TMS Meeting Minutes, April 21, 2022

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

Two TMS chances:
  • Pupil images with thermal gradients on the mirror
  • Olga, in response to Heejoo's question via email:
    • FPIA can be run on synthetic pupils generated from Zemax models, with a caveat
    • The Zemax synthetic pupils don't match the on-sky pupils, where John notes that the engineering is what screws it all up
    • Olga's question: When we were setting the magical factors that converts estimates of the pupils, on-sky differential offsets were applied to provide the calibration terms
    • On-sky differentials from pupils historically taken as a truth, where the central hole size was the primary discrepancy
  • Pupil images from Sunday night, where first achieved collimation, then drove on known amounts of Z11 and Z22
  • Olga drove in everything individual, pure Z22 +/-, then pure Z11 +/-; wonders if we should consider doing a mixed run as well
  • Andrew believes that this should be just fine, as Z11 and Z22 are separable
  • Olga thinks we may get into a situation with negative Z11 and negative Z22
  • Andrew suggests it could be interesting to see Z11, then Z22, then Z4 each as a function of the central hole diameter
  • Then, when you do a combination, you should be able to fit to these three terms to get the appropriate hole size for a given wavefront decomposition
Proposed work on Synthetic Pupil to On-Sky Pupil Mismatch:
  • Heejoo, Trenton, and Andrew to sit down and have a working group thinking through the math of Z4, Z11, and Z22
  • Heejoo wonders if their should be a field dependence in these effects
  • Andrew remarks that there is a good paper by Lota Murda about the balancing of aberrations across the field in fast astronomical telescopes
  • When close to the center of the field, fifth order aberrations should grow pretty slowly
  • Andrew proposes bringing on a student to work on this project over the summer; will provide a brief write-up for the project
  • FPIA uses the intensity of Z22 by dividing the field into three equal radii annuli and measuring the distribution across each annulus
  • A totally flat annulus would have a Z22 est. of 2, most of the time it's around 1.7, where the units are in terms of the ratio of the closest minus center divided by outer minus the center
  • The inner and outer hole diameters are only used to set the ratios of equal radii annuli, not to directly measure the Z11
  • This is effectively a distortion correction, measuring the relative size of the hole and the outer diameter of the pupil
  • More discussion...
Plots from Engineering Nights, only One Side:
  • Thermal pupil plots show both the sent Z11 along with the calculated
  • Z11 in yellow is just what's calculated from the ratio of the outer to inner diameters
  • Z11 in crimson includes the crosstalk terms + other offsets
  • Z22 in dark green in light green shows the sent and calculated with only small offsets
  • During this period, FPIA wasn't sending anything, rather we were watching the mirror thermally evolve over time with TMS correcting actively
  • Olga first collimated, then injected Z4 and Z11 and took a TMS reference
  • Then, she ran a script which took a series of extrafocal pupil images with all four chips while TMS was running in active mode
  • Olga verified with a watcher script that we were sending TMS corrections but no PSF wavefront corrections on the mirror
  • It appears that Z11 was bouncing around, along with other aberrations which were bouncing around
  • Andrew suggests that there could be some correlation analysis of ratios in Zernike mean trend, where the measurements are bouncing around the trend
  • Olga is curious about the Z4, where it starts at 1000 nm, which is strange; the whole thing ran for about 2 hours
  • At the end, Olga backed out her Z4 and Z11 correction and took a pupil image, basically ran FPIA again
  • FPIA said that it needed the 1000 nm of Z4 back, which Olga had just removed
  • Olga was able to see a clear pointing change over time in Rx (Z7) which was obvious because a pupil walked into the field from the edge of the chip
  • The strongest trends in the dataset were in Z7 and Z8, possible thermal tilts of the mirror, over a two hour period
  • Started at 3:30 UT, where the glass was a little over 1 degree too cold and had to warm up; with about half a degree sine wave over the first hour
  • The drop was when the ventilation kicked on to warm the mirror; TMS active was doing a fairly good job of controlling collimation even in a bad thermal environment
  • TMS active was simply not doing enough when the mirror was changing too fast
Last Night, Active Mode, Both Sides:
  • Most interesting behavior is between 8-9 UT, where it seemed like left TMS was fighting with the active optics correction and maybe something else
