Telescope

MWA antennas are a dipole design, made of aluminium, optimized for the 70-300 MHz frequency range. Each antenna is a pair of orthogonally crossed, vertical bowties (having dual-linear-polarisation), with a span of 74 cm and a height of 55 cm.

Each antenna is fitted with a low noise amplifier (LNA), placed between each pair of bowtie arms. The LNA amplifies incoming signals while adding less noise than is received from the coldest regions of our Galaxy.

The LNA for each element is housed in a protective, UV-resistant hub, to which the vertices of the bowtie arms are also attached.

MWA antennas are arranged into ’tiles’. Each tile is a small phased array of 16 antennas in a planar, 4×4 square grid, with 1.10-meter spacing corresponding to half a wavelength at 136 MHz.

The antenna elements are are held ~10 cm above the ground screen by dielectric “feet” that clip to the ground screen mesh. The mesh, which has a 5×5-cm square grid pattern, is made of 3.15-mm-diameter galvanized wire. Three mesh panels, each measuring 2×5 meters, are overlapped slightly to form a 5×5-meter ground screen.

The mirror effect of the ground screen causes the pattern to roll off rapidly at elevations below 30°, with consequent enhanced rejection of terrestrial RF interference.

The beamformer is a device (the small white box next to each tile) that allows the telescope to track objects in the sky. 

The beamformer receives signals from all 16 antennas in a tile, and applies independent delays to each signal in a manner appropriate to form a tile beam in a particular direction on the sky. True delay steering is employed, rather than phase steering, in order to point the beam properly over the full operating frequency range. The beam is 15-50° wide (FWHM), depending on frequency.

The delays are generated passively in coplanar waveguide transmission lines laid out on a printed circuit board inside the beamformer. Delay sections of different lengths can be switched in or out of each signal path as required to steer the tile beam in the desired direction. The delayed signals are combined and amplified by the beamformer, then sent over coaxial cable to the receiver.

The animation below shows the beam pattern of an MWA antenna tile, which changes shape as a function of frequency. The beam pattern represents where the antenna tile is sensitive to receiving signals, and it can be electronically steered in almost any direction. In this way, we can point the telescope without any moving parts!

The signal data is filtered and digitised by the receiver before being sent to the correlator over fibre-optic cable.

Each receiver accepts the analog data streams from 8 MWA tiles over coaxial cable, and outputs digital data streams on optical fibre. To do this, it digitizes the input RF signal after appropriate signal conditioning, and passes the data through a coarse polyphase filter to obtain 1.28 MHz wide “coarse” channels. Astronomers can then select 24 of these coarse channels (within the total tile bandwidth of 70-300MHz) for transmission, giving the MWA a total instantaneous bandwidth of 30.72MHz.

The receiver enclosure is weather tight, and shielded against radio emissions.

Some of the tiles in the MWA are located at distances in excess of 500m from any receiver. Usually the receiver would send power to the tile’s beamformer, but for these far-away tiles, solar power is far more effective. And instead of sending data back along coaxial cables (which would introduce losses and noise over those long distances),  these tiles are connected to their receivers by optical fibre. These far-away tiles have a device called a Beamformer Interface, or BFIF, which manages the communications over the optical fibre link, and the solar panel power supply.

The data packets from the receivers need to be organised before being processed further. Currently this is achieved by a switch and a ‘media conversion’ server, shown in the animation below. After the MWA antenna signals are digitised by a receiver (turning the analog signals into 1’s and 0’s), the media conversion server helps to sort all this data by frequency. Different coarse channels are represented here as different colours. A switch then sends each coarse channel to a dedicated correlator server. This process is known as the corner-turn operation.

The correlator is the ‘brain’ of the telescope; a huge bank of servers that perform the mathematical operations to correlate the information from all MWA antennas together at once. The correlator, called ‘MWAX’, is situated on-site at the Murchison in the CSIRO control building.

More technical information can be found on the wiki, here: https://mwatelescope.atlassian.net/wiki/spaces/MP/pages/24970246/MWAX+Correlator+and+VCS