HOT" state, the ITMs need a 25uD lens.
- To match a thermal lens twice as big (50uD), the SR2-SR3 distance would have to shrink 8mm. None of tf the existing mode actuators can get us there.
- I haven't looked whether the PRC needs a simlar fix. I also didn't check whether out sidebands could accomodate this shift with a simple single-optic move.
 

https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=39706

https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=30126

https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=84089

This study looks at the effects of differential ITM/ETM RoC change, common ITM/ETM RoC change, and differential ITEM lenses change on the DARM sensing function as we track several quantities such as the mismatch between the ARM cavities, ARMs-SRC mismatch, and the beam spot size on TMs. 

FINESSE results suggest that the mismatch between the arms through the ITMs (both RoC and lenses) has the biggest effect on the sensing function. 

In all of these cases, we start with a minimized mismatch between all cavities to less than 0.01% by changing the RoC of several optics until the mismatches are minimized. All mismatches are calculated with the “model.cavity_mismatches_table()” function in FINESSE .

Steps to generate these plots:

1. Changing a parameter (RoC or focal length)
2. Lock the model using DC readout after each change
3. Calculate and plot the sensing function

For the RoC, the 3rd line is always the minimized mismatches case. Maximum mismatch I created was ~0.9% 

Case1: Differential ITMs RoC

Starting with the case of differential ITMs RoC changes by ±2%. ITMX RoC increases while ITMY RoC decreases. The mismatch between the ARMs and the SRC also changes, but that is smaller than the mismatches between the ARMs themselves. 

This has a strong effect on the sensing function shape at low frequencies. It seems that there is a stronger optical spring as the mismatch between the arms increases. The label on the plots show the ITMs RoC in diopters calculated as 2/RoC. 

Changing which RoC increases and which decreases results in the same sensing functions. 

The plot attached shows that result. And a table (attached as .png) shows the mismatches, diopters and more. 

Case2: Differential ITM lenses focal lengths

Similarly, I change ITM lenses' focal lengths. Since, as defined in FINESSE at least, ITMXlens is negative and ITMYlens is positive, a differential change means that ITMXlens diverges the beam stronger every time (becomes less negative. e.g -3*10^5 m instead of -5*10^5) , and ITMYlens converges less, meaning it becomes a bigger positive number. Diopters in this case are calculated as 1/f

This creates a differential mismatch between the ARM cavities. The result is very similar to that of differential ITMs RoC changes. 

The plot and the table attached show the sensing functions. 

Case3: Differential ETM RoC 

This also creates a mismatch between the arms that is bigger between the ARMs and the SRC. This will only affect the carrier’s mode, but not the sidebands, like the ITMs do. 

This case does not have a big effect on the sensing function, as the ITMs (RoC and lenses). The sensing function plot and table are attached.

Case4: Common ITM RoC 

Now, I keep the ARMs mode matched to each other but I mismatch the ARMs to the SRC by changing the ITMs RoC. This has a bigger effect on the DARM sensing function compared to ETMs but smaller than the ITM. Interestingly, this is the only case that shows a pro-spring behavior

The sensing function plot and table are attached. 

 

From this investigation, it seems that the mismatch between the ARM cavities has the biggest effect on the sensing function, rather than the mismatch between the ARM cavities and the SRC. 

Qualitatively, this suggests that the sensing function’s shape is an indicator of how well the arms are mode-matched to each other.

Elenna, Georgia, Craig, Stefan, Oli

We saw a very strange oscillation that appears to be beamsplitter related during the engagement of DRMI ASC. I tried turning down various DRMI ASC gains: SRC1, SRC2, MICH P, and it made no difference. We were getting BS saturation warnings. Finally, I just moved from"engage drmi asc" to "turn on BS stage 2" and the oscillations appeared to reduced. We saw large peaks with harmonics in the corner LSC signals starting near 18 Hz. We are wondering if it's an errant beamsplitter bounce mode?

