13
Early indications of performance Max Voronkov Software Scientist – ASKAP 13 th May 2009

Early indications of performance

Embed Size (px)

DESCRIPTION

Early indications of performance. Max Voronkov Software Scientist – ASKAP 13 th May 2009. Performance tests - Main questions. Typical delay scatter in vis plot and its stability Wide bandwidth allows to estimate delays better May see weak effects we have never seen before - PowerPoint PPT Presentation

Citation preview

Page 1: Early indications of performance

Early indications of performance

Max Voronkov

Software Scientist – ASKAP

13th May 2009

Page 2: Early indications of performance

Performance tests - Main questions

• Typical delay scatter in vis plot and its stability• Wide bandwidth allows to estimate delays better

• May see weak effects we have never seen before

• Flux scale calibration• Bandpass shape, spectral indices and software can affect it

• Using 1934-638 at 7mm?

• Astrometric accuracy• Systematic effects (e.g. baseline solution)?

• Accuracy of the pointing solution• Greater sensitivity may improve the solution

• Bandpass shape and spectral indices can make it worse

• Could we use weaker sources?

Page 3: Early indications of performance

Delays in vis plot

Page 4: Early indications of performance

Delays in vis plot (cont’d)

Page 5: Early indications of performance

Flux scale calibration

• We didn’t have a chance to study this in detail• More information will probably come from the calibrator monitoring

• Be prepared to see systematic effects

• Wide bandwidth effects are more serious for CX-band (3 and 6 cm).

• Probably the uncertainty is not worse than a few per cent for CX-band, 10% for 12mm, a few tens per cents for 7mm and 3mm bands.

• Paddle measurement at 3mm• Frequency-dependent Tsys measurement

• Does not converge amplitudes in vis.

• Can 1934-638 be used instead of the planet at 7mm?• The source is easily detectable at the level of about 0.3 Jy

• There is a factor of 1.8 difference in fluxes if calibrated on Uranus and 1934-638!

• The difference is present for narrow bandwidth too. Flux model?

• Previous such experiments with the old system at 36 GHz gave a factor of about 1.3.

Page 6: Early indications of performance

Interference spikes

Things improved a lot since this picture was taken!

Page 7: Early indications of performance

Observations of G19.61-0.23 at 7mm

Page 8: Early indications of performance

Recombination lines in G19.61-0.23 at 7mm

H53

42951.47 MHz

Page 9: Early indications of performance

Pointing accuracy

• Important for mm bands where wide band effects are small (for general users)• Pointing solution is done at 3/6cm after reconfig (staff)• Wide-band effects (spectral index and primary beam dependence on frequency) are not taken into account• Iterative solution for pointing should converge

• Did a number of pointing scans with a 10’’ offset in positions• Compared solutions with 10’’

Page 10: Early indications of performance

Pointing accuracy (contd.)

• Antennas 3 (technical problems) and 6 were excluded from the analysis• Found 3.9 arcsec accuracy• This is similar to the accuracy obtained with the old system • Strangely enough, the pointing procedure always underestimates the offset• It would be interesting to repeat the same experiment in C/X and/or with narrower tvch

It is advisable to use the old strategy to choose pointing calibrators.

Page 11: Early indications of performance

W33 - combining old and new data

• CABB data completely outweigh the data from the old system

• Blue contours - old system, 6km array• Red contours - the same 6km array data combined with the new CABB data (using H168)• Grayscale - VLA ammonia data (Keto & Ho, 1989)

Page 12: Early indications of performance

Summary

• Usually delays are spread by less than 10 ps or so• Baseline errors can give a residual delay of the order of tens of

picoseconds (especially 6km baselines at 7mm). This should not be a problem for a typical science experiment

• The weather and other effects we don’t yet understand usually give a smaller delay

• Flux scale calibration and ATCA astrometry need further study• Calibration on 1934-638 and Uranus at 7mm gives different result

• There are interference spikes and lines in the spectra (flagging)• Always observe bandpass calibrator for every frequency setup• Pointing accuracy is no better than before

• Use the same strategy as before (1 Jy or brighter source close to the target)

• Merging old correlator data with CABB data is non-trivial

Page 13: Early indications of performance

Contact UsPhone: 1300 363 400 or +61 3 9545 2176

Email: [email protected] Web: www.csiro.au

Thank you

Australia Telescope National FacilityMax VoronkovSoftware Scientist (ASKAP)

Phone: 02 9372 4427Email: [email protected]: http://www.narrabri.atnf.csiro.au/~vor010