I changed the make-up air speed back to its nominal setting of 20% at ~19:15 UTC (12:15 PDT) on 8/7/2024.  The make-up air being on did cause the Laser Room temperature to drop by a couple degF, which in turn caused the output power of both Amp1 and Amp2 to drop by ~2W each; will monitor over the next several hours to ensure things return to normal as the enclosure returns to its normal thermal equilibrium.

Since this happened, Ryan C went into EX to reset VFD Panel, which allowed HEPI ETMX to stop tripping, but we noticed the T240s railing right after and then switching between extremes - this happened at 13:12 UTC. Screenshot below.

As of 13:31 PT (20:31) UTC, it looks like the T240 counts are coming back down.

A look at the sensors during this trip.

Was epics set point for the hepi pump beckhoff set to zero before resetting the vfd? If not, the t240 saturations are probably caused by the changing pressure in the hepi fluid while the pump servo ramps to operating pressure. Normally just resetting the pump vfd without zeroing out the set point would trip the pump, but maybe we got lucky.

Power glitch Monday 19:41:21 UTC.

Caught by MSR UPS, corner station mains_mon DQ channels for all three phases shown.

H1:HPI-PUMP_EX_CTRL_LOC_SETPT was set to zero but H1:HPI-PUMP_EX_CTRL_RMT_SETPT was still set to 70. I noticed the pressure came back on its own without me having to use the controller, that setpoint must be why. I'll make sure to zero it first next time.

The last 2 times I've run the OPLEV charge measurement for ETMX, about an hour later we've had a power glitch that takes down the VFD at the end station, 08/05 alog79455, 07/23 alog79288. Weird coincidence?

I agree with Oli that SRM will be "as good as it has been between 2021 and 2024" if we leave it as is, but for different, more expanded reasons covered below.

To expand on Oli's "RT OSEM has a problem" -- the issue is that with the alignment offsets OFF, the RT (right) OSEM sensor / flag has expanded enough that the flag is only barely occulting the LED / PD beam, close to "open light." Thus the drop in magnitude response for Length and Yaw, the only two DOFs that use the RT OSEM. In fact, the M2 UL and M3 UL OSEMs, i.e. the lower stage left OSEMs are further *engaged*, close to "closed light." This is consistent with an overall positive yaw (+RZ) of the suspension w.r.t. the cage.

The *current* SRM alignment offsets are 
    (P: 1663, Y: -3095 [urad]) 2024-08-01, post OFI recovery and aux laser alignment, in air, HEPI locked 

This digital request of *negative* ~3 millirads of yaw *centers* the OSEMs, implying that the OSEMs have been centered to the *alignment offsets ON* position. With the OSEMs centered, the magnitude of the response increases and looks cleaner because both LF and RT are contributing to measurement of the L and Y DOFs.

This checks out with the last time that we moved the OSEMs w.r.t. to the flags -- in 2021 -- see LHO:60576 where our goal was to keep the physical alignment of the optic the same, even though moving the OSEMs changes the digital request to get there. In that case, we kept the physical alignment and centered the top mass OSEMs to the alignment offsets ON, preserved physical alignment.

Importantly -- the result of the 2021 adventure of reproducing in-vac alignment while the IFO was in air was to leave the alignment offsets at
    (P: 1738.28, Y: -3296 [urad]) 2021-11-08 (LHO:60578)
which is within 200 [urad] of the current slider values, and where the OSEMs are reasonably centered and responding well now.

These are also consistent with pre-vent in 2024 with OFI KTP damage in place, where with HAM5/6 HEPIs locked, a single-bounce IFO beam was used to find a good alignment
    (P: 2192, Y: -3166 [urad]) 2024-07-17 (LHO:79246)

The offsets that Ibrahim used only on the L DOF during the July 17th health check, as Oli quotes, are:
    (P: 4254, Y: -9231 [urad]) 2024-07-17 (LHO:79202)
which I believe were some very-lost, very-brief, temporary "I didn't realize HEPI was locked" alignment state used just prior to the official single bounce restoration later that day, LHO:79246, so the values of these should be ignored.

Finally, I attach a trend of the SRM alignment offsets over the last 4 months of sordid history with the OFI's KTP damage, covering both in-air and in-vac, HEPI unlocked and unlocked.
This also confirms that SRM's alignment, in all of these scenarios, is never needed to be outside of the 
    (P: 1600-2400 [urad], Y: 3000-3500 [urad])
range, i.e. less than 1 [mrad] from the 2021 positions (and the I suspect the large low pitch request after venting is the need to account for bouyancy of the optic, so really the alignment should stay within 0.5 pmrad] = 500 [urad] of those 2021 values.)

Note that this requested alignment range consumes a little less than half of the DAC range at ~50000 [ct] for the pitch OSEMs (T2 and T3) and yaw OSEMs (LF and RT).

Thus, if we just leave SRM alone, I suspect we'll be just as happy the SRM alignment and OSEM performance as we were between Nov 2021 until Jul 2024. This was the same decision we made in 2021. We'll just have to somehow remember that, for SRM, in order to get centered OSEMs to provide a OK performance on the top-mass "health check" transfer functions, we need the alignment offsets ON and in within 500 [urad] of (P: 2200, Y: -3300 [urad]). We can test where the boundaries of the alignment slider requests vs. RT OSEM performance decay during pump-down.

