Binocular TCS Testing Roadmap
The purpose of this document is to layout the overall plan for testing the Binocular version(s) of the Telescope Control System.
This document is not a specification for Binocular PCS. See the Documents section at the end for additional information.
PCS/PSF Code Checkout
This section summarizes the levels of testing that each build of Binocular TCS (or any subsystem) goes through.
(Apologies to software developers if I have condensed their lives down to a single bullet.)
(Michele will have to fill in the details about how the PCS telescope simulations work.)
Off-sky Tests at the Telescope
See the Testing Plan for each TCS Build: http://wiki.lbto.org/twiki/bin/view/Software/SWTestingStatus
On-Sky Tests with the Telescope
The final round of testing for each TCS Build includes verification of its functionality on-sky.
For Binocular TCS, this will be done for both Prime Focus observing and Bent Gregorian observing, and
eventually Direct Gregorian observing. The details of the on-sky performance tests are described
in the next section.
Performance Tests On-Sky
Binocular Presets and Acquisitions
The Preset(s) should ALWAYS succeed in moving the telescope and optics, acquiring the star, and positioning the target in the focal plane as the instrument requested.
The Preset(s) should succeed MOST OF THE TIME in moving the telescope and optics, acquiring the star, and positioning the target in the focal plane as the instrument requested;
AND should return intelligible error messages when they don't succeed. The time for binocular presets should not exceed the time for monocular presets by more than 10%.
The absolute position of the targets in the focal plane depends on the coordinates of the guide stars and the accuracy of the AGw probes (while guiding). PCS should not degrade those positions.
The test is very simple: send a tuple of Synchronized GUIDE Presets, and see if the telescope acquires the star on both sides. Life gets harder when you have to figure out what went wrong if it doesn't work. Repeat for TRACK, ACQUIRE, ACTIVE and ADAPTIVE presets as appropriate. (Try mixing the preset modes on the two sides if you really want to torture the software.)
Again the test is simple: Send Asynchronous Presets to the two sides with various timings and see if the system responds as expected. Various target coordinate differences should be used to explore the cross-eye/wall-eye co-pointing limits (see the table in 481s066).
Pointing Corrections and Logging
Pointing (Model) Corrections should not be happening during routine observing (until we get automatic guide offloads). So the tests here are to see if IE/CA corrections to the two sides move the star(s) in TRACK mode as expected.
Verify that Pointing Logging writes the appropriate files(s).
The Offset(s) should ALWAYS succeed in moving the telescope and optics, re-acquiring the star, and positioning the target in the focal plane as the instrument requested.
The Offset(s) should succeed MOST OF THE TIME in moving the telescope and optics, re-acquiring the star, and positioning the target in the focal plane as the instrument requested; AND should return intelligible error messages when they don't succeed.
The absolute precision of the offsets in the focal plane depends on the accuracy of the AGw probes (while guiding). PCS should not degrade those positions by more than 10% of the inherent probe performance. This small degradation is difficult to measure because the accuracy of measured centroids of stars in the focal plane depends on atmospheric seeing and wind shake.
- Small (within the co-pointing limit)
- Large (beyond the co-pointing limit)
- Small (within the co-pointing limit)
Binocular guiding residuals should be no more than 10% larger than monocular guiding residuals under the same atmospheric conditions.
Guiding residuals depend strongly on atmospheric image motion and telescope image motion, so it is difficult to make a definite specification here. Also note that guiding at a slower rate may improve the binocular performance (by relaxing the timing issues) while degrading the monocular guiding performance (by not correcting higher frequency errors).
Initial Guiding Function Tests
- different position angles
- different guiding exposure times
- different rates on the two sides
Guiding Performance Tests
- good seeing
- bad seeing
- good tracking (normal pointing model in thermal equilibrium)
- deliberately bad tracking (by upsetting the pointing model)
- long science exposures (up to 1 hour)
Monocular Preset and Acquisition
Future Code Developments
Binocular TCS is being developed and tested incrementally with layers of performance being added in a measured way.
One of the motivations for this is to allow science to continue in a monocular fashion as the full-Binocular TCS is developed and improved.
Fall 2010 should see the last of the monocular TCS releases (AO5) as the binocular TCS releases are verified (starting with BP3).
This section provides a list (incomplete, and perhaps out of order) of what functions will be provided in future releases.
What Build BP3 includes
(This is not a complete list - needs to be refined September 10.)
