The design of an antenna array for the Argus system poses a number of engineering challenges, but based on work reported below we are confident that an acceptable array can be fabricated with our proposed resources.

The goal of this sub-task is to construct a 64-element array to operate in the frequency range 400 MHz to 2 GHz with a VSWR of at most 2:1. A highly uniform pattern over a hemisphere is desired in this band, with a null on the horizon. To achieve the lowest possible noise temperature, the antenna should have negligible backlobes, and lossy components must be avoided. In our initial work we have considered elements that are sensitive to a single polarization, but (as we show below) the extension to fully polarimetric operation is not a problem. The geometry of the array is intentionally unconstrained; it should be sufficiently flexible to support experiments into novel imaging concepts. Finally, we require a low fabrication cost (<$ 2k for the array) and the ability to easily manufacture the elements in quantity.

In a previous effort [1] OSU employed an antenna array comprised of bent dipoles arrayed over a ground plane. Based on a design by Lin [2], that array has the remarkable ability to exploit mutual coupling between elements to improve its performance. Measurements performed with an antenna designed for operation at 1.5 GHz showed many attractive characteristics. The performance of the bent dipole array, however, is closely linked to its periodic, rectilinear geometry. To achieve more flexibility in array geometries an investigation of alternative designs was initiated.

Since the mid-1950s there has been intensive investigation of wideband antenna concepts for several applications, including electronic intelligence (ELINT), electronic countermeasures (ECM), and measurement of electromagnetic interference (EMI). For radio astronomy applications, among the most attractive of these concepts is the planar spiral. First described in a 1959 paper by Dyson [3], the two-arm logarithmic (equiangular) spiral is a self-complementary structure, for which very wideband operation is possible. These balanced antennas are capable of VSWRs below 2:1 over a decade of bandwidth. Spirals exhibit circular polarization with axial ratios better than 1.5:1 on axis, the sense of the polarization being determined by the sense of the spiral winding. They are also known to exhibit weak mutual coupling [4]. The Archimedean spiral, with a linear (rather than exponential) rate of expansion, is not truly frequency independent, but it has very similar properties and is easier to manufacture.

A significant limitation of spiral antennas for radio astronomy applications is that they are bidirectional. It has long been known, however, that by placing a ground plane one quarter wavelength behind the antenna one could obtain unidirectional radiation at the cost of a severe reduction in bandwidth [5]. An important contribution was made by Wang and Tripp [6,7] who showed that the presence of the ground plane is compatible with the desired antenna current distribution, although it tends to enhance the contribution of currents reflected from the spiral terminations. Moreover, they found that performance was not a strong function of the antenna-ground plane separation, provided the separation was a quarter wavelength or less. They were able to construct antennas with 10:1 bandwidth by using a resistive loading to attenuate those reflected waves.

Through a combination of simulations and experiments we have found that a spiral over a ground plane without resistive loading has roughly a factor of two reduction in bandwidth (at the lower end) with respect to a loaded spiral of the same design. Since the lower bandwidth is directly proportional to the outer circumference of the spiral, one can achieve a bandwidth comparable to that of a loaded spiral (but with lower noise temperature) by doubling the diameter of the unloaded antenna.

A prototype spiral antenna for the band 400 MHz to 2 GHz was built and is shown in Figure 1. This element was constructed with 3/8" O.D. copper refrigeration tubing and has a diameter of roughly 60 cm. The tubing is held to a dielectric substrate with inexpensive tie-wrap mounts. The total cost of the antenna components is roughly $30. The antenna is placed roughly 5 cm above a ground plane and is fed through a wideband tapered coaxial balun [8]. The impedance bandwidth of the antenna is apparent from the VSWR measurements shown in Figure 2, where we find VSWR<2.2 over most of the desired band with especially good performance near the critical 1.4 GHz hydrogen line. Below 400 MHz the length of the antenna limits performance. Above 2 GHz, the details of the feed construction are significant. The pattern of the antenna was measured by receiving GPS signals over a period of 3.5 hours as described in [1]. The resulting data are shown in Figure 3, where we have plotted the apparent position of the GPS satellites during the measurement period. Detection strength (on a linear scale) is color-coded. The results show a very smooth pattern over much of the hemisphere, as desired.

Several issues must be examined to complete the design work. We have an interest in improving the VSWR, particularly in the region of the hydroxyl line at 1.7 GHz. Performance at the upper end of the band depends on the antenna design near the feed, which is in turn constrained by the balun dimensions. We are presently redesigning the balun. Although our current design has a very wide bandwidth, it is based on a physically large coaxial line and it is difficult to construct. An alternative design based on surface-mounted wideband transformers has been developed and is now being fabricated. The cost of the components, circuit card, and connectors is estimated to be less than $10/antenna.

Although these antenna elements have been developed with a flexible array geometry in mind, it remains to develop some useful array configurations and to verify the performance in such configurations. We are currently exploring dense array concepts, in which smaller replicas of the spiral are packed in voids between larger spirals, leading to reduced grating lobes over wide bandwidths. Arrays for fully polarimetric sensing can be constructed by using spirals wound in both directions. These concepts are also being studied.

References

[1] S. W. Ellingson and R. S. Dixon, The OSU Omnidirectional Radio Telescope Project, OSU ElectroScience Laboratory Technical Report 531393-1, Jan 1999.

[2] S. J. Lin, On the Scan Impedance of an Array of V-Dipoles and the Effect of the Feedlines, Ph.D. Dissertation, The Ohio State University, 1985.

[3] J. D. Dyson, "The equiangular spiral antenna," IEEE Trans. Antennas and Propagat, pp. 181-187, April 1959.

[4] J. D. Dyson, "The coupling and mutual impedance between conical log-spiral antennas in simple arrays," 1962 IRE International Convention Record, Part 1, pp. 165-182.

[5] H. Nakano, K. Nogami, S. Arai, H. Mimaki and J. Yamauchi, "A spiral antenna backed by a conducting reflector," IEEE Trans.Antennas and Propagat., AP-34(6), pp. 791-796, June 1986.

[6] J. J. H. Wang and V. K. Tripp, "Design of multioctave spiral-mode microstrip antennas," IEEE Trans. Antennas and Propagat, AP39(3), pp. 332-335, March 1991.

[7] J. J. H. Wang and V. K. Tripp, "Multioctave Microstrip Antenna ", US Patent 5,313,216, May 17, 1994.

[8] J. W. Duncan and V. P. Minerva, "100:1 bandwidth balun transformer," Proc. IEEE, Vol. 48, pp. 156-164, Feb. 1960.