TMS Meeting Minutes, January 27, 2022

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

Channel Loss:
  • Previously, ch. 10 and 11 were inoperable after our last alignment effort
  • Recently, we lost ch. 5, and we were unable to use our previous reference (unable to obtain collimation with smaller set)
  • On 1/21, when we ran in active mode, we realized the reference we used from a previous night didn't match the channels
  • Some observers asked Olga if it was possible for them to start the night collimated, possibly using TMS
  • Andrew notes that if we drop a channel in the middle of the night, we don't have a simple recovery available
  • Andrew has an action to look at statistics for which channels drop more and which are most noisy
  • Heejoo to look into data from 1/26 to analyze the noise in the channels
  • On 1/26, Dx mode run with active TMS and Sx run without
  • Sporadic elongation of spots on LBCr noted at various times (064521, 064635, etc.)
  • 40 second exposures, with a significant dRY, dRX, dY correction applied at 06:46:39
  • TMS DOES apply corrections during exposures (at present, though there was an idea that it may be possible to pause correction during exposures)
  • John notes that you have to read telemetry data constantly to determine if the shutter is open and prevent corrections during that interval, which is a lot of work
  • John suggests that something is wrong somewhere and that it shouldn't be doing
  • The errors shown on 1/26 indicate pose rotations around 0.5 arcsec, resulting in 1 arcsec on sky (rather alarming)
  • Andrew notes that we may need to look into our filtering algorithm and consider if we are allowing intolerable amounts of noise into the pose calculation
Passive Mode 1/27:
  • Around 10:15:00, there was a significant spike in Rx and Ry without a major jump in X or Y (plotting scale may be off)
  • It is possible that the telescope moved to an FPIA target before going to a new field during this spike
  • The Dx side was using the minimum channel set, 7 channels
  • The Sx data looks much more realistic (this side also had much more channels than Dx)
  • Note, the channel realignment is planned for the mountain crew to complete on Monday, 1/31
  • Andrew thinks we should look into doing a trend fit, subtract the trend, and look at RMS values for both Sx and Dx
  • Predicting that the Dx side, where we have one retro with only a single channel aligned to it, will have far degraded performance in Ry compared to Sx
  • On Sx side, using the smaller Thorlabs collimators, with only 7 operational channels
  • On Dx side, using larger Etalon collimators, with 9 operational channels
  • This discrepancy could result in very different performance
  • Andrew suggests re-analyzing the Sx side with only 7 channels to see if it makes the data jumpier in a similar way
  • Trenton to analyze the Sx data with channels dropped which are similar to the dropped channels on Dx
Mountain Trip Postponement:
  • LeRoy will be checking collimator alignment on Monday
  • Heejoo to have a meeting ahead of time to ensure all tasks are covered and things are clear
Mirror Thermal:
  • Trenton has an ANSYS model that he is working with
  • Andrew offered his son's help as a summer intern to work on the thermal modeling with ANSYS
  • Breann offered help with connections to her colleagues at GMT
  • It was also suggested that Trenton should reach out to RFCML
Request for Engineering:
  • Heejoo requests Olga put in a request to Barry for a commissioning run
  • Heejoo to continue working with LeRoy to improve channel alignment performance
  • Yang working on improving the options with a single config file as one, centralized location where settings can be adjusted
  • Yang planning to release the missing channel handling, make it switchable, and add the config
In-Focus Images w/ Least-Squares Best Fit:
  • Andrew has been unable to get a hold of Ron, sadly
  • To make progress, Andrew plans to do this with a Zemax model (and possibly with Breann's help to link up Python through the API)
  • By looking at departure from nominal ellipticity over the field, able to obtain surface normals
  • Using double-sided finite difference and damped least squares
  • Andrew suggests sampling of 8 squares per chip on LBC
  • Principle has alread been proven to work
Software:
  • Q: Yang wonders if the missing channel processing should be incorporated into the software yet or not (ability to handle a dropped channel)
  • Andrew notes that, though we want it, we would like to have the ability to turn the feature on and off
  • Yang plans to push the code with the missing channel feature, with the ability to turn it on and off, as well as the config file

-- TrentonBrendel - 27 Jan 2022
Topic revision: r1 - 27 Jan 2022, 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