TMS Meeting Minutes, January 7, 2021

Attendees: Andrew, Breann, Heejoo, John, Trenton, Matthieu, Olga (Zoom Meeting)

Meeting Notes:

  • Andrew recently took LBC data in the past few days (also has Mark Wiseman tee'd up for tests)
    • Seems we might have a few chances in the next few days after midnight
    • There appears to be some sort of error when we start up
    • Usually have to do at least a software restart, if not a full laser restart/system restart
    • These issues have been logged, and Mark Wiseman is to come on tonight to help solve these problems
    • This issue appears to be happening more regularly, and it didn't used to happen
  • Critical Action Items for Next Week's Mountain Visit
    • Ch. 11,17,23 alignment is crucial, realy need to lock it in
    • Fiber routing to SX
    • Testing MPO Cassette, power meter
      • So far, failing to transmit MPO fiber to single mode fiber
      • If this new unit works well, we should let eTalon know about the failure of their unit
      • GMTO should have this marked up in their account; eTalon owes them one
    • Retro from Andrew (Heejoo picking up next Monday); Andrew would like a few photos when Heejoo picks up the H/W
      • Mangin Mirror (essentially a cemented doublet), with a protective paint on the rear gold coat
      • Andrew requests two tests
        • Does it work with only 14% return?
        • Set retro inside of SMR nest and rotate to test angular range. Instead of 45 deg from standard, hollow corner cube, the full range of the solid corner cube should be 100 degrees across.
      • Andrew suggests testing the return with Chang Jin Oh's Leica IR tracker
    • On the channels with weak signal:
      • Take a port cleaner and a fiber cleaner to these channels
  • Matthieu has completed the TMS laser power monitor alarm
    • System incorporated in the last two weeks
    • Effective at determining when the power meter is on and when the laser is off
    • There are some issues with the power source of the power meter
    • However, generates too many emails right now, but this will be amended and the rest of us will be added shortly
    • The hardware setup is currently in beta, but once it progresses, we can push it out
  • Andrew Analyzed TMS Data
    • Looking at channel dropping, jumps, etc.
    • Joel Nyssen thinks we should not be seeing jumps as the number of channels changes, but Andrew is convinced we are seeing jumps
    • Mathematica program with 9 channels, imports big data set, calculates pose for each measurment for a given number of channels
    • Each line does a health check
    • Up to now, we have just been looking at an absolute value
    • What seems to be a good check is to turn the system on and look at the standard deviation of the first three measurments (takes 20 seconds for first 3)
    • After health check, drops channels as necessary and reconfigures the pose
    • Skips measurement when more than 3 channels are dropped (criterion at 1 micron)
    • In individual results, we do have substantial jumps based on the distinct Jacobian used, based on the number of channels available
    • Jumps appear to be fairly consistent
    • Data over 4-5 hours, with no big drifts, just jumps
    • This results appears to confirm that we indeed need an offset for different channel dropping configurations, but this offset was presumed to be constant (appears this is not the case)
    • Next step, planning to evaluate all of the data with individual channel configurations, then look at the variance in the offset
    • Planning to write this up in a comprehensible description of results
    • Variance in the LBC results was showing up in the TMS results
    • Steps visible after running FPIA and reconfiguring
    • Andrew has a few ideas for the system to make sure we never command this outlier points
    • After you start your reference and start measurement, if you obtain a large SD from the last three measurements, don't send a correction
    • This standard deviation check could be quite effective for mitigating jumps
    • If you have commanded a pointing change, TMS will allow this change, noting a large SD, and will resume corrections once the new set point normalizes over a set of three measurements
    • All it costs us is a couple iterations at 7 seconds each
    • Trenton and Breann to assist Andrew with this analysis
  • Note: At the moment, we can't configure the number of channels used in our reference, meaning we can't run the system if we don't start with a full complement of channels
    • We would like to be able to set the number of channels upon setting the reference; Yang will need to inform Andrew of the correct syntax
  • Data from a couple nights ago
    • Plan A: run FPIA w/o sending corrections to the mirror and to control position with TMS and look at what evolved in WFS data compared to a stable TMS
    • B/c weren't able to run Yang's software w/o a full complement of channels and couldn't set a reference, settled on Plan B
    • Decided to monitor position with TMS instead of correct with TMS
    • Ran FPIA in closed-loop with a convergence criteria that was impossible to meet (running all the time assuming collimated telescope)