- binocular PCS and PCSGUI
- binocular presets, syncronized and asynchronous
- binocular offsets
- binocular guiding
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP3
What Build BP4 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP4
We got to the on-sky point of PCS sending optics pointing offsets to PSFL and PSFR on 19-OCT-2010 UT.
What Build BP5 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP5
What Build BP6 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP6
What Build BP7 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP7
BP7 was tested on-sky 2010 Dec 14-15 UT. It is functional for monocular work, but still has problems with the optics offsets for binocular work.
What Build BP8 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP8
Optics Demand Passed from PCS to PSFs
- optics repointing (demand) vectors available from PCS to PSF
- Is there a handshake?
- Who tells M1, M2 when to move?
- optics demand feedback direct (without waiting for actual optics positions)
Leaking Back to the Optimized Position
DJT says: "Leaking fixes everything." But, we have the problem that without fast tip-tilt from the adaptive secondary, the practical leak rate is too small to fix the significant problems. So DJT is also correct when he says: "Leaking does nothing." We should get the binocular and locking algorithms working correctly before we add leaking into the mix. What leaking actually does is drift the optics back toward a well-centered position regardless of what type of
dither sequence the observer does.
What Build BP9 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP9
BP9 was tested on-sky 2011 Feb 12-13 UT. It can successfully manage binocular ACQUIRE presets at bent Gregorian, but there are issues
to be resolved relating to the timing of PCS-PSF-GCS interactions when moving the optics.
Additional tests on patched BP9 were done on-sky on 2011 Feb 23 UT. Successful binocular synchronous and asynchronous presets and
offsets were demonstrated including acquisition, guiding and active optics in an "unlocked" mode with leaking enabled. PCS supplied optics tip-tilt updates to PSFs at 5-sec intervals. Some preliminary tests were done in "locked" mode.
What Build BP10 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP10
Binocular TCS testing of version BP10 resulted in a successful pair of
synchronized active presets sent from the left and right sides on 23-March-2011. TCS
Build BP10 was also put into science operation for monocular Gregorian and
binocular LBC work.
What Build BP11 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP11
BP11 was tested on-sky 16-21 May 2011. Tests included Guiding LBC in full binocular mode, and on each side individually.
Coordination between PCS and the PSF was smooth.
Weather and other technical problems limited the amount of binocular testing that could be accomplished. See Issues 3372 and 3374.
What Build BP13 includes (there is no BP12)
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP13
What Build BP14 includes
See the Release Notes: http://wiki.lbto.org/twiki/bin/view/Software/SWRelBuildBP14
What Build 2012C includes (2012B was not released)
See the Release Notes: http://wiki.lbto.arizona.org/twiki/bin/view/Software/SWRelBuild2012B
See the Release Notes: http://wiki.lbto.arizona.org/twiki/bin/view/Software/SWRelBuild2012C
What Build 2012F includes (2012D and 2012E were not deployed)
See the Release Notes: http://wiki.lbto.arizona.org/twiki/bin/view/Software/SWRelBuild2012D
See the Release Notes: http://wiki.lbto.arizona.org/twiki/bin/view/Software/SWRelBuild2012E
See the Release Notes: http://wiki.lbto.arizona.org/twiki/bin/view/Software/SWRelBuild2012F
Locking Sides, or Not
Verify that the "locked" side (the one possibly integrating) is not disturbed when the other side does an offset.
This has been tested on-sky in March 2012 (2012C) with LUCI1 offsetting while LBC-Red is taking data. Additional tests are needed with additional configurations.
Additional tests of locking and leaking were carried out during April 2012 during AZ LBTI observations. Things worked well except for the
tertiary mirror collimation.
Range Centering and Collimation Limit Avoidance
This involves integrating the "preferred" range-centered position of the optics calculated by PSF back into the PCS pointing calculation.
This provides a mechanism for offloading common hexapod displacements back onto the mount. Each PSF should provide a range-centered estimate of where it
would like to be for pointing. PCS will then include the common part of those two vectors as a bias in its optics centering calculation.
The Range-Centered condition should be the best that you can do to avoid collimation limits, so further improvement needs
repointing schemes beyond Mode 1. See below, and see 481s066.
The PSF part of this was implemented in BP13 in October 2011.
The PCS part is planned for the next TCS build.
Mixed Instrument Binocular Mode
With the advent of TCS build BP10, binocular observations with LUCI1 and LBC-Red are at least possible for exoplanet transit observations.