  • There are a row of black X's between 8.2 and 8.5 UT, where left TMS appears to be fighting with the active correction
  • In the next hour, between 9-10 UT, there was a nice trend on the right TMS
  • From Christian's IQ plot, the DIMM seeing was actually much better than what was observed on either LBCr or LBCb
  • Andrew notes that this could be caused by FPIA not converging to a good collimation solution at the start of the night
  • Right at the beginning, when FPIA was running, the seeing was poor, which could result in a poor convergence in collimation
  • Then, right after that, seeing got better, and TMS just held the poor focus FPIA gave it
  • Between 8-9 UT, the DIMM seeing was atrocious at around 2-3", and here TMS actually did a much better job than the guide chip at holding focus
  • Olga notes that there were quite a few times when FPIA was run throughout the night and new references were taken, including at 8.2, and again at 8.6, right when the seeing was really bad
  • Currently, FPIA is not consistently improving the telescope alignment beyond just seeing
  • Christian notes that we don't want to compromise science data in deference to TMS, but we want to allow the system to work the best it can
  • John notes that the observer's log mentioned that they were aware that they didn't have a great reference at the start, and things didn't improve when the DIMM seeing improved
  • Olga remarks that FPIA stated that it was well within the seeing constraints for applying a good correction
  • It is quite possible that FPIA said we had a good correction but we actually didn't
  • Andrew remarks that the way to break through this problem would be to oscillate on either side of focus to get phase diversity
  • Even just having a single shot on the other side of focus would break the degeneracy inherent in a one-sided extra focal analysis
  • Andrew proposes just running FPIA normally and then at the end take only one pupil image on the other side of focus to resolve the inconsistencies
  • This extra shot would provide a significant amount of data to calibrate how much is seeing blurring at the edge and some of the uncertainties
  • Olga thinks there may be a way to remove the Z4, Z11 backout correction, which has always seemed a little bit fuzzy
  • This whole process may add only 3ish minutes of overhead and could result in substantial improvements in FPIA
  • The ambiguity in spherical and focus could be entirely broken with this method
  • Andrew and others to write up an experimental procedure to test this idea on the telescope
  • Andrew to start a conversation on the extra focal pupil
Channel Dropping:
  • Heejoo hoping to write up a procedure to test channel dropping
  • Yang has been working on updating the code
  • Andrew states that, in the current system, TMS assumes you are always returning to your reference with active TMS
  • When a truncated channel drop occurs, in the presence of uncorrelated noise, a big change in the pose change occurs
  • An iterative change in the past delta (which can't be separated from noise) + an error term which can be applied
  • End up taking the truncated set of channels minus the original set of channels and apply that as a correction
  • In the absence of noise, when you drop channels, no error
  • In the presence of noise, on the order of a few microns, you get a jump on the order of hundreds of microns
  • Basically ignore that jump, and subtract it from the original to obtain a corrected pose change with a reduced Jacobian
  • Can think of it like the first channel where you drop a channel is now your new reference + some noise, but the big jump from magnified noise can be removed
  • When you do inverse kinematics on a new Jacobian with a small amount of noise injected, you get a huge magnification of noise; this method solves that problem
  • Yang thinks this should be ready soon, either by tomorrow or Monday
  • Yang to touch base with Heejoo to set up some testing time
  • Closed dome testing with dropped channels and continuous, smooth curves and no jumps is the goal
  • Yang is implementing a new dropped channel simulation where you can iteratively drop and add in channels on the fly
  • Yang to forward email chain to Heejoo for upload to the Google Drive
LBC Time This Week:
  • Ongoing INAF run has more LBC observations ongoing
  • Andrew C. has been conducting these observations, and has been using the TMS without us even asking him to use it
  • Though last night was a real struggle, we have had a couple of really nice nights with TMS running superb
  • Olga thinks that there is an engineering night on Monday; there are other priorities for LBT as a whole, but it could be possible to run TMS

  • TMS SPIE paper was accepted!
  • Important to point out how TMS at LBT is supporting AOM on GMT
-- TrentonBrendel - 21 Apr 2022
Topic revision: r1 - 22 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