A pilot 20-GHz survey at the Compact Array

On the inauspicious date of Friday 13 September 2002, a pilot 20-GHz survey (C1049) commenced at the Compact Array. Over a continuous observing stretch of 96 hours, about 1000 square degrees of sky were surveyed in a "blind" survey between declinations -60 and -70 degrees. The instrumentation and observing techniques were unusual: the correlator was not the regular ATCA correlator, but was an experimental single-baseline analogue device with a bandwidth of 4 GHz; the survey was conducted by scanning the telescopes up and down the meridian at a rate of 10 degrees per minute; and the antennas were controlled and monitored by new generation Antenna Control Computers (ACCs). The hardware, the survey method, the early results, and future plans are described below.

Figure 1: Ron Ekers (Principle Investigator, C1049) and Warwick Wilson discussing the 4-GHz experimental analogue correlator that lies on a bench in the screened room at Narrabri.

Both the 12-mm and 3-mm receivers currently installed and operating on the ATCA provide intermediate frequency signals with a maximum instantaneous bandwidth of around 8 GHz in the frequency band from 4 - 12 GHz. The current ATCA correlator is only able to make use of 128 MHz of this bandwidth in each of two frequency bands, or 256 MHz altogether. However, when surveying large parts of the sky at high frequencies and high sensitivity, maximum bandwidth is required. To make use of more of the available bandwidth, signals have to be transported over hundreds of meters without significant degradation. One possible means of achieving this would be to digitise the signals and send the data over a high bit rate communications channel. An alternative approach, which operates in the analogue domain, has become practical in recent years due to the development of broadband fibre-optic modulators. These devices modulate a laser signal in a single mode optical fibre with a broadband radio frequency signal. The modulated light wave can then be transported over long distances with very little loss. At the destination, an optical detector is used to recover the original broadband RF signal. Thanks to the superb efforts of Mark Leach, this was the method used for this experiment.

In a return to techniques employed in the very early radio interferometers, we were able to remain in the analogue domain by using an analogue correlator to combine the signals from the two antennas. This correlator, built by the ATNF Electronics Group, is composed of sixteen analogue multipliers arranged to sample the correlation function of the input signals over a range of delay of about 2 nanoseconds. In the current system, the multipliers are commercial silicon Gilbert Cell devices. These multipliers limit the overall bandwidth of the system to around 3.8 GHz.

We attached the experimental correlator to antennas CA02 and CA03 both of which were equipped with 12-mm receivers. This gave us a single-baseline sensitivity of 35 mJy, for an integration time of 80 milliseconds, at our central observing frequency of 18 GHz. The downside of having an analogue correlator is that delay compensation (the correction for the variation of geometric delay with time) is not easy. Therefore we dispensed with delay tracking, and made observations solely on the meridian where the delay for an east-west interferometer is zero.

Given the small ATCA primary beam at 20 GHz (about 2.3 arcmin), the most convenient way for us to cover a large solid angle of sky was to scan the antennas up and down the meridian as rapidly as possible. Although continuous scanning is more efficient than the normal point-and-shoot technique of mosaicing, it required the use of the newly commissioned ACCs in order to drive the antennas and constantly monitor their positions. Instead of the usual 10-second cycle, positions were recorded every 100 milliseconds, commensurate with the antenna drive rate of 10 degrees per minute, and the correlator dump rate of 125 Hz. The reliability of the new ACCs, even in this commissioning phase, was encouragingly good.

After commissioning the software associated with the vital antenna-control, scheduling and communication software (mainly Michael Kesteven, David Brodrick and Mark Wieringa), and doing various calibrations, 65 hours of survey data were taken, covering 1000 square degrees of sky. Even for a pilot survey, this is quite a respectable sky coverage. Several-hundred candidate sources were located and will have their identifications and positions confirmed in separate follow-up observations. For comparison, the largest published high-frequency survey is that of Taylor et al (MNRAS, 327, L1, 2001) who used the Ryle Telescope to find 66 sources in 63 square degrees at 15 GHz (but to higher sensitivity than the present survey). During their spare time following commissioning, Mike Kesteven and Warwick Wilson were able to develop split-array software which allowed the remaining four ATCA antennas to be used for instantaneous follow-up observations of candidates at lower frequencies, as well as being used for override observations of two X-ray transients.

Figure 2: The response of the wideband correlator as Mars crosses the meridian. Antennas 2 and 3 were both tracking the planet, so the width of the peak represents the delay window outside of which the interferometer fringe cannot be seen, except by decreasing the bandwidth. Survey observations were conducted solely on the meridian.

Figure 3: The large (!) observing team present at Narrabri for the September pilot survey. From left to right: Lister Staveley-Smith, Ron Ekers, Jenn Donley, Kate Smith, Mike Kesteven, Elaine Sadler, Carole Jackson, Warwick Wilson. Present in later observations were: Roberto Ricci, Ravi Subrahmanyan and Mark Walker. In the background, antennas CA02 and CA03 are scanning up and down the meridian between declinations -60 and -70 degrees.

Plans for the near future include replacing the correlator with a device capable of processing the entire 8-GHz bandwidth and extending the system to three antennas, with an expected doubling in sensitivity. It is with this system that we hope to proceed with an all-sky survey in 2003. In the longer term (4 - 5 years), and as part of MNRF 2001, a wideband digital correlator will provide equivalent bandwidth for all antennas at much higher spectral resolution. And, importantly, it will come with a digital delay system so that observing away from the meridian is possible!

Lister Staveley-Smith and Warwick Wilson, on behalf of the 20-GHz survey team.
Lister.Staveley-Smith@csiro.au
Warwick.Wilson@csiro.au