NRAO Home  >  Green Bank  |  Wiki Topic:    GB > PTCS > InformationForProjectTeam > PTCSProjectPlan > PointingFocusDevelopmentFall2006

Changes | Index | Search | Statistics | Go

Continued Pointing/Focus and Antenna Characterization Development Fall 2006

• Introduction
• Summary of Current Status and Approaches
  • Development and application of pointing/focus models
  • Other commissioning experiments we do/have performed.
• Required and Desired Upgrades
• Proposed "biggish picture" development plan - through Fall 2007
• Possible Pointing Experiments
• The summer shutdown
  • During the track work
  • After the track work
• Proposed upgrades to the "Antenna Characterization Database"
  • General Considerations
  • Specific Proposal
• One Year Goals, Tasks and Milestones

Introduction

The GBT Pointing/Focus models are associated software are currently stable. As we are now entering a new phase of PTCS development, it is worth laying out the plans for the next stage of model development, and this wiki page is an attempt to do so. This page is primarily concerned with the near-term development and application of pointing and focus models, using astronomical measurements, coupled in the future with inclinometer measurements to address track irregularities and other effects. The intent is to lay out short to medium (~ up to one year) developments which will be undertaken primarily by GBT staff. At the same time, Kim Constantikes is prototyping and developing more sophisticated approaches; as these are firmed up they will need to be merged in to the production software.

As well as explict pointing model development, we anticipate renewed efforts will be put into characterizing various aspects of antenna performance - e.g. slow-speed servo tracking capabilities. This will require considerable engineering and technical work independent of the "astronomical" work. The main overlap here is the need to use the antenna to acquire data, potentially simultaneously with astronomical commissioning experiments, and the development of some sort of database to at least index, if not store these data, along with derived results.

The prime focus (sorry) of this page is therefore to discuss experiments which will require antenna commissioning periods, often with associated direct astronomical measurements, and how we will execute these measurements (development of observing scripts for example), process, reduce, store, archive and index the resulting raw data and derived results. Future work should build on existing techniques and approaches, which have proved extremely successful, and a secondary goal of this page is to summarize those approaches and techniques.

Comments, corrections and even major changes in content or emphasis of this page are welcome.

Summary of Current Status and Approaches

Development and application of pointing/focus models

The approaches, algorithms and software implementation to produce the current pointing models have been built up over a number of years. The basic approach is now as follows (glossing over many, many details):

• A series of pointing runs, including all-sky pointing, "single source tracks" and "North Celestial Pole" experiments are executed, performing multiple successive jack scans (peak followed by radial focus scans). Astronomical data are logged as well as Archivist-generated FITS files for a wide range of ancillary data, including structural temperatures, wind speed and direction, and so on. Structural and air temperature data are collected by the Antenna Characerization Manager. A log of all commissioning runs is maintained at CommissioningHistory (note this is in principle every commissioning run, not just pointing). A classification of commissioning runs by experiment type is maintained at CommissioningExperiments, although this is out of date and needs to be updated.
• The astronomical datasets and ancillary Archivist-generated FITS files are processed by Dana Balser via glish scripts to produce a set of comma-separated-variable text files, which collectively form the AntennaCharacterizationDataBase. Note that the contents of the ascii text files are described in tables at the end of this page.
• Although the implementation is different, the general approach to fitting the pointing data is described in PTCS/PN/15.
• The Antenna Characterization Database is imported by Kim Constantikes into an Access database. Kim then uses matlab applications to create fits for both pointing and focus to a gravity model, a thermal model and an "empirical residuals" look-up table. Data may be selected on the basis of other factors (time of day, wind speed, etc).
• The thermally neutral gravity model and the thermal corrections are described in PTCS/PN/25
• The results of the modelling are stored in the Pointing/Focus Model (PFM) database. Currently, we use model 4c. The PFM is described at AntennaManagerUpgradesFall2003, although this is now hugely obsolete - is there a better link? I now curse myself that I never finished the one outstanding PTCS Project Note. Would someone else like to do it?
• The thermal model coefficients, together with the current temperatures are used by the Antenna Characterization Manager to create "dynamic corrections", which are sent in real time to the Antenna Manager, and used to continually update the pointing.
• The Antenna Characterization Manager and dynamic corrections system are briefly described in PTCS/PN/17 - is there a better link?

The net result of all of the above is that we have a combined pointing/focus/track model which, together with dynamic pointing/focus corrections, largely meets our 50 GHz goals, and meets our 90 GHz goals under benign night-time conditions. Currently, performance drops off during the day (dynamic corrections for pointing don't perform so well, although focus does remarkably well), and under windy conditions (we are not compensating for wind). Under benign, night-time conditions, we currently believe track irregularities are the largest single contributor to the pointing error budget.

The current (August 2006) performance of the antenna is described at PTCS/PN/49 and numerous references therein.

