It glitched again in the same manner at 00:32:12 UTC,

Since this looks like a force noise, and my spider-sense thinks this is some sort of electronics failure, I trended the coil current for the ETMX PUM and ETMX TST (i.e. the primary stages of the DARM actuator), and found nothing that didn't look the same is it did in the previous lock stretch. See attached.

Here are some dtt's of the broadband moise in the darm spectrum.  I had a look at the spectrum during the unexplained range drop from Niko's alog 51400.  These ones we're having now seem to be at a lower frequency.

Sheila to the rescue -- the ALS X shutter was open. (And the clue came from ALS X trying to lock, which changed the ALS X SERVO GAIN and knocked us out of observation intent...)

Shutter is now closed and DELTAL_EXTERNAL is back to normal.

Now investigating why the shutter didn't get closed during these sets of lock acquisitions............

J. Kissel, S. Dwyer

The future is now! With the watchful eye of Sheila, I edited the above mentioned X and Y arm codes such that once the transmitted (green) Arm Power is above 0.9 [normalized counts], and the I phase outputs of the (green) WFS A and B are less than 10 [DAC counts], then
 (a) read the current offsets,
 (b) round the QPD offsets to a precision of 1e-6
 (c) write those rounded offsets, and
 (d) break out of the while loop

(a), (b), and (c) are all done in one line via the magic of ezca python bindings, and (d) is new -- the script used to just ask the user to CTRL+C out of the script.

This code change is completely untested, and again we likely won't use it for another 3-6 months, but hopefully I've left enough breakcrumbs in 
 - this aLOG,
 - the instructions that reference these scripts, LHO aLOG 50968,
 - the comments of the code, and 
 - in the log of the svn commit 
that people will come find me when it doesn't work. Apologies in advance!

I'm also in the process of updating the FOM wiki page.

Patrick loaded these changes after a lockloss, and didn't have guardian problems when relocking.  

Request for operators:  I am hoping that this might solve the problem of the IFO lossing lock in the early CARM reduction steps in the first locking attempt after each time we run initial alignment.  I would appreciate if you would report relocking attempts over the weekend, including if you run initial alignment, and if on the first relocking attempt you make it to the CARM offset reduction do you survive those steps?

As a reminder, there is a spreadsheet which could be used to help us keep track of problems encountered when relocking:

https://docs.google.com/spreadsheets/d/1byXR-s2ATegciJCEjORb1FEZhZTp77j6DS-fZuvEY-8/edit#gid=0

In the first locking attempt after Tuesday maintence, initial alignment was run, and the interferometer was relocked without loosing lock, on the first attempt https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=51561 (Also confirmed by trending the guardian state at the times Jeff aloged)

This means that this solution has worked, and we shouldn't have these failed locking attempts after each initial alignment. 

J. Kissel

In the above aLOG, I've compared all of the measurements to the same, no spring, low-frequency optical response model to emphasize how much it changes with spot position and srcl offset. We see that there is not only a spring-like effect, but also some (what we believe to be) parasitic L2A2L coupling of some unknown shape or form (especially evident in the requested 100ct SRCL Offset, August Spot position data).

We also know that our MCMC algorithm for fitting this data, for which we only give it the poles and zeros to model a spring (2 zeros at 0.0 Hz, and an arbitrary set of complex poles at frequency f_s with phase separation defined by Q), cannot accurately give us approximations for what at least the spring part is doing because it's getting confused by the parasitic L2A2L response.

However, in order to make progress on how much detuning we have, and how the requested 100 ct SRCL offset changes that detuning in physical units, we need *something*.

So -- I've noodled around with the optical plant parameters that we *do* have in order to make a "best" by-hand fit of the data. Here, "best" is based on minimizing the residual (ratio) between measurement and model, acknowledging that I will never be able to get an actually good fit, because I'm not using enough poles and zeros to cover the L2A2L effect. Also, acknowledge that my by-eye happiness is determining the uncertainty in the parameters, so we have no nice neat, multi-dimensional posterior distribution that exposes the (very real) covariance between the parameters.

Given those caveats, I attach the results.

Here's a table of the resulting fit parameters.
                                                                 No Offset                    100 ct SRCL Offset
                                                         Aug spots       Jul Spots        Aug Spots      Jul Spots
    optical gain, H_c               [1e6 ct/m]           3.13 (0.01)     3.16 (0.01)      3.13 (0.01)    3.16 (0.01)
    caviy pole frequency, f_cc      [Hz]                 395  (3)        405  (3)         409  (3)       415  (3)
    optical spring frequency, f_s   [Hz]                 7.2  (0.5)      6.0  (0.5)       5.0  (1.0)     0.0  (1)
    optical spring quality, Q       [n.a.]               10   (5)        35   (5)         10   (5)       1.0  (1)
    spring type                     [n.a.]               pro             pro              anti?          no?
    measurement date                [YYYY-MM-DD]         2019-08-19      2019-08-21       2019-08-19     2019-08-21
    attachment page                 page 1               page 2          page 3           page 4

Again, in the 100ct SRCL offset data, there's not enough poles and zeros in the fit, and/or not a visible enough spring to really trust the spring-related assessment last two columns, and I thoroughly recommend we map out the parameter space better and develop a model (or get rid of) the supposed L2A2L coupling so we can really do a proper fit on this.

