Camilla, Sophie

We reprofiled the QCL unit 0970 because we were concerned the SNR was too low and effecting the profile of the beam (see alog 92242). The 99:1 BS placed in front of the QCL was replaced by the CO2 99:1 BS which at 4.6um acted effectively as a 70:30, with the 30% beam transmitted. We profiled the 30% beam at 1A and 25C over a 50cm range. 
The strange beamsize fluctuations around the waist were observed again so we concluded this was a not a measurement error and slightly non-gaussian behaviour of the beam.
The complex q parameter for the laser was found using a non linear fit to be:

--- Horizontal (A1) Fit ---

Data range: z = [97.0, 597.0] mm
Beam radii: [422.0, 1412.3] µm
Fit converged: w₀=411.4+/-9.2 µm, z₀=224.9+/-3.9 mm
q = -225+/-4 + 116+/-5i mm

--- Vertical (A2) Fit ---

Data range: z = [97.0, 597.0] mm
Beam radii: [413.6, 1183.7] µm
Fit converged: w₀=421.0+/-10.9 µm, z₀=293.7+/-3.9 mm
q = -294+/-4 + 121+/-6i mm
Note: Louis pointed out the plots of the fit have weird error bars which I will investigate in the QCL fitting code. 

The measured q-parameter is very similar to the Y-arm unit and a very similar telescope optimisation could be completed. This layout is listed in the git and has the following beamsize and astigmatisms. 

Beamsize at ITM
wx_ITM: 53.299 mm
wy_ITM: 52.042 mm
w_ITM_mean: 52.671 mm
w_ITM_astig: 1.257 mm

This is an improvement in astigmatism of approximately 7mm from the 0922 unit we swapped out.

We are currently aligning the table and the beam is aligned up to the PDBS. Unfortunately, many of the other optics had to move to ensure the PD intensity noise measurement optics would not be positioned in the waist so they do not get burnt.

During the latter part of the day, Louis and I tried to get back to resonance, but have yet to get there. First, we had problems with TR carm (again), and then we struggled at the carm offset reduction.

For TR CARM, we thought the gain might be a bit low in start_tr_carm, so we upped the analog H1:LSC-REFL_SUM_A_IN1GAIN by 3 dB. This seems to help! Louis also increased the time in the servo to 10 seconds.

CARM_OFFSET_REDUCTION problems:

• We ran carm offset reduction and failed. We tried stepping up slowly in the offset and adjusting the gain, but kept losing lock. We found we can step to -35 offset and a gain of 60, and then up to -40 and a gain of 61. This is 3 dB lower than the original final gain of 86, which makes sense based on our gain increase in TR carm.
  • Even though this is guardianized, this still seems to fail unless done by hand. I don't know why because the guardian code should be exactly replicating what I do by hand: step to -7, wait, step to -12, wait, engage darm boost, step to -35 and 60 gain, wait, step to -40 and 61 gain, engage reflbias boost
• I have tried stepping the alignment around to see if I can make the recycling gain better.
  • Looking at the transmitted powers both yesterday and today, I am concerned that the refl power is too low and that we have poor recycling gain.
  • In O4c, I measured arm transmission and refl ratios to see if our carm offset reduction sequence was having a problem, see here. Turns out we were fine at the time. But we can compare those powers to now, reminder we had recycling gain of about 54.
  • For arm transmission of 800, the refl power should be about 1.96 mW, compared to 3.7 mW off resonance, so a ratio of 0.52.
  • Tonight, for a transmitted power of 800, refl power is 1.3 mW compared to 3.5 mW off resonance, so a ratio of 0.37. We saw similar powers last night, although we were able to make it all the way to resonance. This is still in the region of a gain of 30, but that does seem low compared to a year ago.
  • I tried stepping the DHARD alignment around, but it did not improve the refl power. CHARD alignment seems to help, but it's a small improvement. Since the improvement is so small when the transmitted power is low, I think I stepped the wrong way and caused a lockloss by reducing the carm offset with very low recycling gain.
• I tried one more time to walk CHARD around, looked like I was going in the right direction, but we still lost lock.

We had one lockloss at carm 5 pm after completing carm_offset_reduction by hand. Don't know why, but this was another state we sometimes had to carefully do by hand.

Want to reiterate that the IMC is still being fussy when relocking.

TITLE: 10/10 Eve Shift: 2330-0500 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: None
SHIFT SUMMARY:

IFO is in at DARM_TO_RF

Elenna and Louis worked hard to get past IDLE_ALS, whioch is the state right before we lose lock. The current hypothesis is that the PR_Gain ramp up is wrong so they've been measuring and then stepping up the gain (more negative) slowly to see what is enough.

It turned out that the arm alignment was bad, and we successfully got past DHARD_WFS. Though the problem is still happening so Elenna is troubleshooting.

We've relocked fully automatically with DRMI over 10 times. 

