Array combiner For GMRT
Abstract
Today, Radio Astronomy has developed into a highly complex, interdisciplinary subject. Many radio astronomical problems need technologies from different engineering fields and knowledge from many scientific fields, at times well beyond the state-of-the-art. A large number of unsolved problems in the area of meter-wave radio astronomy led to the conception of the world's biggest meter-wave telescope array facility in India, known as the Giant Meterwave Radio Telescope (GMRT). It consists of 30 dish antennas operating in six different center frequencies of astronomical interest. Some of the important studies planned with GMRT are the interplanetary scintillation study, pulsar searches, studies of known pulsars, radio recombination line observations, and lunar occultation observations, etc. All these observations can be efficiently carried out only if a suitably combined output is available from the GMRT array. The instrumentation design and development details presented in this thesis facilitate the above-described observations with GMRT by providing such single-beam outputs in two modes.
The Design and Development:
The design requirements were carefully studied. The GMRT array simultaneously requires the Incoherent and Phased Array Mode Single Beam outputs for efficiently conducting certain important astronomical observations. After the array outputs are digitized, delay-corrected, and Fourier transformed, the signals can be tapped for producing the single-beam mode outputs. These are complex spectra, represented by digital numbers.
Large Number of Inputs: There are about 120 complex numbers, each being 12 bits wide, contributing 1,440 input signal bits to the system. Handling so many inputs is a major task.
Large Processing Power: There will be about 16 million such 120 complex numbers to be processed every second. This calls for high-speed digital processing hardware with capabilities to do 17,280 million multiplications and 8,488 million additions every second. The computational power required in producing the two single-beam modes cannot be met by any readily available commercial hardware.
The major engineering challenges arise from the magnitude of the number of inputs to the system and from the required speed of operation. It is required to design a suitable instrument to meet these requirements. The design was carefully analyzed and the two main sections were identified. Techniques were found to simplify the design complexities. To validate the design, a reduced version of the instrument, working at 16 MHz, was built and tested at the telescope site with astronomical signals. The tests showed that the basic design was working satisfactorily. Currently, it is being used for astronomical observations. In this design, the major computations were simplified using look-up tables, and erasable programmable logic devices were used for performing the addition. In the final design, two more important considerations were included, viz.,
simplification of the hardware complexity by running the system at twice the speed,
optimization of the quantization levels (in the look-up tables) in accordance with each independent input-signal level.
Then the final instrument design work was taken up. It has been designed with a scalable architecture. The system speed of operation is about 32 MHz. Running the system at the higher speed introduced new constraints. The hardware technology required was completely different from what was used in the prototype design. The realization of the final system hinged on the availability of faster components in the market and the knowledge from literature for building such high-speed systems. For this purpose, the high-speed circuit design aspects were carefully studied. The techniques for performing processing of signals at the required speed were identified and circuit layouts were carefully made to suit the high-speed design. The full system design was completed. The design, taking into account all the additional factors, is implemented in the final version of the hardware. The PCBs required for a main section of the instrument, supporting an eight-dish system (two polarizations), were made, assembled, and tested.
The details about the design and development of this instrumentation outlined above are presented in the thesis.
Results and Conclusions:
The Array Combiner operations were analyzed by simulating the system operation using computer programs. These simulation studies yielded useful information about the choice of parameters to optimize the performance of the Array Combiner. After the instrument was brought to the final form, it was tested with simulated signals in the laboratory as well as at the telescope site. Tests carried out include observation of the noise sources. The test results show that the design goals are adequately met in the hardware. Using the prototype, even pulsar signals were observed by connecting the outputs to the Pulsar search pre-processor back-end. The following are the two typical results (observed pulsar profiles).