    • Still have a continuous set of data
  • Data Visualization for 20201201 IQ Plot
    • Zernikes now included, along with the glass thermistor average temperature
    • Zernikes from PSF system in DMS, with the total value reported; updated when FPIA
    • Bit of a mystery why the LBCr IQ drifts so much from 10:30 to 11:00
    • -4000 nm of focus jump at 10:45; corresponds to a 100 um measurement in TMS Z
    • John notes that even small jumps shouldn't make any systematic trends because they have to jump back eventually
    • Andrew suspects bugs in TMS software that we do not yet understand; suggests getting a clearer idea from eTalon data
    • Very useful if we could more directly compare TMS TCP/IP code to eTalon system
    • Andrew requests Yang prepare a version of his code which does not send any corrections
    • Run eTalon code in normal way with Yang sifting data out with TCP/IP, then we can get parallel sets of data and match them
    • TMS measures the focus correction applied by FPIA, then TMS appears to correct it out
    • 0.8 mm step noted on TMS Z corresponds to FPIA step
    • When we are controlling with TMS, we should be able to exclude rigid body motions
    • Either we are not controlling rigid body motion correctly, or there is some other mechanism for IQ which is not rigid body motion
    • Instead of plotting the TMS readout, we instead plot the group rotations from the TMS correction
    • About 100 micron group rotation Z value jump at 10:50 (excluding defocus for FPIA)
    • The rise in group rot. Z value from 10:45 to 11:15 follows the upwards slope of the LBCr IQ (FWHM)
    • John hypothesizes that this is a change in focus for a non-rigid body motion
    • One of the suspects is that we are doing something wrong in our newly written TMS code, and we haven't properly qualified this code against existing eTalon code
    • Either something is affecting red but not blue that's not related to position or the TMS is moving off gradually from where it should be by about 100 um
    • Given the accuracy of the basic hardware, it's likely that the issue is in the code
    • John reminds us that TMS does not try to unfocus when FPIA defocuses, but if you don't take a new reference, it will undo the added correction
    • The point is that image degradation happens before those FPIA steps, around 10:30, where there is no group rotation in Z slope
    • Looking at Rx, Ry, not much coma added in over this period of time
    • Recall that this was a period when Olga was seeing donuts
    • Stable situation with glasstherms, assuming z meas. constant and good, assume corrected glass also stable thermally
    • What could happen with detectors to produce piston?
    • Could get into a funny situation where the corrector lenses were a vastly different temperature yesterday and they have a long thermal time constant (the ventilated primary mirror has a time constant of less than an hour, in comparison)
    • This would be a very unusual situation, but this is possible
    • Looking at the ambient thermistor average for the past 24 hours, we see a plummet from 6 degC at 00:00 Dec. 1 to 1.5 degC at 07:30 Dec. 1
    • It would be nice to hook up some temperature sensors to the corrector
    • Peter Grey has some iMonit temperature sensors; would like to put these inside the barrel, but not sure if the signal would transmit out
    • John suggests placing temperature sensor in gap outside of L1 of the corrector
    • Andrew recalls that GMTO has thermal modeling of the GMT WADC with temperature dynamics over a couple hours, will share with team
    • The only way this would work, is if we have a cliff in the temperature
    • A very distinct upwards cliff is observed in the ambient thermistor average from 9:30 to 11:00, which could explain the degradation in IQ over this period
    • Olga reminds us that the FWHM is an average for the entire focal plane; we do see gradients in PSF across the focal plane
    • Supposedly "in-focus" data, though we did see donuts
    • Olga wondering if we can do any more with that data
    • Andrew proposes that we will eventually get to a point where we can look at field-dependent IQ and compared that to Rx,Ry,x,y,z
  • Upcoming LBC Nights
    • Andrew coordinating with Eric and Mike Wiseman in Germany, so he plans to start up the system
    • The TeamViewer issues seem to be persisting
  • 20210101-02 Measurements with Olga and Andrew
    • 48 hours of ambient thermistor aveage
    • Appears to be a larger effect
    • For this dataset, not only TMS data, but also WFS data
    • Continuous TMS data, WFS data, and temperature for ambient and glass thermistors
    • Note that solar heating during the day heats the dome steadily while the mirror temperature only increases marginally over the same period (big jumps only when mirror ventilation is on)
  • Trenton to polish MATLAB code to synthesize future datasets with I/O
    • Use getdata to prompt user to find data files on personal drive to import into MATLAB
    • TMS x,y,z,Rx,Ry,Rz, PSF, temperature, group rotations, elev., azi., etc.

-- TrentonBrendel - 07 Jan 2021
Topic revision: r2 - 08 Jan 2021, TrentonBrendel
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback