Practical Aspects of Focal Plane Array Testing Lessons from an FPA Test-bed at CSIRO, Marsfield Douglas B. Hayman1-3, Trevor S. Bird2,3, Karu P. Esselle3 and Peter J. Hall4 1 2 3 CSIRO Astronomy and Space Science, CSIRO ICT Centre, Electronic Engineering, Macquarie University, Dept Electrical and Computer Engineering, Curtin University of Technology Contact email: douglas.hayman (at) csiro.au 4 A test-bed to study Focal Plane Array (FPA) systems has been built at the CSIRO Radiophysics laboratory at Marsfield in Sydney. Called the New Technology Demonstrator (NTD) Interferometer, it was developed by a team from CSIRO, many of whom are now involved in the design of the Australian Square Kilometre Array Pathfinder (ASKAP) [1]. The test-bed consists of two 14 m antennas, one fitted with a single horn and the other employing an 8 8 array of Vivaldi elements, which was designed by the Dutch astronomy organization ASTRON. The primary weighting used for beamforming was the maximum sensitivity or gain (G) on system temperature (T). This presentation will concentrate on some of the practical aspects of evaluating the testbed's operational performance, including: Dealing with a high level of radio frequency interference (RFI): Filters were fitted to the system to reduce intermodulation distortion products and the operational frequencies were restricted. Choice of reference source: A number of choices of reference (calibration) sources were explored including geostationary and LEO satellites as well as astronomical sources. G, T and G/T measurement: Methods for absolute and relative G/T determination were explored and uncertainty estimates were calculated. Radiation patterns: Methods to determine aperture illumination from short radiation pattern cuts were explored. Lessons from both the design and evaluation of the NTD Interferometer have contributed to the development of a second test-bed at Parkes and the design of ASKAP. [1] D. B. Hayman, R. Beresford, J. Bunton, C. Cantrall, T. Cornwell, A. Grancea, C. Granet, J. Joseph, M. Kesteven, J. O Sullivan, J. Pathikulangara, T. Sweetnam, and M. Voronkov, The NTD interferometer: A phased array feed test bed, Workshop on Applications of Radio Science, 2008. http://www.ncrs.org.au/wars/wars2008/ Hayman%20et%20al%20WARS%202008.pdf International Workshop on Phased Array Antenna Systems for Radio Astronomy, Brigham Young University, Provo, Utah, USA May 3 5, 2010
Practical Aspects of Focal Plane Array Testing Lessons from an FPA Test-bed at CSIRO, Marsfield International Workshop on Phased Array Antenna Systems for Radio Astronomy, Brigham Young University, Provo, Utah, USA May 3 5, 2010 Douglas B. Hayman 1-3, Trevor S. Bird 2,3, Karu P. Esselle 3 and Peter J. Hall 4 1 CSIRO Astronomy and Space Science, 2 CSIRO ICT Centre, 3 Electronic Engineering, Macquarie University, 4 Dept Electrical and Computer Engineering, Curtin University of Technology Contact email: douglas.hayman (at) csiro.au
Overview Historical context Description of the testbed at the Radiophysics Lab, Sydney Some results Lessons/issues confronted 2/19
Historical Context Focal Plane Array (FPA) Work at CSIRO ~2002: Collaboration on FARADAY array with ASTRON: Near field antenna measurements Testing with a Luneburg lens 2003/4: FPA projects Two PhD students start: Receiver on a chip: Suzy Jackson Beamforming FPAs: Doug Hayman Plans for a testbed at Marsfield underway New Technology Demonstrator (NTD) Interferometer Plans for (what was to become) ASKAP Focus on digital beamforming Western Australia site planning 2006: FPA for NTD (THEA tile) arrives from ASTRON 2006-2008: NTD operational 2008 Parkes 12m testbed first light 3/19
NTD 1 Interferometer Overview Get FPA test bench going fast On site dedicated system Support decisions needed for the ASKAP Learn about FPAs 2 dishes at Marsfield Radiophysics Laboratory Recycled Fleurs dishes One with an FPA: ASTRON s 8x8 THEA tile One with a single feed f/d: ~0.4 Diameter: 14m. Large enough for astronomical signals. 24MHz beamformer-correlator 1 New Technology Demonstrator: Project name for CSIROs early PFA development Google Maps 4/19
NTD Interferometer Block Diagram Focal Plane Array Single horn feed 21 cables to pedestal 1.1-1.8 GHz East Receiver West Receiver IF: 70 MHz (BW 24 MHz) Digitizer Beamformer-correlator Digitizer Beamformer (weighted sum) Correlator Integrator 1024 complex data points across 28 MHz per second 5/19
NTD Photo THEA Tile for FPA Downconverter Horn Feed 6/19 Beamformer/Correlator
Front End Filter Significant second and third order distortion: mobile phone base station at 947MHz and a TV station at 527MHz. A filter (1100 to 1700MHz) was added between the first and second stages of amplification resulting in: 5-10dB improvement in distortion products Freq 1200MHz 0.51λ element spacing. 7/19
