TMS Meeting Minutes, February 25, 2021

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

Software Update:
  • Tested code, ready to go for on-sky observation
  • Demonstrated smooth operation of basic processes
  • Collected data for Z11, Z22 measurements; Heejoo planning to upload data to Drive
  • Discovered a bug in the program after the test
  • Does not affect the correctness of the program, but it significantly affects the speed when requesting a raw measurement
  • Accidentally included an additional level of implementation into one of the expressions
  • Causes the history loading standard deviation filter to always fail
  • Every time you request a vector, it requests three measurements to build the std. dev., hence the slowness
  • Yang has since fixed the error and installed the new code on the mountain machine, including more on-screen information
  • Information included: current attempts at requesting a vector, how many historical measurements have loaded for the std. dev. filter, how much time is spent in reading a raw TMS measurement
  • When making a reference vector, supposed to be slow, because it reads the history and requests three readings
Etalon Update:
  • Etalon has fixed the MATLAB bug we were encountering; works well so far
  • Releasing a beta version of the software which means we should be able to leave it running w/o freezing up and not delivering information when we try to run Yang's code
  • Should be available next Monday or so
  • Andrew to talk to Etalon about ensuring we have the ability to revert to a working copy if the beta version doesn't work out
  • After a painful yearly cycle of license problems, we have been granted a sort of license rigidity and we want to ensure we don't screw that up
  • If we don't have the ability to have parallel versions installed, at least we'll want a clear pathway for reverting
Engineering Night:
  • Heejoo planning engineering night with Christian for the 27th of February, Friday night
Andrew's Update on Data Processing:
  • Andrew has not yet had the bandwidth to reduce the Z4, Z11, Z22 calibration data from Etalon and Yang
  • Planning to have that done tomorrow morning
  • Planning to use both raw and corrected data from Yang, along with the applied correction, plus the Etalon corrected data
  • Currently using LBT weather station for temperature and pressure correction
  • Andrew requests the ambient air, station pressure, and three links for each measurement with a timestamp, along with the log giving times that Z4, Z11, and Z22 were applied
  • GMT scaling: On meter scales, 4 microbar gives 1 nanometer
  • Meaning on meter scales, 4 millibar gives 1 micron
  • Not anticipating tens of millibar change between outside of dome to inside of dome
  • Andrew's work on noise analysis gives a new perspective on dome seeing that we haven't seen before
  • Uncorrelated noise is on the order of 200 nm, 150 nm at 1 sigma, around the M1 diameter up to the hub, which is very consistent with expectations
  • Looking ahead, once things have become boring and the system is running smoothly...
  • In each set of data, we can actually record something like 20,000 individual length measurements in a burst of 300 ms
  • Would be really interesting to look at correlating with vibration data from accelerometers
  • Could look at separating vibrational components of noise from dome-seeing components in very high resolution to peer into the different aspects affecting image quality
  • Proposing this as a very interesting study once the system is working normally
  • When able to take these 20,000 quick measurements with an amplitude of around 2 nm, effectively frozen the air, so you've got something working like an interferometer in vacuum
  • Once we start collecting the right type of data with the Etalon, should be able to collect at rates of 3 Hz through about 20 kHz
  • Accelerometer samples at 1 kHz, and it rolls off at low frequency
  • Etalon modified their code to make it possible for us to use this huge amount of vibration data, but it's a bit impractical because of the rate of data flow
  • Downsampling to match the accelerometer cutoff would be a good idea to help mitigate this problem
  • Correlated noise from LBT data on averaged measurements is on the order of 1 to several microns, 1 sigma at around 2 microns, which is consistent with physical motion between mirror and the hub
  • A bit larger than the accelerometer measurements indicate, which is around one micron
  • Christian interested in looking at some of the fast sampling files to examine vibrational characteristics, preferably in a few different operational conditions and some with wind
  • Fiber routing on SX side complete
  • Ready to mount collimators
  • Heejoo has designed a couple of collimators based on off-the-shelf designs
  • Has had communication with technical support teams from Edmund Optics and Thorlabs
  • Currently reviewing Heejoo's designs to see if they can simply tweak their hardware to fit our application
  • Based on their evaluation, they will create a quote with 20-30 units of this collimator
  • Heejoo currently waiting on these quotes
  • Designed for high throughput on IR and a 5-6 cm visible spot at a distance of 10 m
  • These suppliers may be able to leverage other degrees of freedom, such as different lens materials or radii of curvature to improve performance
  • Basically a simple tweak that's needed from current designs
  • Questions about the budget
    • Standard collimator around $200 to $300
    • We want to keep it under around $6000
    • Not likely that GMTO will be receptive to spending that much on new collimators at the moment
    • LBTO willing to spend on these collimators
  • Planning to have a prototype model first, then planning to order 20-30 units
  • May be able to make some compromise between collimator performance and price
  • Andrew requests to be in the loop with the new collimator ordering process to see if his contacts might be able to help
  • Heejoo suggests starting a few tests on the SX side with the current off-the-shelf collimators, noting that we may have poor performance to start
  • Andrew remarks that we have had relatively good success with the larger aperture collimators over the smaller aperture models, though not across the board
  • Andrew suggests going ahead and turning off the dome lights and going through the pain of spending a few extra hours aligning just to get things going
  • The current collimators are perfectly capable of making a laser-truss, it's just more labor intensive than it could be
  • Heejoo recalls that we have around nine collimators available
  • Heejoo planning to visit the summit in early March to run some preliminary tests on the SX side and prepare for binocular TMS
  • Ch. 11 is gone, ch. 8 appears to be working; reasonable data for motion of the primary mirror
  • Took ch. 8 out of measurements and then put it back in
  • Took a 7 channel dataset with ch. 8 and ch. 11 dropped
  • Planning to include ch. 8 tomorrow, as it seems to be working fine (ch. 11 is the only one that has sporadic performance dropouts)
On-Sky Plan for Tomorrow (21:35-00:27 UT, on Polycom):
  • Run the software, close the loop, and understand whether or not focus runs away
  • Obtain collimation at a fixed elevation
  • Test 1: Qualitative comparison between IQ of blue and red with TMS running (1.5 hrs)
  • Once collimated, shoot a reference and run the system for a suitable period of time, where the period is determined by when/if we see the red IQ degrading
  • If red is looking good and blue is degreaded, run for a sufficient period of time, perhaps long enough so that there is a clear relation between red and blue IQ
  • Olga notes that there are unexplained reasons why one side degrades faster than the other at times
  • Then run FPIA again and look at what corrections have been applied
  • Andrew would really like to see data where we're able to run FPIA continuously w/o sending corrections, then run TMS continuously while holding things in place
  • If everything is working well, barring shape changes in the mirror, FPIA won't want to make any more corrections, showing that the TMS is doing it's job
  • Test 2: Once test 1 seems to have been passed and IQ isn't spuriosly degrading on red, deliberately send spherical aberration on PSF to make sure that's handled properly (SA should give zero Z change on TMS) (very quick)
  • PSF command of 500 nm of Z11, 500 nm of Z22, along with the necessary change in Z, shouldn't be visible on TMS Z measurement
  • Test 3: Take pupil images and later on process them with VFAST Extra (first with constant elevation, then Test 4: varying elevation if there's time) (1 hr on same field)
  • With a defocused telescope, is TMS able to maintain the given defocus?
  • Relevant to science cases such as transit measurements where a deliberate, maintained defocus is required
  • Test 5: Brief scripts to vary offsets to make sure normal dithering done with optics functions properly (next time)

-- TrentonBrendel - 28 Feb 2021
Topic revision: r2 - 04 Mar 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