TMS Meeting Minutes, February 11, 2021


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

  • Andrew, John, and Olga obtained hysteresis data and Andrew is working on data processing
    • When mirror driven back to reference, observed <4 pixels of transverse focus (about +60 micron mismatch)
    • Andrew is convinced TMS was seeing correctly, so it's most likely thermal focus
    • FPIA converged to a different value for focus than TMS suggested by 60 microns
  • Andrew has reduced the data from Jan. 2nd, where FPIA was run continuously
    • B/c of TCP/IP communication error, weren't able to use V Fast Extra
    • Instead, ran FPIA continuously for hours while monitoring with TMS
    • Andrew has set up a program which takes the vacuum Etalon output with all channel lengths and calculates the vacuum corrected lengths using Etalon values and our DMS values
    • Planning to construct a one-stop-shop to get pose out of Etalon, do thermal corrections, and analyze
    • On Jan. 2nd, had air ang glass temperatures (glasstherm, airtherm around mirror, airtherm from weather station)
    • Air temp induced change in line length of a few microns (while temperature was changing by a few points of a degree)
    • Computed glass therms change, assuming a geometrical scale
      • With glass therms change of a 0.1 degrees, see about 2 microns of mirror focal length change
      • 2-3 pixels of focus, when considering only the CTE of the glass
    • Referencing FPIA log
      • Matched timestamps of Etalon data to timestamps of FPIA
      • Integrated focus from FPIA and converted to microns/millimeters
      • Plotted change in vacuum length along Z with FPIA and Etalon pose
      • Substantial discontinuity between FPIA and Etalon Z vacuum lengths
      • Queried John for more data
      • Realized that when we have an applied Z from telescope focus change, have Z4 plus Z11 plus Z22
      • Andrew still needs to compile the effect of all of these terms
      • Expecting glass therms to have a gradient related to a finite focus shift on the mirror from mirror shape change
      • Expects mismatch of TMS Z correction to FPIA Z correction to correlate to mirror thermal changes
    • Andrew would like to request engineering time to gain better insight into these effects
      • Either using TMS with V Fast Extra
      • Or by using FPIA with mirror ventilation running to artificially produce thermal gradients in the mirror
      • Suggests selecting a field with very small elevation change to mitigate variability caused by elevation
    • Olga notes that the matrix which computes the Z4, Z11, and Z22 terms sent to the mirror may not be correct, so how can we disentangle this from other effects
    • Andrew remarks that we may directly look into the effects of thermal focus (plus thermal Z11 and Z22), deflection-free
    • Buddy Martin once stated that it's something like a 40 micron Z4 for 1 deg of thermal _
    • Andrew requests the specifications locating all glass thermistors

  • Etalon is preparing a beta version of the Etalon software to fix the TCP/IP connection issues which multiple customers are experiencing (we should have this within the next week)

  • Software discussion from John, Matthieu, and Yang
    • Yang planning to clean up the long email he sent and post it to the Wiki as the current software spec
    • The software diagram on the Wiki page is somewhat outdated, so he is planning to redraw it
    • At present, when a measurement can't be made for whatever reason, retries set to 100 times
    • Yang plans to change the retry count to only 3 instead of 100
      • Yang would prefer not to use a prompt to the user, as the current method is simpler
    • John proposed introducing a brief pause between tries
      • Yang suggests that the repetition rate isn't really taxing the hardware, and there is already a slight delay just sending commands over the network, even with no pause coded in
      • John proposes a policy of not hammering hard on other's software needlessly
      • Andrew proposes that Yang take this up with the auther of the software at Etalon to confirm that continuous retries won't be an issue
    • John mentioned that it would be a good idea to request 10 readings of the temperature telemetry and average or median the values
      • There could be an issue with being pulled away by a sudden jump from a misread
      • This change could make the software more robust and mitigate the risk of errors
      • John remarks that you may need to change TCS so that it spits out the running median reading (the weather station reportedly polls data every second)
      • Yang suggesting internally caching the weather telemetry within the TMS software
      • John also suggested polling data from DMS, with data populated every 5-6 seconds, perhaps simpler to ask DMS for the last 20
      • Yang to ask Matthieu about computing the median in TCS with frequent sampling
      • As of now, the temperature is only sampled when a measurement is received
    • Computation of standard deviation: do we keep a timestamp on the past states used to compute the standard deviation
      • There are cases under which the spread in time between measurements used for std. dev. could be large
      • John proposes we limit the allowable timegap for std. dev. correction to reasonable values
      • Andrew suggests that there may be no need to control for this, as a large change will be rejected in short order by new measurements with a rolling std. dev.
      • Yang remarks that you need some assumptions to make this work well, and we need to be sure that the assumptions hold under most cases
      • This consideration depends on whether or not you plan to flag any of the three measurements which are bad or not
      • If we don't discard any bad measurements from history, than there is no need to consider the timestep between good measurements used in the std. dev. computation
      • The 2 micron parameter seems to be good as is, for now
    • Comment on configuring which channels should be used as a reference
      • Andrew remarks that there is great need to be able to manually specify the channels used for reference creation
      • The primary reason for manual channel removal is for the case when we know we have a channel which is failing intermittently
      • Yang planning to add an option in the command line to specify channels to disinclude from the reference specifically for engineering mode
      • At the moment, TMS is only operating with 7 channels (Ch. 8 and Ch. 11 are malfunctioning)


-- TrentonBrendel - 11 Feb 2021
Topic revision: r1 - 12 Feb 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