That being said, if we really want to mechanically relieve the yaw offset and restore centering of M1 RT (and M2 & M3 UL) we can...

Opened IIET Ticket 31769, to record that we've thought carefully about this decision to "let the 2021 OSEM position and SRM mechanical alignment ride," but that we reserve the right to try to mechanically offload *next* time.

Hopefully the ticket and this aLOG series provides enough breadcrumbs to remember that we should relieve the alignment when we have the AUX laser set up already. 

Sheila's original video is ~2sec footage of fast shutter, supposedly at 150 fps according to the file metadata.

Attached is an edited video that shows the 50ms of FS action slowed down by a factor of 10. "SRC TC" is the time elapsed since the start of (Sheila's original) video multiplied by a factor of 10, i.e. 1 sec in SRC TC is actually 100ms. This timestamp is not to be trusted at the lowest digit of sub-second (ms level in reality) because there's some frame interpolation going on due to mismatch between the standard output frame rate of 24FPS VS the source frame rate (150 slowed down to 15 FPS). Anyway, the mirror goes down lower than the closed position right after the initial thrust of the HV fast shutter motion, and even a few bounces after that, before the shutter eventually settles to the permanently closed position.

Also attached is an annotated png file of four frames ("Open", "HV thrust", "Bounce to the bottom" and "Close").

In light of recent failure of the replacement FS at LLO where the mirror broke off due to the hard stops that were set too low, and a comment by Rich (LLO alog 73584) I watched the video of the LHO unit again (see my entry above).

Though LHO video was shot from the side where we cannot see the position of the hard stop, the bobbin might be hitting the stops right after the launch even though the height of the stops looks good to my eyes. See TC~00:00:05:06. And the bobbing oscillates several times though it never hits the stops again.

Does this mean that the magnetic braking is too weak?

I converted this microphone DQ signal to an audio wav file. The microphone saturates at +/-32768 for a fraction of a second, but you can hear the echo fade away after this.

Here is the accelerometer as a wav audio:

(new file added at 13:52, orig file was created using the incorrect frequency of 16kHz, this is actually a 8kHz channel)

The ITMY Oplev (aligned at the time) didn't see this hard close, plot attached. 

Remember GV1 is closed, this makes ITMY Oplev blind, Sum on trend above is very small.

Here're a bunch of trends that I used to inform how much DAC / coil / magnet drive I should be putting on M1 in order to achieve an alignment of H1 SUS SRM that replicates the most recent good in-vacuum, nominal low noise alignment. 

The time I used was around the time of May 09 2021 20:30 UTC.
Though the alignment swims around slowly on the ~12 hour timescale during the lock stretch, it's not that much, so I used the following DAC request to recreate the roughly equivalent alignment:
                                      [urad]              [DAC ct]       
H1:SUS-SRM_M1_OPTICALIGN_P_OFFSET     1738.28   OUTPUT    3259.45
H1:SUS-SRM_M1_OPTICALIGN_Y_OFFSET    -3296.41   OUTPUT    -8836.36
  
                                 [DAC ct]
H1:SUS-SRM_M1_LOCK_L_OUTPUT      -5000
H1:SUS-SRM_M1_LOCK_P_OUTPUT         10
H1:SUS-SRM_M1_LOCK_Y_OUTPUT       -120

H1:SUS-SRM_M1_COILOUTF_T1_INMON      0
H1:SUS-SRM_M1_COILOUTF_T2_INMON  54769
H1:SUS-SRM_M1_COILOUTF_T3_INMON -54769
H1:SUS-SRM_M1_COILOUTF_LF_INMON  53488
H1:SUS-SRM_M1_COILOUTF_RT_INMON -58488
H1:SUS-SRM_M1_COILOUTF_SD_INMON      0
(where, in case it's not clear, the LF RT OSEMs drive Longitudinal and Yaw, and the T2 T3 OSEMs drive Pitch.)

Just because I'm insatiably curious... ("were the OSEMs actually 'open lighted'?" so we're not confused between "actual LED decay" and just "a little bit of flag is still occluding the LED light") I also attach here a trend of the H1 SUS SRM M1 BOSEMs while Betsy has used the PEEK CAM nuts to have them longitudinally "backed off" from their flags.

It's ... too interesting if I look too closely (at the "10s of counts" level), so I'm going to ignore this, and justify it with the same argument I gave recently to a different SUS team in LHO aLOG 60179: 
"you need not worry about a 50 / 30000 = 0.001 or 0.1% error in the calibration gain of 30000/OLC.

Conclusion -- 10s of counts error (or uncertainty due to environmental conditions) don't matter."
See that aLOG for full discussion.

Just prior to this recentering, I took a health check of the SUS in air with the ISI locked, using data from templates marked with the time stamp 2021-11-08_2000. Attached are the post-processed results. Nothing alarming here, just posting for comparison of data we *after* this change to the top mass OSEMS. (Well, not *change*, but re-centering as best we could around the in-observation alignment.) 

Cross-referencing IIET Ticket 31769, which reminds us to revisit the mechanical alignment of SRM in 2025-2026, because in 2024, we decided to "let it ride," where "it" is the mechanical alignment kept, and OSEM positions we set, in this aLOG.