A 50 MHz System for GMRT

1. A 50 MHz System for GMRT N. Udaya Shankar K. S. Dwarakanath Raman Research Institute, Bangalore, INDIA MWA Meeting, Canberra, 18-23 Jan 2009

2. TEAM S. Amiri, R. Somasekar, B.S. Girish, Arvind Nayak, Wenny Laus, S. Kasturi, Sujatha.

3. Outline • Objectives of a 50 MHz system for GMRT. • Feed design. • Summary of observations with the new feed. • Direct voltage recording system. • Experiments with the direct voltage recording system in the interferometric mode. • Conclusions.

4. Objectives of a 50 MHz system • Design and build feeds for operation at frequencies below about 100 MHz. • Carry out a low frequency survey of the sky visible to GMRT with arcmin resolution and unprecedented surface brightness sensitivity. • Make the survey results available to the astronomy community as an online resource. • Make the observing system available to GMRT user community.

5. Survey Parameters • Assuming: • Bandwidth: 10 MHz, • System temperature: 4000 °K, • Synthesized beam: 1’, and • RMS sensitivity (4 hr, dual polarization): 2 mJy/beam (Thermal Noise) 10 mJy/beam (Dynamic Range Limited) • A mosaic of about 300 pointings to cover the Northern sky requires 1200 hrs of observation • Primary beam area of 0.03 steradians.

6. Survey comparison • Larger collecting area and higher antenna efficiency means the low frequency GMRT system will be more sensitive than the 74 MHz VLA system.

7. 5s confusion Sensitivity Vs Angular size

8. Development of high dynamic range, interference tolerant receivers. Detection and mitigation of radio frequency interference. Calibration issues arising out of ionospheric effects. Problems of imaging wide-fields with non-coplanar baselines. Deconvolution of images with large scale features. Main challenges

9. A Low Frequency Feed for GMRT Wish list: Covering the frequency range 30 MHz to 90 MHz. Sky noise dominates the system temperature (Tsky> 5 Tsys). Reasonable aperture efficiency (0.5 < ap < 0.7). Symmetrical E & H plane patterns. Physical dimension is suitable for mounting it along with one of the existing feeds on the GMRT antenna turret, with minimum interference to its operation.

10. Feed Design Simulations were done to determine the configuration which cause minimum interference to the operation of the existing feed. After field trials, it was decided to co-locate the new feed with the existing 327 MHz feed. The desired frequency of operation (30 to 90 MHz) was chosen to: Facilitate imaging in the band protected for Radio Astronomy around 38 MHz, Have an overlap with the 74 MHz system on VLA, and Minimize the RFI due to the FM Radio band starting around 90 MHz.

11. Feed Design Boxing ring configuration chosen since it has more symmetric radiation characteristics (E and H plane patterns) than a single dipole like feed. Following parameters were tuned: Distance between two parallel dipoles in the boxing ring. Shape of the dipoles to achieve the required bandwidth. Height of dipoles above the reflector. Impedance mismatch at frequencies away from the resonance.

12. Present Status The first phase of the project : A system for 4 GMRT antennas has been designed, built and installed and its performance has been examined. The design would be developed further, if necessary, based on our experience. Four folded V-dipoles each of length ~2.4m. Its feed point is positioned 1m above a 3m X 3m square reflector with a 50mm mesh. The feed co-located with existing 327 MHz feed

13. Summary of observations • GMRT antennas C04, C11, E02 & W02 equipped with new feeds. • 3C sources observed: Cyg-A, Cas-A and a few other sources. • Maximum EW baseline: 6 km. • C04-C11 NS baseline: ~0.5 km. • Aperture efficiency: • ~65% (55 MHz) • ~30% (40 MHz) W02 E02 C04 C11

14. Observed several 3C sources. Over an hour of observing of Cyg-A, the cross correlation coefficients showed: A steady behaviour as has been seen at other observing bands of GMRT. Measured rms of ~6% on the visibilities at 55 MHz implies an rms of ~12° in closure phases and ~12% in closure amplitudes. Closure phases showed an rms of 1.4° and the closure amplitudes showed an rms of 7%. Observed rms in the closure quantities are much less implying that a significant part of the rms in the visibilities are due to systematics. Summary of observations

15. Baseline: C04-E02 Baseline: C11-E02 Phase (Degrees) Baseline: C04-C11 Time (Hours) Closure phases • Looking at one good channel across 1 hr time. • RMS: ~1.5°.