LOG:

TJ, Jim, Arnaud

Implemented a new HAM3_CRS guardian. The purpose of this guardian is to :

1) monitor the state of the CRS on HAM3 (Laser status, 25mHz tilt amplitude, and readout noise (3_10Hz)
2) adjust the damping gain depending on the tilt amplitude, and
3) take it to "SAFE" (Damping OFF) when maintenance is necessary 

This morning, we tested the condition for a rung up CRS, see attached screenshot. The CRS was manually rung up, and we let the guardian take over. After the 1urad rms threshold was reached, guardian jumped to CRS_RUNG_UP state, changed the damping gain to 1 (from nominal 0.01), and went back to CRS_GOOD after the RMS threshold came back under threshold. The earthquake that happened later today also triggered the CRS_RUNG_UP state as expected, see second plot from this alog

Guardian logic description is written under: T2600479

TITLE: 10/09 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Earthquake
INCOMING OPERATOR: Ibrahim
SHIFT SUMMARY: 

We are back to trying to relock after the big earthquake earlier today.

Locking notes:
Morning Initial Alignment
- Had to help XARM because it kept catching on wrong mode, but probably would've caught on its own eventually

Relocking
- Kept losing DRMI, mostly due to mode hopping - this was because the green arm references hadn't been sdf'd. After resetting the refs to what they were yesterday we were good
- Huge 7.6 earthquake come through and kept us down for multiple hours. It was in the BIG RED BUTTON zone, so I pressed the Big Red Button™.

LOG:                                                                                                                                                                                                                                                                                                                                 

TITLE: 10/09 Eve Shift: 2330-0500 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Earthquake
OUTGOING OPERATOR: Oli
CURRENT ENVIRONMENT:
    SEI_ENV state: CALM
    Wind: 17mph Gusts, 12mph 3min avg
    Primary useism: 0.12 μm/s
    Secondary useism: 0.17 μm/s 
QUICK SUMMARY:

IFO is back to locking after Panam M7.7 (and 6.6 aftershock).

We're getting to DRMI.

Oli, TJ, Arnaud

We used the earthquake down time opportunity to evaluate safe IMC thresholds for a rung up, in loop CRS. A threshold of 1urad rms at 25mHz seems safe for alerting the seismic nodes to take the CRS out of loop. We will have to verify those thresholds are not too conservative, and that they still work with a full ifo. 

For this test :

HAM2-HAM3 were set to their nominal blends (no sensor correction), and the CRS was put in loop for HAM3. The CRS was rung up to 80 urad pp, 28 urad rms, by flipping the sign of the CRS damping gain. We monitored the IMC stayed locked throughout (fig1). The mode cleaner controls (on MC2) reached at most +/- 50e6 out of the +/- 134e6 limit (less than half the total dac range). In comparison, the Panama earthquake from this afternoon (or the change of ISI state) rung up the CRS to +/- 2.2 urad, 1.5 urad rms (fig2). 

Louis and I started morning locking once Oli re-re-ran initial alignments with the "correct" green camera offsets (ones set by the baffle scripts).

We got to DRMI ASC, and the SRC P error signals looked close to zero, so I tried engaging SRC1 and SRC2 P. However, we saw slow large oscillations that went away when the SRC P loops were turned off. I also saw that the alignment wasn't brought to the best buildups- I had to move SRM pitch more to maximize. I also reset the POP A offsets (SDF attached), which is a minor change but we have the best buildup now (ignoring the >200 buildup from last Friday)- POP18 is 167 and POP90 is 20. As an aside, I think we need to pico onto the POP QPD sled. The POP A offset required is very large, about 0.98 in both pitch and yaw. We're probably close to or falling off of POP A and the same is true for POP B. Perhaps we can lock DRMI, close PRC1 with the current offsets, let it converge, open the loop, turn off the offsets, and pico to bring the beam closer to the center of the QPDs.

I have now created separate "use DRMI ASC" flags for SRC1 and SRC2 pitch and yaw. The yaw flags are set to TRUE, and the pitch are set to FALSE. The ISC_DRMI guardian will now separately engage these loops in ENGAGE_DRMI_ASC. This means that only SRM and maybe SR2 pitch will need to be hand-adjusted for buildups in DRMI ASC. Tagging opsinfo.

Louis and I tried to go through TR CARM four times:

• Try #1: tried running straight through carm to TR, lockloss
• Try #2: went to start TR carm, saw that the REFL_DC_BIAS output signal had an offset. Tried adjusting the TR CARM offset to remove the offset. Had to lower it to -0.3. REFL DC BIAS was around zero. Tried measuring TR CARM using the template, had no coherence, bumped up the template excitation by factor of 100. Measurement showed low TR carm loop gain, by roughly a factor of 3. Tried raising the REFL DC BIAS gain from 6 to 12, lockloss. See attached "TR carm fail 2" for measurement.
• Try #3: Decided maybe we railed the input to the common mode board on last try, decided to raise the analog gain instead. Went up 3 dB, remeasured. TR carm gain looked better, but still low. Went up 1 more dB step. Immediate lockloss. See attached "tr carm fail 3" for measurement.
• Try #4: Oli alerted us to an incoming earthquake from Panama, big enough for big red button. Oli spoke to Jim, was advised to press big red button asap. No chance to try carm on tr.

