TMS Meeting Minutes, April 14, 2022

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

Possible Improvements in AO at LBT:
  • The current voice coil actuators on LBT are a bit outdated compared to the new hardware from TMO
  • The stroke of the current actuators is rather limited
Olga's Email:
  • Olga has been thinking a lot about the problems we are having with collimation
  • It has been costing a lot of time
  • Olga found a whole folder full of synthetic pupil images
  • Andrew wonders if, for the next level of modelling, we can extend to higher orders of fitting by using thermocouple data
  • There are some interesting cases were you could use mirror thermal to diagnose Z22 thermal issues without TMS data
  • Andrew requests data from nights were Olga has pupils and thermal data could be really useful to fit Zernike modes to using Zemax
  • Andrew's impression is that TMS, as it is, is okay to run for a while in the mode that it's running
  • John remarks that at the beginning of the night you can't go too long because you run into Z11 model fitting errors
  • Andrew seeks to prioritize the Z11 fitting model that could be applied as an offset to the TMS correction
  • Possible that there is another systematic error going on that could cause the physical telescope to converge to a different spherical aberration solution than the model
  • There is an elliptical coma term that could be interfering with the FPIA correction, particularly if we are biasing the correction for the middle of the field
  • Andrew is working on making our Zemax model match what we see on the telescope
  • He requests the pupils along with their offline FPIA fits and radial intensity plots, along with the thermal telemetry
TMS Updates:
  • Plan to keep running TMS in the current mode
  • Andrew thinks we could really work on beating down the start of the night accuracy and the dropped channel algorithm
  • The dropped channel algorithm works on paper, but it really doesn't when uncorrelated noise is introduced
  • One thing to look at could be the residuals, which may grow over the course of the night as uncorrelated noise builds up
  • Could run a bunch of matrices wwith different channels dropped to identify the most problematic sources of noise
  • It may be time to consider updating the TMS collimator mounts
  • Andrew has great interest in getting more students to measure the vibration and dome-seeing in the channels during the day (requests student from Heejoo)
  • Peter Gray is very interested in vibration data with TMS (could compare with vibrometer)
  • The Etalon system can control kHz level vibration data, which allows you to measure atmospheric data, it's just a pain to pull out the data from software
  • Each channel can measure amplitude, and to get the directional information, you can use the TMS construction geometry
  • This could be used to compare and correlate the vibrational amplitude measurements directly from TMS against double-integrated accelerometer data
  • With a closed dome, you should be able to see low-order modes evolving over time by fitting sinusoids to the ring of points, enabling the determination of astigmatism
TMS Code Update:
  • John updated the code to permit TMS functionality during extra-focal configuration to compute the pose when the mirror is dropped
  • Andrew would like to request that we use this to take pupil data only while we do a deliberate thermal imbalance on the primary mirror
  • The idea being to drive thermal gradients into the mirror while measuring pupils with FPIA and TMS
  • Olga wondering if we need to constrain the seeing to less than 1-2"
  • For the data to be useful, the seeing should be below the cutoff level of 1.8" for spherical and 1.5" in general, the point at which you only use radially symmetric modes
  • It would also be useful to consider the case of channel dropping, where we have both a bad seeing case and a good seeing case
  • Could run TMS for a while, maybe 10 cycles or so, and then beginning iteratively dropping channels while sensing the bumps in the pose correction
  • Could run in this state with a correction vector for a period of time, then return to the original case with no dropped channels
  • Able to drop channels in either the reference or in the actual live data
  • Currently, when we drop a channel, the algorithm will recalculate the Jacobian matrix by dropping a row, but there will be an erroneous step in the pose
  • To correct for this, a correction factor is applied based on the change in pose from the prior step with the channel to next step with the channel dropped
  • Right now, the channel dropping code only works with probabilities, not with a toggle where a channel is dropped on demand
  • Yang suggests that it could be possible to enable active channel dropping on command by incorporating a polling of the config file while the TMS code loop is running
  • Andrew really wants to test that we don't incur any large, sudden jumps when a channel drop is induced, which could be done in a closed dome
  • Heejoo notes that it is possible that dropping one channel by walking the beam off the collimator upon return from the retro can cause the other channels to change length
  • Andrew remarks that the absolute solution changes a bit, but the relative solution from the reference will be nearly unchanged upon dropped channels
  • Even a millimeter error in the laser tracker position of a fiber tip will result in almost no change in the measured pose relative to the reference, though it will affect the absolute error
  • This fact could have very important consequences for the GMT requirements on the commissioning of the AOM at GMT
  • Breann's pointing model inherently includes a pointing error, and she plans to merge the AOM and the laser tracker models
  • Andrew notes that it is interesting to note how accurately we need to survey the retros and collimators
Engineering Night:
  • 2.5 nights coming up this weekend
  • Friday taken, but Sat. and Sun. there is a chance of using the LBC
  • Plans to look at asteroids
  • Run FPIA to converge, then inject positive and negative Z11 and Z22 to examine pupils, which will only work in really good seeing
  • If we have mediocre seeing, Andrew suggests that we look at pupils all the time while driving in thermal gradients (could significantly improve performance of LBC with well-built model)
  • However, we can't do this we are planning to hand the mirror over to someone else immediately afterwards
  • Andrew thinks that it could take about 2 hours to get 90 degrees on a sinusoid with 2.5 Kelvin drivne into the mirror
  • Olga unsure as to whether or not we need any more details to drive in the temps
  • Andrew remarks that the TO (OSA) should just be able to change the setpoints
  • No gaurantees, but we'll hope for the best
Thermal Analysis:
  • Trenton in contact with the thermal engineer at Steward who is working on GMT mirror thermals for the Mirror Lab, thanks to Buddy Martin

-- TrentonBrendel - 14 Apr 2022
Topic revision: r1 - 15 Apr 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