LBC software is ready (March 2011) for binocular observations since it automatically sends the Binocular Control commands when only authorized on one side of the telescope. For the 19-May-2011 version of the LBC code, the BinocularControl command is sent for preset and all sizes of offsets. (LBC always does synchronized offsets, but the IIF can be set to ignore the BinocularControl commands.)
Marcus J. and Tom S. made a version of the monocular LUCI software that also sends BinocularControl commands in front of presets, but not offsets. (LUCI1 always does asynchronous offsets.) This was tested for monocular operation in May 2011 and released for monocular science.
While the parts have been tested individually, the combination of LUCI1 and LBC-Red remains to be tested on sky.
This was planned for September 2011, but it rained alot. On-sky tests were carried out in March 2012 with Build 2012C without problems.
Below elevations of ~50 degrees the LUCI1 and LBC-R pointing models are different enough to cause problems. A manual pointing adjustment to LBC-R Primary Mirror is needed to maintain collimation range.
Olga and John tested mixed mode with apparent success on 20120511. But Olga and Dave had problems on 20120607 running into M1 limits, so there still may be issues to resolve.
Optics Position Feedback to PCS
We started with the demand optics tip-tilt being fed back into the pointing calculation(s). We can further improve the situation by feeding back the actual positions of the optics pointing (only the pointing part of the collimation, not Total Collimation). This makes the mount more synchronized with the optics (although M1 and M2 are not well synchronized with each other).
What happens when M1 or M2 is busy from a previous command?
Managing non-sidereal observations will be included in the binocular development at least to the point of making the same non-sidereal observations on both sides. (Mixed sidereal and non-sidereal observations are not considered high priority, although nothing in the PCS kernel explicitly prevents them.)
Implemented in October 2011 with BP13, tests pending.
I'm declaring that the following additional features are beyond the scope of Phase 1 Binocular development.........
The most important one will be OPD aware collimation which will be a Phase 2 Binocular development.
Guiding with the Adaptive Secondary Shell
Given the present (Feb 2011) response time of the M2 hexapod, optics pointing corrections for guiding can be done no faster than every 5 seconds.
If we are able to make the guide corrections with small tip-tilts of the adaptive secondary shells, then
guiding can proceed at any rate permitted by PCS-AOS (up to a few Hz) and the guide cameras without any additional
delays for mirror motion. Of course, this only becomes interesting when we have two adaptive secondaries, and when we are observing
in seeing-limited (flat shell) mode.
Other guiding improvements include such things as non-linear guider gains, although this is not exclusively a binocular development.
Non-linear guiding gains were implemented in 2012F in June 2012.
Beyond Mode 1 Re-Pointing
This involves adding the smarts to PSF so it smoothly transitions into Mode 3b pointing (or whatever mode) as Mode 1 pointing
begins to run out of range. See 481s066 for defintions of the pointing adjustment modes.
Dynamic Pointing Change Limits
Replace the fixed (configurable) Pointing Change Limit in PCS with Dynamic Pointing Change Limits that are computed by PSF.
(This item may be dropped, as we feel it is important for OB preparation to have deterministic limits. That means the dynamic limits
have little value for science observations. The math is mostly done when we implement range balancing.)
OPD Aware Collimation
In this case, we are moving beyond Modes 1,2,3 to provide repointing motions which are OPD neutral.
Calbrations which are Independent of TCS Code
In parallel to the binocular-aware machinery of the TCS code, there are a number of calibrations that are required to make the binocular
TCS work efficiently on-sky. These are summarized in this section.
Update for New Telescope Loading
With the addition of MODS1 and LBTI (and the corresponding counterweights) to the telescope over summer 2010, the pointing and collimation models
need to be updated to accomodate the different gravitational flexures. To start with, these are traditional monocular pointing models. For LBC, this also should include offsets to set the co-pointing at high elevation.
Update Collimation Models for new telescope loading conditions - September 2010
Update Pointing Models for new telescope loading conditions - September 2010
Update for Laser Tracker M1 Positions
This activity is to make the two primary mirrors point better coaxially with the mount. Monocular pointing models tend to be degenerate
between optics and mount pointing, although we are within an arcminute or so.
DavidThompson summarizes the steps in a recent (18-August-2010) email. These steps are paraphrased from that
Measure M1 Positions relative to Instrument Rails (Mount)
- 1a: Laser Tracker: M1-Instrument rail survey
Update Collimation Models for Laser Tracker based M1 Positions
- Generate new M1 collimation models from the survey data.
