1. Widefield Astronomy and Technologies for the SKA 4 - 6 November 2009 at Limelette, Belgium The SKA AA-lo array; E.M. simulation and design Eloy de Lera Acedo (UCAM) Nima Razavi-Ghods (UCAM) David Gonzalez-Ovejero (UCL) Luis Enrique Garcia (UC3M) Christophe Craeye (UCL) Peter J. Duffett-Smith (UCAM) Paul Alexander (UCAM)

2. Overview Towards the SKA: The SKA-AAlo The BLU antenna The BLU antenna in a regular array Simulating large random finite arrays… Numerical results Current state specifications Future work Conclusions

3. Towards the SKA: The SKA-AAlo It is part of Prep-SKA. Frequency range from 70 MHz to 450 MHz and +/- 45 o scan range:  Irregular vs Regular / Dense vs Sparse (Memo 87).  Sky noise limited?  At 100 MHz: 4000 m 2 /K.  Up to at least 10000 elements per station.  For more specifications: Memo 111 (R. Bolton et al). 10020030040045070MHz10

4. Overview Towards the SKA: The SKA-AAlo The BLU antenna The BLU antenna in a regular array Simulating large random finite arrays… Numerical results Current state specifications Future work Conclusions

5. The BLU antenna Bow-tie Low-frequency Ultra-wideband antenna. Low profile and differential feeding. Wide beam width. Good matching at the high end of the band where the sky noise does not limit the performance. Small and cheap.

6. Overview Towards the SKA: The SKA-AAlo The BLU antenna The BLU antenna in a regular array Simulating large random finite arrays… Numerical results Current state specifications Future work Conclusions

7. The BLU antenna in a regular array Infinite array simulations were carried out to analyze the sensitivity of a unit cell containing a BLU antenna versus the inter-element spacing, the antenna size and the tilt angle of the arms. + Info - GND @ λ/4 @ the highest freq. - No dielectric. L L Z = 200Ω d

8. The BLU antenna in a regular array The sky will dominate a large part of the band. Furthermore, grating lobes will show up in the band. + Info - Sky noise - Receiver noise - System noise Table 1: Idealize LNA parameters. F min RnRn Z opt Z amp 0.210 Ω200 Ω Physical size of the unit cell Table 1: Idealize LNA parameters. F min RnRn Z opt Z amp 0.210 Ω200 Ω System noise Receiver noise

9. The BLU antenna in a regular array In the infinite array the sensitivity of a unit cell in the regular array improves as sparser is the array in the dense regime while it saturates for the sparse regime. The transition band is the key. Sensitivity for different Inter-element spacing distances dL λ/2 @: 70-300 MHz Fixed to λ/2 @ 300MHz

10. The BLU antenna in a regular array Larger antennas bring a multi-lobulation issue at high frequencies. Sensitivity for different antenna sizes.

11. The BLU antenna in a regular array An optimum tilt angle can be found for the antenna arms. The impedance match improves and so does the receiver noise. This is of special interest for the region in between the dense and sparse regions. T rec for different antenna angles α

12. Overview Towards the SKA: The SKA-AAlo The BLU antenna The BLU antenna in a regular array Simulating large random finite arrays… Numerical results Current state specifications Future work Conclusions

13. Simulating large random finite arrays… Based on Method of Moments + MBFs (CBFs) and the interpolation technique presented in [1], where the computation of interactions between MBFs is carried out by interpolating exact data obtained on a simple grid. [1] D. Gonzalez-Ovejero and C. Craeye, “Fast computation of Macro Basis Functions interactions in non-uniform arrays,” in Proc. IEEE AP-S Soc. Int. Symp., San Diego, CA, Jul. 2008.

14. Simulating large random finite arrays… A so-called “radius of influence” is defined for every antenna in the array. The interactions between MBFs are computed only within that region reducing drastically the number of unknowns. The system is solved within each region of influence for each antenna. 30λ

15. Overview Towards the SKA: The SKA-AAlo The BLU antenna The BLU antenna in a regular array Simulating large random finite arrays… Numerical results Current state specifications Future work Conclusions

16. Numerical results - Distance to ground plane = λ 0 /4. - No dielectric. - Array radius = 30λ 0. - Number of elements = 1000. - Minimum average separation = 1.5λ 0. λ0λ0 λ0λ0 Z = 200Ω

17. Numerical results λ = 3λ 0 *Embedded Element Pattern for radius of influence = 10λ

18. Numerical results EEP- SEP = 0.379 dB in broadsideEEP- SEP = 0.2 dB in broadside

19. Numerical results

20. Current state specifications For previous simulation: Computation time:  Preprocess (computing data to interpolate): 37 min  Computing 1000 EEP: 100036 s ~ 10h. Memory:  Only 17.8 MB of contiguous memory required. Speed  10h per simulation: excessive.  Sparse matrices (initial results): Simulation time reduced to 1h. Requires more contiguous memory: 1.77 GB

21. Future work Full EM simulation of a real size SKA station. Optimization:  Use of different NFFT (Non-uniform Fast Fourier Transform) schemes for the array pattern calculation.  Paralelization of the code in a cluster of computers.  Full migration to C code.  Use of sparse matrices solvers. Validation of the code with a scaled prototype of an SKA AA-lo station. Analysis of the mutual coupling effects in irregular arrays (random sparse vs dense regular ?). Station+antenna design. Work towards the telescope calibration… - Scaled and small SKA AA-lo station. - 400 elements. - Scale: 30:1 of the BLU antenna. - 1m 2. - UWB baluns. - Band: 2.1 GHz to 13.5 GHz.

22. Conclusions MoM/MBF based method to simulate large finite irregular arrays. Objective: Full fast EM characterization of a SKA station. Analysis and design (randomness, sparseness, etc.). Can we help? Analysis of an infinite regular array of BLU antennas proving its suitability for the SKA AAlo.

23. End Thank you! Eloy de Lera Acedo eloy@mrao.cam.ac.uk