16. Direct voltage recording system • Digitize a signal of ~5 MHzbandwidth centred at 55 MHz. • Data rate: ~80 GB/hr. • Systems synchronized using signals from GPS-disciplined Rubidium oscillators. • Two direct voltage recording systems were installed at C04 and C11 antenna bases.

17. Rf Rf 8-bit A/D Converter 8-bit A/D Converter ~80 GB/Hr Stored On Hard Disk Stored On Hard Disk 8192-point Fourier Transform 8192-point Fourier Transform Laboratory Test Setup Offline Data Processing Delay Correction Delay Correction Visibilities Post Integration Setup At GMRT Antenna Base Direct voltage recording system

18. ~2000 central channels used for analysis. Typical cross power dynamic spectra • ~4000 channels. Spectral resolution of 1.36 KHz. • ~4 min of data. Each time frame with ~1 second integration. • Notice the RFI stripes. • Only ~2000 central channels marked in plot used for analysis.

19. -20 -25 SELF ANT-1 -30 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 CHANNEL INDEX -20 -25 SELF ANT-2 -30 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 CHANNEL INDEX -28 -30 Un-NORM CROSS -32 -34 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 CHANNEL INDEX Typical Self & Cross spectra • Integrated for ~4 min, only central ~2000 channels shown. • Notice the comb in self power of Antenna-2 (C11) and is relatively affected more by RFI.

20. 600 500 400 Number of interference points 300 200 100 0 1 2 10 10 10 Duration of interference (seconds) RFI Scenario • RFI Scenario in ~1s integrated visibility for ~4 min data. • ~0.5% RFI points above 4-σ.

21. Dynamic spectra after RFI excision • Central ~2000 channels. • Notice, the vertical RFI stripes (across time) have disappeared.

22. Structures in Time-Frequency • ~4 min integrated spectra. • RFI channels: ~20% • Period: ~ 45 s • Variations: ± 5%

23. Our experience shows that a multi-resolution filter provides a possible optimal solution for interference mitigation. • Using this approach we are in the process of analysing simultaneous observations carried out using: • GMRT hardware correlator, • New GMRT software correlator which is under testing, and • RRI direct voltage recording system. • We hope these investigations will throw more light on the effect of systematics on thermal noise.

24. Conclusions • We have successfully designed, installed and tested a 50 MHz system on 4 GMRT antennas. • Aperture efficiency is ~65% in the 50 MHz band. • Test observations indicate satisfactory performance of the feed system. • Investigations are under way to understand the mitigation of systematics and RFI to achieve performance limited by thermal noise.

25. Acknowledgements We thank our colleagues at GMRT site and NCRA for their excellent support.

28. Structures in Time-Frequency • Integrated for ~4 min, only central ~2000 channels shown. • First row: • Second row: Variations across time.

29. Aperture efficiency Cyg-A (Sep 07 observations) • Baseline- C04:C11 • Centre frequency: 55 MHz • Aperture efficiency: ~65 %

30. Aperture efficiency Cyg-A (Sep 07 observations) • Baseline- C04:C11 • Centre frequency: 38 MHz • Aperture efficiency: ~30 %

41. Surface Brightness Sensitivity of other surveys Surface Brightness sensitivity of MRT 7 X 10 (-22) Watts/m2/Hz/ Ster (Theoretical) 2 X 10(-21) Watts/m2/Hz/ Ster (Achieved) 4.2 X 10 (-22) Watts/m2/Hz/ Ster (Eq at 1.4 GHz) Typical SB of SNR = 1.5 X 10 (-21) Watts/m2/Hz/ Ster SNR cat complete = 10(-20) Watts/m2/Hz/ Ster at 1 GHz VLA 4mass 100 mJy/beam Beam Size 80’’