TITLE: 06/03 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Planned Engineering
INCOMING OPERATOR: Oli
SHIFT SUMMARY:

IFO is in PLANNED ENGINEERING and LOCKING

Productive light maintenance tuesdays in which most of the work was done checking annulus pumps (VAC) and doing cable work on ISCT1 (EE). After this, we begun locking again.

There was also some GC/CDS work done with the internet, which has concluded. Other than that, normal Tues Maint. activities such as tumbleweed removal and NORCO took place.

We were able to sit at MAX_POWER for quite a while, before comissioner-caused lockloss.

DRMI ASC now works but more work needs to be done on PRMI ASC. This means that after a manual initial alignment, we can get to MAX_POWER automatically.

The majority of the work now is being done fixing an issue thought to be mainly with ADS (and soft loops that engage during and after POWER_UP.

The plan for then night can be found in Oli's alog 84765

LOG:

TITLE: 06/03 Eve Shift: 2330-0500 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Planned Engineering
OUTGOING OPERATOR: Ibrahim
CURRENT ENVIRONMENT:
    SEI_ENV state: CALM
    Wind: 12mph Gusts, 6mph 3min avg
    Primary useism: 0.02 μm/s
    Secondary useism: 0.22 μm/s
QUICK SUMMARY:

We lost lock from LOWNOISE_LENGTH_CONTROL a bit ago, so we are working on relocking. Plan for the rest of the day:

• Check PRMI offset in DRMI ASC
• Craig & Matt rephase LSC POP in lownoise
• Craig & Matt check PRCL/SRCL sensing/inputmatrix
• Elenna measure LSC FF
• Vicky & Kevin sqz measurement of SRCL offset
• Check laser noise suppression state
• Vicky & Kevin other sqz stuff (in parallel w/LSC stuff)
• * Matt overnight IMC measurement

Today at ~11:10 am, IP3 (LVEA 2x 1250 l/s IP) failed leading to a slight pressure rise in the corner. Seen on all LVEA pressure gauges (BSC2, HAM6, etc.). Attached snapshot shows the two pumps behavior at the time of failure, pump ion current on left column, output voltage on right.

Gerardo and I went to the mechanical room to diagnose the controller, upon arrival both channels on the Dual Vac controller had tripped off. We reset the high voltage and powered on the IP. Channel 1 (corresponding to Pump A in CDS) would only reach 300 V and then would trip the controller. Channel 2 (Pump B would reach 7kV no issue but would trip when Channel 1 tripped). We then swapped to a Gamma MPC controller to test if the controller was the problem. We saw the same exact behavior where Pump A would only reach 300V and Pump B was ok at 7kV.

We then swapped back to the DualVac controller after confirming it was not the issue, but with only Pump B connected and powered on. We are continuing to monitor to make sure Pump B remains steady. In parallel, Travis and Mark are testing the HV cable to the pump to see if there are any shorts. Results to follow.

If the cable is not the culprit we will HiPot the pump to see if there are any shorts, pump will be isolated from main volume during this test. To be continued.

I changed that way that the SQZ_MANAGER SCAN_SQZANG gen state worked to now take a brms_channel argument. Nominally for SCAN_SQZANG_FDS this is SQZ-DCPD_RATIO_3_DB_MON which tunes the squeezing around the 350Hz BLRMS to give the max IFO range. This is not yet tested, changes in SQZ_MANAGER.py svn.

I added a new state SCAN_SQZANG_KHZ (state number 85) from FIS that uses SQZ-DCPD_RATIO_6_DB_MON to maximize sqz at high frequency ~1700Hz. It makes sure ASC is off before scanning angle and can be be used when we take a sqz data set with SRCL offset steps e.g.  83569 to make 83572 brontosaurus plots.

