20200924 TMS Meeting Notes

[trentjbrendel]

Andrew's Results

• Mathematica results
• Took the nominal vectors and rounded to the nearest 100
• b1-b14 nominals all rounded to the nearest 100
• Used Yang's Mathematica code to test channel dropping
• Took 4 consecutive lines of data from eTalon output
• Nominal is a very different vector (formerly based on laser tracker measurements)
• Computes first pose acceleration (change in the change of pose) for nine channel change in pose between nominal and IP1
• Then computes for 8 channel, 7 ch., 6 ch., etc.
• When you do your first dropped channel, massive errors, like a reset, zero offset (like degrees and meters of error)
• Take this error and accept as constant for subsequent meas.
• If you use this as a constant correction, makes errors much smaller on continued dropped channels
• Basically, if we drop a channel, do the 8 ch. change vector, don't apply it immediately, but apply it to the next 8 ch. measurements
• In 7 sec, assume essentially no change
• In instances where a channel is dropped once, we just miss that measurement, basically
• All we've done is effectively missed one step
• If we calc pose w/ 8 ch., there will be a large constant offset
• John suggests using the previous 9 ch. meas. to calculate the offset, not the 8 ch.
• AKA Use last good measurement to apply correction to 8 ch. measurements
• In this case, we would not miss a step
• In the case of an 8 ch., just use the last good 9 ch. (or last two good 9 ch.'s)
• If we are off by about 1.5 deg. C, the Ciddor correction becomes very important (for 0.5 deg C, not as much of a problem)
• Have to be careful when experiencing large Jacobian changes

Yang and Andrew's Discussion:

• Evaluating image stability best done with WFS data

John's Evaluation of the Tool:

• Some non-physical telescope optical phenomenon observed on sky (2018)

Engineering Time w/ LBC on Sunday:

• Have FPIA and LBC running
• Andrew suggests we would like to test channel dropping before that
• Yang not able to make the time, lots to work on
• Andrew suggests using guide cam images, centroiding, and comparing these with what our measurements would drive mirror to
• However, not able to differentiate the actual telescope tracking, atmospheric image motion from pose changes
• We can show the non-existence of [some sort of problem]
• Can apply difference in channel number to the pose reconstruction after the fact, with a real star
• Normally, when running LBC, guide camera images go to DMS and centroiding is performed
• Sorry, the guide camera corrections are stored on DMS forever, not the guide camera images
• Just need to know the data range to look at the guide camera correction data
• So what to do with Sunday's engineering time...
• Yang suggests passively collecting TMS data for offline analysis
• There are some LBC image issues, and the engineering team is wondering if TMS team can help
• Apparently there are some ongoing LBC image problems which need solving
• Yang also mentioned FPIA problems... we want to fix this w/ our intra/extra focal method