Other commissioning experiments we do/have performed.

As well as pointing runs, we frequently perform half-power tracking experiments to characterize the tracking performance of the antenna. See PTCS/PN/49 and e.g. PTCS/PN/19. A variety of other more specialized experiments and analyses are desctibed in the PTCS ProjectNotes series - anyone interested in GBT pointing and antenna performance in general needs to carefully scrutinize the documentation index, and read the appropriate memos.

Required and Desired Upgrades

The current system is extremely stable, and has delivered consistently good pointing performance for the last few years. Nevertheless, numerous upgrades are required. In rough priority order from an operational perspective (rather than a scientific priority), these include:

• The pointing/focus data is currently processed by Dana Balser (who is moving on to other activities) using Glish scriprs (Glish is now essentially obsolete as far as Green Bank is concerned).
• The pointing/focus data is analyzed by Kim Constantikes (still available as a consultant, but no longer on our permanent staff) using a custom Matlab application (Matlab is a fully supported commercial product, but only Kim knows the details of this application).
• We intend to destroy and rebuild the complete track in Summer 2007, so we can anticipate major changes in pointing performance.
• The current implementations, were built up over time, in many cases while we were exploring alternative approaches. For exampke, the Antenna Manager now explictly calculates many required quantities which were not available in earlier versions. We should rationalize our analysis software accordingly.
• We would like to start incorporating additional data into our analyses. Specifically contemporaneous Inclinometer data, probably also Accelerometer and Quadrant Detector data. These devices were not uniformly available and/or logged in earlier pointing runs.
• We would like to develop the "next generation" of modelling and fitting.
• The current infrastructure for maintaining our knowledge of, and data from, commisioning runs (e.g. the CommissioningHistory wiki page, the ascii text file Antenna Characterization Data Base) is in many cases incomplete and/or marginally adaquate for the task. We should put all of this on a much firmer footing.

In addition to the above points, the current infrastructure is very much dominated around the concerns of processing pointing/focus data. There is some infrastructure to support half-power tracking data; almost every other experiment has been supported in an ad-hoc way. While not strictly necessary for the continued development of pointing/focus models, I believe it would be advantageous to broaden our infrastructure to support most, if not all antenna characterization activities.

Proposed "biggish picture" development plan - through Fall 2007

I propose there should be three threads of effort, which may be executed in parallel:

• GB staff develop the new standard infrastructure to support processing of pointing/focus data.
• Kim Constantikes develops the "next generation" fitting software to include more sophisticated treatments of track effects and alidade distortions. This is merged into the "production" software as appropriate.
• The antenna characterization database, and associated tools, are developed to provide infrastructural support for additional antenna characterization experiments (e.g. half-power tracking).
• Neeed to develop more complete error budget, and start specifically testing against this.

These should culiminate in a prototype test of the new pointing modelling software in Spring 2007 (before the track shutdown), and production use of the new pointing modelling software in Augusyt 2007.

Possible Pointing Experiments

With astronomy data:

• all-sky pointing: look at a mesh of pointing calibrators well distributed over the whole sky.
• single-source tracks: continuous peak/focus measurements on a single calibrator as it rises, transits and sets. Repeat with sources that transit high and low in the north and south.
• North Celestial Pole experiments: continuous peak/focus measurements on a single source very close to the NCP, so that the telescope is an an essentially fixed azimuth/elevation.
• Condon double/triples - point on a nearby calibrator (or calibrators) after having peaked up on a "reference" calibrator - I think this is our best test of true offset-pointing capability.
• half-power tracks: continous data-acquisition while the telescope tracks the half-power point (in azimuth and then in elevation) of a strong calibrator. Variations in antenna temperature are assumed to be due to variations in pointing.
  • With X-band and DCR as now
  • With Ka-band and CCB, which has a true beamswitch for atmospheric cancellation, and four 4 GHz channels. Should be a much more discriminating test.
• half-power moon scans: Do the same on the limb of the moon, provides a knife-edge with extremely high signal. This should be the most sensitive test of tracking of all - why don't we always do this?
• pointing with Penn Array. Should be able to instantaneously sample a bright calibrator, so will provide fully-sampled high-time resolution equivalent of half-power tracks.
• Servo performance tests - e.g. step demands onto a point source (or limb of moon) to look for overshoot and ringing.
• We should use the moon more!

Without astronomy data:

• servo performance tests
• dedicated inclinometer and environmental data acquizition while telescope motion is under our control.
• serendipitous inclinometer and environmental data acquizition while the telescope is performing normal astronomical obvserving

The summer shutdown

During the track work

After the track work

This time could also be used for non-pointing PTCS work (e.g. OOF holography), although it will still be summer, so the weather may rule this out.