To take SRCL offset FIS dataset (e.g. 83569):

• No SQZ (for subtraction in models)
  • Good time to measure NLG (instructions in 76542) and change NLG if wanted,
  • For SRCL offset measurements, increase NLG (NLG of 52 used in 83569 by increasing sqzparams OPO trans setpoint up to 120uW, SHG waveplate might need adjusting)
• Take IFO to FREQ_INDEP_SQZ
• For each SRCL offset:
  • Turn off SQZ ASC (H1:SQZ-ASC_WFS_SWITCH)
  • Change SRCL offset H1:LSC-SRCL1_OFFSET (prev used 0, -90, -190, -290, -390)
  • Go to SCAN_SQZANG_KHZ, will take 2mins (can check scan data at /ligo/www/www/exports/SQZ/GRD/SQZANG_SCAN/)
  • Go to FREQ_INDEP_SQZ, ASC should turn on as long it's in on in sqzparams, let it converge for ~1minute
  • Take 3 minutes of data, without glitches, note time down.

Take NLG back to nominal ~11, once done.

As a part of trying to diagnose what could be wrong with CHARD Y, I checked the ITMY A2L gains, in case they were so far off that it was responsible for some of the instabilities we've seen. Just based on the magnitude of the changes, I don't think it has been causing any problems.

While we sat at 25 W, after the move spots was completed, I drove a pitch and yaw dither on ITMY using the DOF2 paths in the ADS bank. The pitch line was driven at 19.125 Hz with a magnitude of 5000 and the yaw line at 21.287 Hz with a magnitude of 3000.

I made steps of 0.2 in both the pitch and yaw gains in both directions, and watched the line heights change. I was able to make the lines both higher and lower, so I am convinced that I am at least close to, if not at the minimum. Overall, the change in the gains is not that significant, so I think that it wasn't causing us any problems.

In the screenshot below, cursors mark the location of each line on the red trace.

Sheila, Elenna, Caroline, Julia

We tested and reconfigured the DRMI ASC, and now we can run all of the DRMI ASC!

I trended the POP A offset yesterday to determine what the new values should be based on the alignment of the PRM once we are in full lock. I set those so we can use the PRC1 ASC loop (attachment 1 and attachment 2).

Then, we sat with DRMI locked and Sheila and I tried moving around PR2, IM4 and PRM to check if the error signals made sense. PRM uses POP A QPD, IM4 and PR2 use REFL A and B 9 MHz signals. We found that the PR2 and IM4 error signals seemed flipped, so we checked all of the REFL WFS signals and even thge POPX signals to see what might be the best new sensing matrix. Eventually, we chose to remain on REFL 9 I, but we flipped the sensing.

We were able to engage all the SRC ASC, except that after a minute or two og engagement, we saw an oscillation in the SRC signals. We eventually turned down the SRC1 P and Y gains, and the SRC2 P gain.

I tried moving the PRM to confirm the PRC1 error signal was ok, and I found that the pitch offset I chose was good, but the yaw offset was not good. So instead I moved the PRM yaw to maximize the buildups, and then reset the the POP A yaw offset (attachment 3).

Then, Sheila and I tried engaging the PRC2 and INP1 yaw ASC, but used the wrong sign for PRC2 Y and caused the DRMI lockloss. Luckily, we quickly recovered, and changed the sign and reengaged. We followed a similar procedure for the pitch ASC, but I stepped PR2 a few times to ensure the sign of the input matrix was correct.

We were then able to engage all the DRMI ASC, using lower SRC1 and SRC2 gains to prevent ringing. Sheila updated the ISC DRMI guardian to use these new input matrix values.

Except for some overall magnitude, the DRMI INP1 and PRC2 sensing has had this change in both pitch and yaw:

old sensing: INP1 = + REFL 9I A - REFL 9I B and PRC2 = +REFL 9I A + REFL 9I B

new sensing: INP1 = +REFL 9I A + REFL 9I B and PRC2 = -REFL 9I A + REFL 9I B

With the ASC engaged, we were able to raise the SRC1 pitch and yaw gains back to their nominal values (updated again in guardian). The SRC2 P loop still has a lower gain.

