NRAO Home  >  Green Bank  |  Wiki Topic:    GB > PTCS > InformationForProjectTeam > PTCSProjectPlan > OOFDevelopmentFall2006

Changes | Index | Search | Statistics | Go

Continued Out-of-focus (OOF) Holography Development Winter 2006 / 2007

• November 2006 OOF Confirmation Experiment
• Spring 2006 Status and Overview
• The Approaches
• The Big Picture
  • Short term (Winter 2006/2007)
  • Intermediate term (Summer 2007 through Winter 2007/2008)
  • Long term (vague and ill-defined)
• Discussion topics for the November "mini-OOF workshop"
  • Question and Answer Session
  • Goddard approach
  • Data Formats and shared pre-processing software
  • Scanning strategies
  • "Big-bang" data fitting, different basis vectors.
  • Use of Ka-band receiver
  • Use of the Penn Array
• Detailed Goals, Tasks and Timescales for Short-Term (Octonber 2006 through July 2007) Development

November 2006 OOF Confirmation Experiment

Spring 2006 Status and Overview

Out-of-focus (OOF) holography is now an integral part of the overall PTCS system to deliver 50 GHz, and ultimately 115 GHz (3mm) operation. As of Spring 2006, we have a stable mechanism for performing and applying the results of OOF holography; this has allowed us correct for large-scale static deformations (at the riggling angle) and large-scale elevation dependent deformations (gravity). The technique is fully capable of measuring large-scale deformations due to thermal gradients, but the current approach is too slow to allow "real-time" correction of these.

Application of OOF Holography has allowed us to reach the "phase II" goal of acceptable 50 GHz operation under benign night-time conditions. The Spring 2006 status of the oof holography work is summarized in the following links:

• OOF paper I (to be attached)
• OOF Paper II (to be attached)
• PTCS/PN/47: "Out-Of-Focus Holography at the Green Bank Telescope, The Gain-Elevation Curve and Absolute Efficiency", Nikolic, Balser, Prestage, June 2006
• PTCS/PN/50: "Astigmatism and Thermal effects at the GBT", Nikolic, Balser, Prestage, June 2006
• PTCS/PN/51: "Preliminary Analysis of the Correlation Scales of Wavefront Errors of the GBT Using the Moon", Nikolic, Balser, Prestage, June 2006 (Not really OOF results, but related).
• Current Performance and Active Surface N.B. The instructions on observing procedures is out of date, and needs to be corrected!
• AntennaCharacterizationAstridScripts descibes current Astrid Scripts; does this include performing OOF mapes?
• Add other good links here....

As of now, the system is stable, and the gravity model can be applied as-is once we confirm this fall that the existing coefficients (e.g. 2005WinterV3) are still valid (assuming that indeed they are).

Although the system is stable, there are numerous enhancements and new developments we would like to make to improve the capabilities of the OOF system. The purpose of this wiki page is to outline what these required enhancements are, and to start a frame work of specific goals, tasks and timescales.

The Approaches

Currently, we are using the techniques (and software) developed by the Cambridge, UK group and described in the above links. We have recently been approached by staff from the James Webb Space Telescope Wavefront Sensing and Controls (JWST WFS&C) Group. who would be interested in trying to apply their technique to our data. (See Phase retrieval algorithm for JWST Flight and Testbed Telescope by Dean et al.)

In what follows, I'll use "OOF holography" to refer to generically to either of these approaches, or indicate explictly when referring to a specific approach.