We can at least confidently say that Jul spot positions, with a requested 100 ct SRCL offset is best for the optical gain and cavity pole.

JeffK noted that we don't have a good all in one place how-to on setting the initial alignment offsets.  So, here's an attempt at it.  This assumes that the QPDs are all close enough to center (or can be made to be so by moving TMS) that you do not need to use the picomotors.  Using the picos requires more thought and care.

The attachment is my desktop when I'm doing this, so you can see all of the windows that I'm looking at. 

• ndscopes:
  • bottom left: reacquire lock 'scope from my Locking screen
  • bottom middle: ADS 'scope from my Locking screen
  • right side: New template in ..../jenne.driggers/Templates/Scope/green_wfs_setpoints.yml
• medm screens:
  • sitemap, just because I always have the sitemap open
  • ALS camera setpoint screens, from each ALS end overview screen, then CAM ITM PIT green button.
  • tiny A2L screen: these are the A2L spot position gains from each test mass's PUM stage.  You can pull them each open individually from each SUS, or get this compact screen from my Locking screen, near the top right.
  • Compact QPD screen for the 4 IR transmon QPDs, to make sure that we're not falling off the IR QPDs.  Get this from my Locking screen, "ASC Pointing" button.
  • Green QPD offset sliders, get from either of the ALS end overview screens.
  • Sliders for the 2 TMSs.
• Terminal windows:
  • One running setEndGreenQPDOffsets_XARM.py
  • One running setEndGreenQPDOffsets_YARM.py

Process for each arm (can be done in parallel since each arm is independant of the other for this work):

• Ensure IFO is at 2W with all ASC engaged and converged (DC readout is a good spot for this).  Make sure that the ADS error signals are close to zero.
• Ensure green shutters are open (sitemap -> LSC -> Shutters, then "Open" on the two shutters in the second from right column)
• Ensure that the green laser is locked on the TEM00 mode to the cavity.
  • If it is not, you can noodle around the green QPD offset sliders.  Since these are in-loop with the pointing PZTs on the ALS in-air table, this effectively changes the pointing of the green laser into the vacuum, and into the cavity.
• Run the setEndGreenQPDOffsets_[X,Y]ARM.py script, which will adjust the green QPD offset sliders to adjust the green laser input pointing to match the arm cavity (the metric in the script is zeroing the ALS WFS signals on the bottom row of the right-most ndscope). 
  • Watch the QPD offsets to make sure they aren't moving the spots too close to the edges of the QPDs.  Top two rows of the rightmost ndscope have Ylims set to +/-1, so it should be easy to tell when you're getting too close.  The third row is the sum of light on the QPDs, so you know that if that starts dropping you're getting too close to the edge.  We've been operating at offsets of more than 0.8 for all of O3, so while that's not great, it's not unacceptable (since these are unused in Observation mode).
• Slowly move the TMS (steps of 0.1 urad seem fine with the standard 2 sec ramp) to get the beam closer to center on the IR QPDs.  Hopefully the green offset script will be pulling the green offsets closer to center, if they were close to their edges.  If they don't, you should stop here and reconsider things.  You may need to use the picos. 
• Once you've got the TMS moved to center the IR QPDs and the script has converged such that the green wfs signals are close to zero, you should set the camera offsets.
• Now, you're done and you should ensure that your new setpoints are saved in the down and observe SDF snapshots of the ALS end stations.

If you must pico, I'd recommend only pico-ing the IR QPDs since they're much easier to think about and independent of other beams.  I'd start the green QPD setpoint scripts to make the green QPDs follow your movement.  Then I'd make a small step in TMS alignment in the direction that will put the green QPD offsets closer to zero, and then pico the IR steering picos to get the IR QPDs closer to zero.

I've edited the code that sets the green QPD offsets (/opt/rtcds/userapps/trunk/als/h1/scripts/setEndGreenQPDOffsets_XARM.py and setEndGreenQPDOffsets_YARM.py) -- check out LHO aLOG 51476 for further commentary.

This procedure needs to be modified because we are now setting green references with our spots centered on the optics.  We would like to center the IR on the IR QPDs once we have our spots in their final positions, however, not necessarily with the spots centered.

Since Craig added in the TMS servoing to center the IR QPD, we should still be able to follow this procedure at DC readout.  At DC readout, in the current situation, we are servoing the spots to the centers of all 4 test masses at 2W input power.  This is the position we want to set the initial alignment references to. 

Later in the acquisition sequence the power is increased to 10W, and then the beam spots are moved to their final high power positions, then we further increase the power and continue on to NomLowNoise.

Before setting the green camera references, we should remember to check that the camera image is good and that any mask applied is appropriate.  We can check this by opening up the digital camera screen and taking a snapshot, then finding the snapshot in /ligo/data/camera