We're holding in down while this large earthquake passes.

I want to re-emphasize that we are still having trouble relocking the IMC after lockloss. The IMC will acquire lock, and the boost will kill it. This can get intense enough that the boost kicks the FSS which then kicks JAC. The only way around it is to either wait a really long time, or to hold the IMC guardian in acquire, wait a minute, and then engage boost. Doesn't always work, but can be faster than just waiting forever. Tagging IOO, someone needs to look into this boost.

TITLE: 10/09 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Commissioning
OUTGOING OPERATOR: None
CURRENT ENVIRONMENT:
    SEI_ENV state: SEISMON_ALERT
    Wind: 7mph Gusts, 4mph 3min avg
    Primary useism: 0.03 μm/s
    Secondary useism: 0.16 μm/s 
QUICK SUMMARY:

Starting an initial alignment as requested

TITLE: 10/09 Eve Shift: 2330-0500 UTC (1630-2200 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: None
SHIFT SUMMARY:

IFO is in COMMISSIONING and FIND_IR

Overall a good shift where Louis and Elenna outlined the steps that worked to get to RESONANCE - alog 92254.

Instructions after they left were to run an initial alignment, which I did. I had some initial troubles with MICH_DARK since my first adjustment of the BS was not enough. After figuring that out, I was able to get all the way through. By this time, Louis came back on at around ENGAGE_DRMI_ASC and took over.

One weird thing is that the power would auto-jump to 10W after SRC_ALIGN was being offloaded. I may have clicked the LOWER_POWER guardian when we were stuck at 10W after MICH to see if it would go down to 2W (instead of going to the LASER_PWR guardian, but it doesn't seem to have worked and may have fudged the state ordering). Either way, SRC did actually offload. So after that, I set PWR back to 2W and started locking. This was fully auto.

LOG:

We have achieved resonance, and could probably make it to carm analog. Unfortunately, the steps are not fully automated just yet. Here is a summary of how we are doing this:

• Lock to "engage DRMI ASC"
  • the SRC Y ASC works, but the SRC P ASC does not yet work, so I take us here, manually engage SRC Y, and then adjust SRM pitch by hand
• MICH ASC is disengaged after DRMI ASC offloaded
  • Louis and I attempted to get MICH ASC running on AS 36, we failed twice and gave up in favor of getting to resonance
  • The beamsplitter oplev servo should be re-engaged at this point, pitch gain is 300, yaw gain is 650. It can be done by hand
  • We had some locklosses due to oscillations which turned out to be the beamsplitter moving too much, the oplev servo helps this
• From here, we can run through to CARM OFFSET REDUCTION
• At this point, the current arm alignment is very far off from the zero point of the DHARD error signals.
  • I was using the move_arms_dev.py script to walk the DHARD error signals to zero. Steps of +- 0.01 work well. Turn off the smooth limiter for both 
  • Then, DHARD WFS can be run
  • Earlier today, after we let DHARD converge, I reset the green camera references. This means that if we run a full initial alignment, this step should no longer be necessary. I recommend doing this asap. Tagging opsinfo
• We have had some trouble running carm 5 pm with the guardian
  • Both Sheila and I have been running through this state by hand. I think I have edited the guardian to work properly, but if it doesn't, I recommend the following steps:
  • Run lines 2747 and 2752 in ther ISC_LOCK guardian to raise the PRCL gain
  • Adjust the TR CARM offset by hand up from -40 to -48. As you do, remeasure the tr carm OLG and check that we are between the two blue reference traces. If the gain is too low, increase the TR CARM gain. I have found that 1.5 works well, this is what is in the guardian
• We can then run carm to refl and resonance in the guardian
• Louis and I think we have adjusted the analog gain settings in carm to analog such that it should run without intervention, but we haven't tested it yet.

I know Sheila had her own alog documenting our work for most of the day, but I don't see her post just yet.

I'm attaching the green camera SDF, and also the rephasing of RF36 that Louis and I did earlier.

I was walking the overpath and heard hammering noise, and there it was, an ice-pecking raven. You can hear the sound as the guy pecks in the attached video (PXL_20261008_233238334TS~2.mp4, use headphone). I was shooting from the middle of the overpath slope, so it was kind of far but the noise was very much audible.

Thick layer of ice formed around the LN2 pipe, see picture (PXL_20261008_233616363MP.jpg).

TITLE: 10/08 Day Shift: 1430-2330 UTC (0730-1630 PST), all times posted in UTC
STATE of H1: Commissioning
INCOMING OPERATOR: Ibrahim
SHIFT SUMMARY: Commissioners have done a great job getting us up to RESONANCE. 

I had some issues in the morning with ALS but we didn't have any issues after that. Relocking has been really fast and smooth.
LOG:                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                              

Sheila, Louis, Elenna, Camilla. Follow on from 92167. 

Louis made this Googledoc. It incudes a table of Powers before and after BS swap, with and without arms. It shows that we now have a higher ratio of powers on POP ports with the arms that without the arms that we used to have in O4. 

There’s still work done on this and finding better comparism times, but see attached table for arms vs no arms ratios.