Proposed upgrades to the "Antenna Characterization Database"

General Considerations

The current database is structured primarily around 30 seconds of time peak and focus subscans, which are then characterized by the UT time for the mid-point of the scan, deduced pointing paramters (e.g. measured azimuth/elevation/focus local pointing corrections, and average quantities (over the 30 second scan time) of various environemtal and other quantities (e.g. structural temperatures, wind speed and direction), and so on.

The current database has served us well, and I believe the first step should be to replicate this structure, or something close to it, in MySQL.

Is there any benefit to recreating the contents of the current ACDB using MySQL? I don't think so - the existing is already available to the applications that want to access it.

To date, the database has been populated by running the Archivist to create per-scan FITS files of the ancillary data, and converting these to CSV files using Dana's glish scripts. Since we are currently developing python scripts to process the data, and we have a "fresh start" we have proposed to switch to using sampler2log FITS files instead of the Archivist FITS files. This will change the format of the raw engineering FITS files to some extent, but should have no impact further downstream (the data content should be a superset of what the Archivist logged).

There are two areas where I would propose the existing database could be improved:

• provision of more metadata, and "housekeeping" support for all antenna commissioning experiments.
• Treatment of 10 Hz sampled data

Metadata and Housekeeping support: it would be useful if the ACDB were extended to maintain considerably more metadata about the commissioning runs than is currently the case. To date, metadata has been stored in a per-commissioning run wiki page in a free-format, although reasonably consistent form. The quality of this metadata has depended upon the discipline of the observers (sometimes poor in my case).

Potential metadata which might be stored would be:

• start and stop time of a commissioning run
• names of those performing the commissioning run
• more explicit descriptions of the type of the experiment
• the location of the Scanlog, the engineering FITS files, ranges of scan numbers
• an ascii text field for qualitative information (e.g. cloud cover, presense of fog, etc).
• what else?

Treatment of 10 Hz sampled data:. I'm not sure what is currently in the ACDB for 10 Hz sampled data (the main experiments of this type to date have been half-power tracks). We anticipate considerably more experiments of this type: servo performance tests, inclinometer measurements versus wind and other quantities, Penn Array pointing data and so on. It would be useful to develop a database which can at least keep track of the metadata for these experiments, if not the actual data themselves.

It would also be useful to be able to keep track of where the data from /home/gbtdata and /home/gbtlogs has gone to once it has been archived by the Computing Division!

We also need a unified way of storing the results of astronomical data analysis - e.g. the measured tracking offsets from half-power tracking experiments to compare to other 10 HZ sampled data. Such experiments to date (e.g. PTCS/PN/13) have to my knowledge been treated in an ad-hoc way.

We may wish to do certain experiments where data are sampled at > 10 Hz (e.g. 50 Hz). I propose we continue to treat such experiments in an ad-hoc manner for now.

Specific Proposal

• implement it in MySQL
• fill it from sampler2log files
• base it initially on Kim's Access database schema
• make it better! (this is as far as I have got).

One Year Goals, Tasks and Milestones

This task list seems very ambitious. Is it unrealistic?

• By December 2006:
  • Complete devlopment of python scripts to process pointing/focus data, document comparison to Dana's scripts in a PTCS project note.
  • documented list of the relevant quantities to be stored via sampler2log; what they are, where they come from.
  • Robust, reliable, simple mechanism for observer/operator to enable/disable sampler2logs, which doesn't require m&c password, or calling Joe up in the middle of the night.
  • Complete development of python scripts to process sampler2log files and create something similar to the current ACDB, results should be identical (since it is only format conversion, not data processing) to current glish scripts; if not, explain discepancies and document in PTCS Project Note.

• • First trial version of new ACDB in MySQL, based on Kim's Access database schema.
  • Add new experiment types (e.g. inclinometerw/wind) to ACDB.

• • Update user documentation and transfer to user's guide from PTCS web. Note we can enable dynamic focus and pointing corrections separately!
  • Support John's servo work while not driving it.

• By March 2007:
  • complete prototype version of new fitting software including inclinometry measurements.
  • Develop plan for PTCS use of track shutdown period.
  • Develop more complete pointing error budget (e.g. actual tracking error is instantaneous tracking error plus offset pointing error) and test against this.

• By May 2007:
  • Complete campaign in spring to characterize "before", and demonstrate ability to generate new pointing/focus models, using prototype version.
  • Agree if/how PTCS team will support track work over summer (e.g. use of inclinometry to check track alignment).

• By August 2007:
  • do lots of fixed azimuth, varying elevation wind/inclinometer/thermal data acquisition over the summer (do we want a transit telescope pointing/focus capability?
• restore pointing performance to current levels in August 2007
• By September 2007:
  • restore pointing performance to pre-shutdown or before levels
  • perform comprehensive commissioning runs (all type of pointing experiments) to fully characterize current performance.

-- RichardPrestage - 13 Oct 2006