Generation of Phase-Coherent Local Oscillator Signals

The GMRT is generally used in the interferometric mode where the correlation between the electric field vectors received by each feed element is measured. To maintain the relative phase between the electric field vectors incident on different antennas, it is essential that the local oscillators used must be phase-coherent^23.5. This implies that the frequency of the LOs at the various antennas must be identical and the variation of phase of the LO at a given antenna (with respect to the phase of some reference antenna) must be precisely known during an observation so that necessary correction can be made in real time. This is acheived by having all the local oscillator signals be generated from a single reference frequency source using phase-locked loop (PLL) techniques.

In detail, the third and fourth LO signals are generated in the CEB from an ultra-high stable and pure quartz oscillator at 5 MHz. The second LO generation uses a voltage controlled crystal oscillator (VCXO) in a PLL, while the first LO needs a phase-coherent frequency synthesiser. Signals at 106 MHz and 201 MHz are broadcast from the CEB to each antenna, and these are used at the antennas to generate the 105 MHz and 200 MHz second LOs. The 105 MHz second LO signal is in turn used to generate the first LO. In this way the phase coherence of all the LOs at all the antennas is maintained.

Despite being derived from a common signal, there are still phase variations of the LOs at the different antennas for a variety of reasons. The physical length of the optical fibre link to various antennas varies from a few hundred meters to about 20 kms. As the temperature coefficient for expansion of the fibre is not zero, there will be a variation of the phase of LO signal broadcast from CEB and received at an antenna. The receiver system in each antenna is housed in an air-conditioned environment and undergoes independent cyclic variation in temperature. This also causes the LO phase between antennas to vary in a random manner. All of this would make it desirable to have a system for estimation of phase of the LO signals at all antenna locations. This could be achieved by bringing back the second LO signals from each antenna to the CEB and comparing its phase with that of the signal originally generated at CEB. From this information, the phase variation introduced by the transmission process could be estimated. Of course, this needs the pair of optical fibre to an antenna to be reciprocal and non-dispersive, which has been independently confirmed. However, this scheme is not yet fully implemented.

Footnotes

... phase-coherent^23.5

i.e.the phase difference between the LO signals at antenna $i$ and antenna $j$ should be constant with time for all antennas $i$, $j$. If this is not acheivable, then at least the phase difference between the LOs at different antennas should be calibratable in real time.