Vertex calibrator Focal Plane Array Radiator 21 cables to pedestal 1.1-1.8 GHz Receiver Noise Source Receiver Digitizer Beamformer-correlator Digitizer Beamformer (weighted sum) Correlator Integrator 1024 complex data points across 28 MHz per second 8/19
Vertex calibrator Amplified noise diode for signal Log periodic antenna Correlated a coupled signal from source with each element s signal to get a magnitude and phase measure Proved a useful diagnostic tool Beamform on the calibration signal Independent of antenna pointing System self test Current focus ASKAP Continuous ~1% T sys cal. signal Radiator choice? Signal to radiator: fibre or coax? Aaronia AG HyperLOG 9/19
Measurement Method Data for Weights Point source for element responses (~20min) Virgo A (M87) in interferometer mode Blank sky for noise covariance matrix (~ 1hr) Correlation between each element Avoid any strong sources in the sky Calculate maximum G/T weights Apply weights in the beamformer Interferometer response... Single dish response... e( θ, φ) C w 1 = C e 10/19
Measurement Method Interferometer Mode Relative G/T against single element (more precise than single antenna measurements): Signal level: Virgo A (M87) in interferometer mode Noise level: Blank sky: autocorrelation Pattern cuts against SGRA* and Virgo A Made with respect to beam centre (coordinate transforms wrt antenna pointing) Try to get as much information from short pattern cuts (~6 beamwidths) Interpolate to get 2D far field Project to aperture (Fourier transform with Hamming filter) Allows qualitative comparisons between different cases single element and weighted cases 11/19
Results for Offset Array + 1 Defect Single Array + 1 defect 12/19 Weighted over single element G/T improvement 2.0dB
Results for Central Array + 1 Defect Single Array + 1 defect Element Weight Amplitude (db) 2 0 1.5-10 1 16 11 17 0.5-20 20 8 3 9 21 0 12 4 1 5 13-0.5-30 18 6 2 7 19-1 14 10 15-40 -1.5-2 -50-2 -1 0 1 2 Element Position (λ) Element Weight Amplitude (db) 2 1.5 1 11 16 17 0.5 20 3 8 9 21 0 1 4 5 12 13-0.5 6 2 7 18 19-1 10 14 15-1.5-2 -2-1 0 1 2 Element Position (λ) 0-10 -20-30 -40-50 13/19 Weighted over single element G/T improvement 2.2dB
Measurement Method G/T Performance Single dish mode Y factor On/Off M87 Y factor On/Off absorber T rx estimate 137 (-17 +29) K Weighting Number of Elements G/T (db(1/k)) T scene T sys η (%) Tsys/η (K) Single Element 1 Max G/T 5 ~Max Gain 21 Max G/T 21 17.2 112 249 41 614 18.8 66 203 49 419 19.6 64 201 57 350 19.5 55 192 54 357 For details see [1] 14/19
Move to Parkes Much lower RFI 12m prime focus dish Test bed for prototype FPAs for ASKAP More powerful beamformer & correlator (data collection much faster) Used with 64m allowing much better SNR 15/19
System considerations Real time vs. store and process Ideally use both providing verification: e.g. Parkes system Set up for observation: Beamforming Prototype for final: Real Time Setup easy to fine tune:fringes clearly visible Beamformer committed at observation time Requires more stability Closer to final system operation Store and Process Setup difficult to fine tune Free to apply different beamformers during processing Easier to re-run processing to match different scenarios Interferometer advantages over single dish Allows much lower strength sources to be used Much higher SNR and phase in pattern cuts Less sensitive to gain variations 16/19
Conclusions Demonstrated digital beamforming in prototype radio telescope (with an interferometer) System performance was modest but explainable. A team was trained in FPA system design and testing FPA characterization techniques were evaluated and refined Project management issues and lessons for rapid prototyping Early definition of port numbering Nomenclature: Ports, channels, beams, primary pattern... Keep it simple Subsystem testing vs. entire system debugging Propose early integration tests Identify interface issues Clarify specifications Early results Follow by comprehensive review and rework if required 17/19
Acknowledgements The NTD Interferometer THEA tile & support from ASTRON CSIRO team Thanks for helpful discussions with: Stuart Hay, John O Sullivan and Mike Kesteven This work is part of a PhD with Macquarie University and funding support from the CSIRO ATNF. More information [1] D. Hayman, R. Beresford, J. Bunton, C. Cantrall, T. Cornwell, A. Grancea, C. Granet, J. Joseph, M. Kesteven, J. O Sullivan, J. Pathikulangara, T. Sweetnam, and M. Voronkov, The NTD interferometer: A phased array feed test bed, Workshop on Applications of Radio Science (WARS08), 2008 [2] Hayman, D. B.; Bird, T. S.; Esselle, K. P.; Hall, P. J.;, "Experimental Demonstration of Focal Plane Array Beamforming in a Prototype Radiotelescope," IEEE Transactions on Antennas and Propagation, June 2010 (already on Xplore) 18/19
Questions? 19/19