I updated the POP_A P and Y offsets so the DRMI ASC will take us closer to where the full IFO ASC will take us

• First screenshot of the POP inmon starting at DRMI_ASC, with a time cursor at full IFO ASC
• Second screenshot of the SDF - I'm fairly certain this is SDF controlled and not guardian controlled

HAM1 annulus system appears to have issues pumping down, currently its annulus ion pump is railed 10 mA, and the pump cart pumping on this system is sitting at 6.04x10^-05 Torr, see attached photo.

Process thus far:

• Last Thursday I managed to go and visit some of the -Y door bolts, found out that most of the bolts from 3 to 7 O'clock were not tigth, I was able to get 90 degrees turn, one bolt was actually loose, I was able to turn it by hand, but those were the only one that I was able to work on, since I was asked to stop work at HAM1.
• Another test done was to check for leaks on the aux cart pumping line, so the cart was isolated via the annulus system isolation valve, the pressure dropped at the aux cart down to 4x10^-06 Torr, meaning that no leak is on the pumping side, the isolation valve was opened again to continue pumping.

I plan to visit HAM1 to check the rest of the bolts on -Y door and +Y door.

TJ, Camilla, Sheila

TJ and Camilla got the HWSs working, and now we turned on the SR3 heater at 15:54 UTC, 2W requested power.  This is to do a check of the SR3 heater calibration and range similar to 27262

This morning we changed the demod phase of the POP 9 sensor.  The overall summary is that this has a very similar impact to adding a PRCL offset, reducing the PRCL to REFL RIN coupling, and slightly increaing power in the PRC, but not changing the coupling of PRCL noise to DARM.  I've turned off the offset, added a new phasing, and updated the PRCL to SRCL subtraction in the LSC input matrix.

More details:

I used a 702.1 Hz frequency noise injection, (template found in userapps/lsc/h1/templates/phase_LSC_sensors_frequency_exc.xml)  and adjusted the POP9 phase to reduce the appearance of the frequency injection in Q (screenshot shows initial to final settings change in Q peak).  I used Elenna's template to make a PRCL injection and measure the coupling to RELF RIN and DARM during this move, see screenshot.  Adjusting the phase reduced the PRCL to REFL RIN coupling in a way that's similar to what Elenna has seen by adjusting the PRCL offset (see 79989 for example). Also similar to what Elenna has seen with the PRCL offset, this has very little impact on the PRCL to DARM coupling.  The ndscope screenshot shows how this impacted REFL and POP powers, you can compare the time when I turned off the offset to the time when I tried reverting the phase to the original setting after finding the new setting, the two things have a simlar impact on the reflected power and the power in the PRC. 

One interesting observation is that the phase that minimized the PRCL to REFL RIN coupling (as shown by the sign flip in the transfer function), is different from the phase that minimzed the frequency injection coupling to POP 9Q, by 2 degrees.  I've left this at the phase that minimizes REFL RIN coupling.  The impact on the power build ups is small enough that this two degree difference can't be evaluated using them. 

Another observation is that the change in phasing doesn't seem to have an impact on the PRCL spectrum, spectrum screenshot. 

Lastly, I did another PRCL excitation to measure the coupling to POP 45 and POP9 I, using the template in userapps/lsc/h1/templates/SRCL/SRCL/SRCL_input_matrix_git_rid_of_PRCL.xml  (excuse the typo).  I used this measurment to try to zero the PRCL coupling to SRCL ((POP45/PRCL)*MTRX_5_3 + (POP9I/PRCL)*MTRX_5_1 = 0, and found that this matrix element needed to be updated from 0.1185 to 0.1408.  This was last updated by Jenne Driggers in 70919.  The update reduced the SRCL to PRCL coupling by about 20dB, shown here. 

Lastly we edited the guardian to not use the PRCL offset, and update the PRCL to SRCL input matrix element, and accepted the new matrix and phasing in SDF (in the observe file, and now also in the safe file).