The Big Picture

When we initiated the OOF project, we were also pursuing traditional (phase-reference) holography, and the laser rangefinders. In some senses, the OOF technique was considered simply a stop-gap solution. However, as with the pointing work, the system design has evolved to a state where astronomically-derived measurements of the figures and collimation of the various reflectors (both during dedicated commissioning perids, and in real-time) have become an integtral part of the overall system. This change in philosophy is important in considering continued developments. We believe there are considerable improvements to the technique which are still possible. Without getting into the details at this stage, examples include:

• As the servo performance is improved, we may be able to significantly improve (shorten) the time required to obtain OOF maps, even while continuing to use the existing Q-band receiver, by performing more efficient "daisy-petal scans". (Currently, servo performance is too poor to get good coverage of the peak of the source at the center of the daisy-petals.)
• By using an improved basis set (e.g. including the predicted effects of translating the subreflector explicitly in X and Z), and revising our observing / fitting strategy, we could come up with a single measurement/analysis cycle which simultaneously measures and solves for pointing offsets, radial focus offsets, subreflector mis-collimations and residual large-scale error (modelled as low-order zernike polynomals as now). This might allow us to reduce the ~ 40 minute peak/focus/oof-map observe/reduce/apply cycle to say a ~15 minute cycle which replaces all of these steps.
• Using the Penn Array might allow us to obtain sets of OOF maps in a few minutes, and potentially achieve improved accuracy and higher spatial (on the surface) resolution.
• Using the Ka-band receiver and the Caltech Continuum Backend, and simultanously fitting to the four 4 GHz chunks between 26 and 40 GHz might allow quicker observations or better results (we essentially dropped our attempts to understand Ka-band, since switching to Q-band seemed a better short-term approach). Is the Ka-band receiver less weather sensitive than the Q band? A further advantage if we can get this working with the Ka-band receiver is that Ka-band observers could use the receiver in place for real-time corrections, rather than the potentially unacceptable overhead of switching receivers (although by then, we might have the turret rotator upgrade...)

I propose a three-stage, phased approach, as described below.

Short term (Winter 2006/2007)

• Continued investigation and incremental improvements to the existing approach:
  • Commissioning periods are used to obtain/validate current format gravitational models, which are automatically applied via look-up tables.
  • We investigate/demonstrate more effective scanning / analysis strategies with each of Ka, Q and Penn Array.
  • We prototype a "real-time" correction approach using one or more of the above receivers which will allow a "real-time" (goal of say 10 minutes total) measurement and application of pointing, radial focus, collimation and large-scale errors simultaneously.
  • Summer 2007 we develop production code to allow the combination of a gravitational model with a real-time thermal correction.
• [Simultaneously, we will be attempting to characterize small-scale errors via "traditional" holography, photogrammetry, or another technique TBD]

Intermediate term (Summer 2007 through Winter 2007/2008)

• Continued incremental improvements to the existing approach:
  • Winter 2007/2008 we continue to refine/implement measurement strategies using one or more of the Ka, Q, Penn Array and potentially the W-band receiver.
• Investigate possibility of a dynamic (temperature based) correction system:
  • Complete the disentanglement of true (main axes) pointing and collimation-dependent pointing effects, so we can improve our pointing models as well as mirror alignment.
  • Develop a system which is capable of predicting required subreflector displacements from measured thermal gradients, to produce a "dynamic thermal correction system" analogous to the current "dynamic pointing correction system" - i.e. having trained the system, use real-time temperature sensor data to predict/correct thermal mis-collimations (or equivalent distortions in the primary) without the need for real-time OOF maps. This might correct for coma and astigmatism, or even go to higher order.
• [Simultaneously, we will be correcting small scale errors using the successful technique demonstrated earlier.]

Long term (vague and ill-defined)

• Develop a scheme which combines OOF holography, thermal sensors, the small scale measuring technique, and laser rangefinders to deliver the ultimate in performance. Hope (or plan) that the earlier developments flow reasonably seamlessly into this approach.

Discussion topics for the November "mini-OOF workshop"

The Goddard phase-retrieval guys are visiting November 20 through 22nd. Bojan at least has agreed to be available for some telecons. We have some limited commissioning time in December when we could obtain some new data if necessary, but we still have a lot of mileage to go in the current data. I propose as a starting point the following topics for discussion.

