TMS Meeting Minutes, January 14, 2021

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

Meeting Notes:

  • Heejoo visited the mountain for various troubleshooting topics (system sanity check, clean fiber tips/collimators, realign channels, check SX fiber bundle, power meter check, MPO fiber cassette check, etc.)
  • Andrew's eTalon Data Analysis
    • For bad data cutoff, using a measurement of std. dev. of the current measurement with the last two measurements
    • For channels where std. dev. is too high, the algorithm drops that specific channel and recomputes the Jacobian with the reduced set
    • Only skipped when there are fewer than six usable channels
    • This week, using all available data with different Jacobian
    • Looking for constancy of the pose acceleration
    • In general, the std. dev. filter is successful in removing the jumpiness of the channel data
    • Expecting 9 to 8 ch. drop to be the most common situation on the telescope
    • When only dropping from 9 to 8, the rotations vary only a small amount, though the translations contain a bit more structure
    • For some reason, pose differences in Y is 10 times what it is in X for the drop 9 to 8 ch. case
    • 20 um of Y is not comfortably small, inducing a substantial amount of coma for FPIA
    • Q: What is the real explanation for these differences and for the structure in the pose difference for 9 to 8 channels?
    • John suggests that the submicron variations could be caused by dome seeing
    • Dropping from 8 to 7 channels, the rotation errors where larger, on the order of arcsec
    • It seems that moving from target to target affects the pose aceleration
    • 400 nm -> 15 microns allowable on FPIA
    • Operationally, would seem that we are borderline concerned about the pose acceleration
    • There appear to be unexplained effects outside of channel dropping causing pose acceleration
    • The fact that the pose diff plots are not flat lines (for 8 to 7 ch.) means the pose offsets are not linear with channel dropping
    • When the departure from constant is inching towards borderline for FPIA
    • Andrew planning to conduct further analysis and put everything down on paper to help contextualize the results
  • Andrew would like to request from Yang a section of the code to set a reference with less than nine channels avaialble
    • Currently, no way to send corrections with TCP/IP code and monitor with VFAST Extra b/c there doesn't seem to be a way to set a reference with fewer than 9 channels
    • LBC_offset sets the reference
  • Standard Deviation Filter
    • Andrew suggests standard deviation filter with three successive measurements to apply sigma cutoff
    • Use only channels that pass the sigma filter to construct the reference
    • John notes that it will take longer than 21 seconds because there are mirror updates that have to run, as well
    • In general, want to set a reference just after FPIA has been run, when the mirror is still
    • Would require the telescope to operate for a minute after FPIA without active control
    • Yang notes that when TCP/IP reports measurement failed, the eTalon system doesn't receive a valid meaurement, prior to any processing
    • Currently, the software is set up to only calculate pose when all 9 channels are available
    • Turns out this is not practical; Andrew would like to now request that we enable the ability to use less than 9 channels
    • Suggests incorporating the standard deviation channels health check with a sigma filter to reject channels of 0.02 um sigma
    • Essentially, take three measurements, calculate sigma for each channel, then reject those channels which fall outside a sigma of 2 microns until a new reference is constructed
    • Andrew also requests making the sigma filter value a user-configurable value
    • Requires three steps of measurement history (current plus previous two)
    • The 1% error allowance for channel dropping is simply two large; the three-measurement sigma filter should solve this problem and reject when needed
    • Yang notes that our current software is stateless, so he will need some time to find a way to implement this filter
  • Next chance for LBC is on the 17th UT with Barry
    • It will be useful to run the system with our newly aligned channels (thanks, Heejoo!)
-- TrentonBrendel - 14 Jan 2021
Topic revision: r2 - 18 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