MOST: The Molonglo Observatory Synthesis Telescope

A severe disadvantage of the original Mills Cross was that it could make only transit observations. It was recognized that a steerable telescope was necessary to obtain extended observing times and greater sensitivity. To achieve this at a reasonable cost it was decided to abandon the $NS$ arm of the cross and provide a new phased system for the $EW$ arm only. With this a two dimensional aperture is synthesised using earth rotation synthesis. If linear polarisation is used, the position angle of the feeds with respect to the sky will also rotate. Hence, the existing linear feeds were replaced by a circularly polarised feeds.

The usual aperture synthesis procedure accumulates data as points in the spatial frequency $(u,v)$ plane and then interpolates them onto a rectangular grid^7.6. The map in the $(\theta,\phi)$ domain is produced by a fast Fourier transform. An important requirement of this method is that the primary beam shape must not vary throughout the observation. This makes it unsuitable for the Molonglo telescope where the primary beam is derived from a rectangular aperture. Because of the mutual coupling problems together with the foreshortening of the effective aperture, the gain of the telescope can vary by over a factor of five as the pointing moves from the meridian to $60^0$ from the meridian. This gain variation can be removed from the sampled data, but, the change in beam widths during observations leading to a large variation in the relative gain, between the center of the map and map edges, cannot be corrected for.

The problem of non-circularity and variability of the primary beam may be overcome by the fan beam synthesis or the beam space beam forming. For this the $E$ and the $W$ reflector, each $778$ m long and 11.6 m wide (separated by a gap of $15$ m) are divided into $44$ sections of length $17.7$ m. The $E$ and $W$ reflectors are tilted about an EW axis by a shaft extending the whole length. To control the direction of response in an east-west direction a phase gradient is set up between the feed elements by differential rotation. Each module output is heterodyned to $11$ MHz. A phase controlled transmission line running the length of each antenna distributes the Local Oscillator. One of these lines is phase switched at $400$ Hz.

The detection and synthesis process involves the formation of a set of contiguous fan beams in each antenna. The $44$ signals are added together in a resistance array to produce $64$ real time fan beams. Signals from corresponding beams from each antenna are multiplied to produce $64$ real time interferometer beams. By switching the phase gradient by a small amount every second, these $64$ beams are time multiplexed to produce either $128$, $256$, or $384$ beams in each 24-second sample. Each beam has an EW width of $43''$ and at meridian passage a $NS$ width of $2^0.3$. The hardware beams have a separation of $22''$ and the time multiplexed beams $11''$, which is just under half the Nyquist sampling requirement.

If observations of a particular field extend over hour angles of $\pm6$ h, the fan beam rotates through all position angles and synthesis may be performed. The field is represented by a square array of points corresponding to the projection of the celestial sphere onto a plane normal to the earth's rotation axis. Every $24$ seconds, the accumulated signal at each of the 4x63 fan beam response angles are added to the nearest $(l,m)$ array points. This process continues throughout the 12 hours of synthesis. The computation apart from summation includes gain, pointing, and phase corrections; cleaning to improve the map; to locate the sources and to measure their flux densities and position.

Footnotes

... grid^7.6

See Chapter 11