Question and Answer Session

Answer any questions Goddard guys might have about GBT Optics, data formats, etc. Give them a tour of the telescope.

Goddard approach

Can the Goddard guys get the same answer that we get for the same data-set? If not, why not?

Data Formats and shared pre-processing software

We should adopt a single approach to data pre-processing (interpolation of antenna positions, calibration, etc). We should build on what Bojan has developed, perhaps also Bill Cotton's OBIT software for making (in-focus) beam maps.

Scanning strategies

Can we agree an improved scanning strategy, e.g. some sort of daisy-petal scan, which will get the required spatial coverage faster?

We should simulate this before we take any new data. This could either be a purely hypothetical simulation, or we could take sub-sets of the existing, fully-sampled data, process those as if they had been produced by a different scanning technique, and see if we get the same answer.

Note that the daisy-petal approach might introduce other issues, e.g. swaying of the feed-arm.

"Big-bang" data fitting, different basis vectors.

We should be able to use the OOF data sets to get all of pointing, focus and large scale error simultaneously. We might also consider fitting directly to the functional forms which might be expected by specific geometrical distortions (x,y,z displacement of the subreflector, etc).

Use of Ka-band receiver

We dropped the Ka-band receiver data since we couldn't (at the time) distinguish the different phase response of the Ka-band and Q-band receivers from actual large-scale error, and we wanted to use Q-band to do absolute (43.1 GHz) efficiency measurements (didn't know flux calibrators for e.g. 38 GHz). Now we understand the large scale error better, we should be able to do better with the Ka-band receiver. It seems like having four simultaneous frequencies, with the highest frequency close to that of Q-band receiver, and a 4:3 ratio in frequencies, should be helpful. Presumably we could fit for the changing phase response as a function of frequency (which should look something like a defocus) by requiring the large scale error to be fixed across all four frequency channels. we should discuss this further.

Use of the Penn Array

At the moment the Penn Array is not available - it will be off the telescope until next fall for engineering work. We did get one example data set on Saturn, which reminded us that there will be optics issues - e.g. the beams will get vignetted when the telescope is out of focus. But, we could discuss whether this receiver is worth pursuing further at this time.

Detailed Goals, Tasks and Timescales for Short-Term (Octonber 2006 through July 2007) Development

• Upate istructions to observers on how to use Zernikes
• Checkout Fred Schwab's new actuator zeropoints
• Finish two OOF papers
• Fix all known bugs in the Active Surface Manager and lower-level actuator control software.
• agree the intermediate format pre-processed data and file format, which by default should be the current OOF format. This is essentially "calibrated" data, (possibly in arbitrary units), in a binary FITS file, with each row being something like:
  • az (or ra) offset
  • el (or dec) offset
  • data value
• This will allow the analysis groups to use the identical input data, and free them from the tiresome house-keeping chores (merging GBT Engineering FITS files, etc).
• Develop GBT-supported standard software for converting raw data from any GBT receiver/backend (inc Penn Array) into this format.
• Do gain-elevation measurements at Q-band under excellent conditions to confirm the Winter 2005 gravity models are still valid.
• Acquire and process some Penn Array data
• Acquire and process some more Ka-band data
• Design the new and improved observing and analysis strategies
• Execute the new and improved observing strategies in the December commissioning period.
• Develop and test the prototype "real-time" correction observe/reduce approach and software.
• Prototpye software that fits for and corrects for mis-collimation in the subreflector rather than the primary (is this useful?)
• Convert all prototype software from above into production software supportable and executable by GBT staff
• Develop the implementation to allow the combination of gravity and "real-time" zernike corrections.
• Consider enhancements to the Active Surface control software (e.g. an improved actuator servo system).

At the end of the above, we have production elevation-dependent lookup tables, and a production real-time correction technique ready for the Winter 2007/2008 observing season.

-- RichardPrestage - 13 Oct 2006