Project Skylight


A few years back, I became fascinated by the concept of phased array antenna used to view the sky at low frequencies. There are three examples of the technology operating at the moment and although they all use different techniques in the end they achieve the same goal.

LOFAR is based in Holland and is the largest of the antenna arrays which also combines antenna arrays in other countries resulting in a huge coverage area.

One LOFAR station with low frequency array in the center and two high frequency arrays either side.

LWA New Mexico is sited near the Very Large Array featured in the movie 'Contact'. The LWA trial site has 256 dual polarity dipole antenna with an amplifier module at the antenna and co-axial cable running back to a Shipping Container converted to a Science Station that houses the electronics and computers used to receive the weak radio signals from space.

The Long Wavelength Array station number 1 in New Mexico.

The Murchison Widefield Array is located in Western Australia and is part of the Square Kilometer Array project. The MWA also uses dual polarity dipoles in a 4x4 'tile' resulting in 32 separate signals that are combined in a box that can perform a phase delay on any of the 32 signals. The phase delay allows the 4x4 array to point electronically to anywhere in the sky. The combined signal is then routed to the main Science building for further data recovery.

Several 4x4 phased arrays with a 32 input phasing box.

The LMRO Low Frequency Array
My personal fascination with viewing the sky at radio frequencies started long ago. Visits to Parkes Radio Telescope in New South Wales and the Australian Telescope Compact Array in Narrabri led me to an interest in Radio Astronomy. Obtaining my Amateur Radio license in 1981 while living in Gunnedah, New South Wales, I had an amazing sky to look at.

Having been working with the Radio Astronomy section of the Astronomical Society of Victoria I have access to a remote location near Heathcote in central Victoria. The RA section has a Science Station on the site with many different experiments running. The website at provides access to the data gathered at the site.

I decided I wanted to build a phased antenna array of my own. So I have started Project Skylight. The parameters are modest. I wanted it to be modular and easy to build with readily available material. It had to be low cost but still capable of achieving good science. This meant each antenna had to be a module capable of being added to as the design developed.

I wanted each antenna to be capable of operating on its own with its own radio receiver and digitizer. In order to make the system frequency agile I couldn't use an analogue receiver in each antenna. With recent developments in Software Defined Radio a cheap method of receiving was devised. By using a 'USB Dongle' type receiver designed for TV and FM radio reception I had a simple and flexible method of receiving signals at each antenna.

I also needed a method of controlling the USB Dongle receiver that was housed inside the antenna. The Raspberry Pi computer with its small size and ability to run a Linux operating system made a good choice for the signal digitizer. By sending control messages to the Raspberry Pi the Radio Dongle could be set to a specific frequency and bandwidth.

Each Raspberry Pi would stream the data through an Ethernet cable to a Data Switch at a central point in the array. The multiple data streams would then arrive at the main storage/processing computer where digital streams can be processed in real time or stored on Hard Drives for later analysis.

One problem I'm trying to overcome is the correlation of data streams. With each antenna sampling at 2 Megasamples per second a method is needed to line up the samples at each antenna. I'm now at the point of developing a method of 'time-stamping' blocks of data samples so when combining sampled data a delay can be computed during processing.

Part of the correlation process involves each Software Defined Radio dongle being provided with a common 28.8 MHz source to keep the phase the same in every receiver. This reference signal can be sent through a spare twisted pair in the ethernet cable that connects to all the antenna. What is still needed then is a time stamp reference preferably embedded in the data packets sent by each antenna.

I have come up with a possible solution. By inverting the phase of the 28.8 MHz reference once per second all receivers will automatically stamp the sample stream when the phase change occurs. I'm hoping this will result in a single sample in the data stream 'standing out' from the rest so it can simply be correlated with all the other streams. This 'marker sample' can be easily found and define the start point of every data packet.

Note: This document is a work in progress. More to come shortly.

06.07.2014. 22:16