Update Pointing Models for Laser Tracker based M1 Positions
- 1b: Make LBC pointing models using the new M1 collimation models
Results of steps 1a-b are primary mirrors that are reasonably co-pointed through their laser-tracker collimation models, and a "mount" pointing model that applies to both telescopes simultaneously (suitable for the next step).
Dec 2010: This activity has been overtaken by range balanced collimation models constructed from mining historical data.
Update Gregorian Collimation Models On-sky
- 1c: Make M2-M1-Instrument rail collimation models (need to do for both sides at Bent Gregorian) using pointing-free coma corrections measured on-sky
At this point we have telescopes that are reasonably co-pointed and independently collimated.
- Adjust LBC M1 collimation models to make LBCs precisely co-pointed at high elevation and adjust or remeasure the LBC pointing models.
Update for Laser Tracker Thermal Corrections
Update Collimation Models for Laser Tracker based thermal corrections
Range-Balancing the M1-M2 Collimation Models
- 2: Range-balanced binocular collimation and pointing models
Astronomically make finer adjustment corrections on sky. While co-pointed and co-collimated also insure optics in middle of optics movement range by applying coordinated Mode 1 pointing corrections to the collimation models, and redoing the pointing models. This is where we tie the two sides together in an astronomically relevant and accurate
sense for a binocular telescope. The laser-tracker measurements may only have the two sides co-pointed to some ~few or more arcseconds. After these new pointing and collimation models are made, the only remaining deviations to co-pointing should be from un-modeled thermal or flexure issues. Hopefully those will be below the remaining range of motion possible with the optics.
BinocularPresets.pdf (not archived as of 27 May 2010)
(441 kB, 7/10) The ICE Instrument Interface Control Document (See section 6.6)
Software Testing to Achieve Binocular Operations (Rev B, 18-Nov-2010)
Arbitrator Design Document' (version of August 2010)
(102 kB, 3/11) Generic Binocular Observing Use Cases: Operational Philosophy
(472 kB, 3/11) Lucifer Use Case Details (flow charts)
(750 kB, 4/11) 11x17 versions of the flow charts from 481s621
(4061 kB, 2/11) Binocular Observing 101: A guide to binocular observing at LBT
(274 kB, 6/06) 'Active Optics and Pointing on the LBT' 'LBT-DLT-001' '08/06/2006'
(182 kB, 8/06) 'Binocular Pointing Kernel Design PCS' 'LBT-DLT-002' ''
(232 kB, 5/07) 'Pointing Algorithms for Binocular Telescopes' 'SPIE 6274-16' ''
(120 kB, 9/09) 'On the Path to Binocular Pointing with LBT' 'ADASS XVIII' ''
(132 kB, 8/10) 'Binocular Pointing' 'LBT-DLT-003' ''
(69 kB, 7/10) 'Practical Considerations for Pointing a Binocular Telescope' 'SPIE 7740-90' ''
(545 kB, 10/11) 'Range-balancing the Large Binocular Telescope' 'SPIE 8128-07' ''
(599 kB, 10/11) 'Maintaining hexapod range while co-pointing the Large Binocular Telescope' 'SPIE 8131-12' ''
Binocular Code in PCS
The PCS code for 2012E is located at https://svn/view/tcs/branches/2012E/pcs/
Look at the routine TrajectoryGenerator.cpp which essentially is the fast loop and a few helper routines.
Appendix: Stages of Binocular TCS Development
-1) Monocular TCS (one side only) -- deployed 2005
0) Monocular TCS (side aware, but still one side only) -- deployed 2007
Master-slave guiding was implemented for binocular LBC observations
Binocular Phase 1
(Apologies for multiple redundant numbering schemes.)
1a) Binocular TCS (single target, asynchronous offsets and guiding, stepwise updates to optics) -- deployed March 2011
1b) Binocular TCS (synchronized or asychronous presets and offsets, stepwise updates to optics) -- deployed May 2011
1c) Binocular TCS (including range-balanced collimation) -- planned for Fall 2011 / Spring 2012
Binocular Phase 2
2) Binocular TCS (with OPD-aware optics motions, stepwise updates to optics, platescale adjustment) -- requirements TBD
Binocular Phase 3
3) Binocular TCS (with smooth trajectories of all optics) -- requirements TBD
- 02 Sep 2010