ADR measurements at 1.8GHz
A Very Well-Behaved Antenna
David J. Jefferies and Aminuddin Mat Ariff

INTRODUCTION
Dan Handelsman N2DT and David Jefferies had exchanged many dozen emails, on the subject of the design, theory, and simulated behaviour of Dan's Asymmetric Double Rectangle or “ADR” antenna design. L B Cebik, W4RNL is credited for coining the ADR moniker. Dan's designs were an adaptation of the so-called "Hentenna", and the adaptations were intended for the Amateur bands at HF frequencies 1.6 to 30MHz.

Dan reported the ADR behaviour in Communications Quarterly for Winter 1999, pages 97ff, (ISSN 1053-9433). This article of Dan’s was followed by a number of other articles on ADR simulations in several issues of antenneX, for example:

"Loops Part 1"  - www.antennex.com/archive4/May00/May6/qloops1.htm
"Loop Antennas Over Ground" www.antennex.com/archive4/Jun00/Jun6/loops2.htm "
An editorial "The Dark Ages of Quad Design" www.antennex.com/shack/Nov00/dan_edit.htm" in which Dan lets his hair down.

Being constitutionally wary of relying on simulation and thought (theory) alone, David was constantly pressing Dan for real-life measurements. Now, these are clearly difficult to perform on an antenna whose perimeter is over a wavelength on the amateur HF bands, and which cannot be easily isolated from the ground and from adjacent structures (see the discussion thread of March 9, 2001 "What is an Antenna?", toward the bottom of the popular Antenna Theory Forum site on antenneX http://www.antennex.com/atheory/   ).

Dan had reported some measurements in his article(s) in antenneX magazine (see above, op cit), however....

MOTIVATION
David was aware that there were applications for novel simple antenna designs in the mobile communications bands around 2GHz, where the wavelength is 15 cm. As a consequence, David decided to set the task of simulation, design, construction, and measurement of a scaled ADR for 1.8GHz as project for a student on the Master's course in mobile and satellite communications at the University of Surrey. It was thought that the measurements would be much easier on antennas at this microwave frequency, even though the laboratory environment would not be anechoic and would be relatively cluttered. It was realised that there would not be significant metallic obstacles (apart from the feed structure) in the near field region of the ADR under test. Any reflections would be generated by surfaces at least 10 wavelengths from the antenna.

This brief report therefore summarises a selection of our measured results after designing, simulating and building a sequence of ADR antennas for use at low GHz frequencies. Useful design guidelines were provided by Dan Handelsman, for which we are most grateful. We have found that, as predicted, they can be made with large useful fractional bandwidths (50% and greater) for which the VSWR is less than 2. Comparison with other designs, relying on micro-strip patch antennas of a comparable overall size, shows that the ADR antennas are much more broad-band, do not require careful sizing to tune them, do not detune (for practical applications) in the presence of adjacent near-field objects, and have useful almost omni-directional radiation patterns around an azimuth plane at an elevation of a few tens of degrees. They also have enhanced boresight gain over equivalent monopoles and dipole arrangements.

CONSTRUCTION
Figure 1a shows a successful design at 1.8GHz, and Figure 1b shows Aminuddin Mat Ariff's hands holding it, to establish the scale. In this example, the wire diameter is 2 millimetres (14 SWG) and the base is a perspex sheet (not a ground plane and so not coated with copper) to provide the mechanical support. Underneath the perspex base is a half-wave coaxial line balun, cut (taking into account of the velocity factor on the coax) to be resonant at a centre frequency of 1.8 GHz. This balun transforms the 200-ohm radiation resistance of the antenna down to the 50-ohm impedance of the cable and the measuring network analyser, and couples the unbalanced coaxial feed to the balanced ADR antenna.

This is a single example chosen from the fifteen antennas that were constructed and measured.

At 1.8 GHz, the free space wavelength is 167 millimetres (16.7 cm or 6.56 inches) –click here for metrics/imperial chart. The wire diameter is 0.012 wavelengths and made of tinned copper. The intermediate wire is soldered to the rectangular frame.

This design is extraordinarily "well-behaved" and tolerant of fabrication error, except in the placing of the intermediate wire with respect to the top wire, which has to be chosen reasonably accurately. The antenna dimensions are 54 mm in width by 57 mm in height. The spacing of the intermediate wire is 12 mm with all figures quoted to the nearest mm. In terms of the free space wavelength, the height is 0.34 lambda by a width of 0.32 lambda. The perimeter is 1.33 lambda. The spacing of the intermediate wire is 0.072 lambda from the end wire.

VSWR AND BANDWIDTH MEASUREMENTS
Figure 2 shows the measured VSWR of this antenna (red line) together with two slight variants where the intermediate-wire to end-wire spacing is altered to 8 mm (blue line) and to 20 mm (green line). The measurements were made on a Hewlett Packard 8714 network analyser. The system was calibrated with a matched load, short-circuited, and open circuit on the coaxial feed, at the SMA connector where the antenna balun was coupled to the feed line. Thus the curves shown in Figure 2 also display the matching effects of the 4:1 impedance transforming coaxial line balun. Tests on the coaxial feed from the network analyser to the antenna showed that the results were unaffected by objects in close proximity to the lead. This indicated that the balun was doing its job, and that there was no appreciable evidence of radiation from any currents that might have been running on the outside of the cable.

Figure 3 shows an expanded VSWR scale for these three antennas, with the same colour coding as before. We see the VSWR = 1.5 bandwidth, for the 12 mm spacing model, is from 1560 to 2320 MHz, and the VSWR = 2 bandwidth is 1500 to 2360 MHz. The VSWR = 2 fractional bandwidth is therefore 2(2360-1500)/(2360+1500) = 0.445 or nearly 45%.

Examination of the curves in Figure 2 shows that the -3dB fractional bandwidth  (at which the VSWR = 5.8) of the 12mm spacing antenna approaches 63%. This antenna design is therefore wideband, as well as being a well-behaved one.

SIMULATION OF RADIATION PATTERN
Turning to the radiation patterns and the gain, the situation for the simulation is presented in Figure 4. This shows an azimuth (horizontal plane) section for an antenna, with the plane of the antenna placed vertical and the intermediate wire in the horizontal plane, together with a half-space radiation pattern surface. Azimuth at zero degrees is located broadside to the plane of the antenna. The horizontally polarised pattern is shown dotted in red, with maximum gain along azimuth zero, and -3dB beam width of 108 degrees. The vertically polarised pattern is shown in dashed green, with nulls along 0, 90, 180, and 270 degrees and maxima along 33, 147, 213, and 327 degrees. The maximum gain for horizontal polarisation is about 0.5 dBi and the maximum total gain is 2.8 dBi at an azimuth angle of 27 degrees for a slant polarisation at 35 degrees to the horizontal plane.

MEASURED CHECK OF RADIATION PATTERN
Measurements, using a horizontally polarised receive antenna, are shown in Figure 5. As indicated, there is reasonable agreement with the simulation, showing  maxima at zero and 180 degrees azimuth directions, with  a half-power beam width of 99 degrees average for the two boresight directions (compare 108 degrees which was predicted from the simulation).

Looking closely at the three-dimensional radiation pattern, we see that at elevation angles between about 15 degrees and 30 degrees the total pattern is approximately omni-directional in a horizontal plane with gain greater than 0dBi everywhere.

THEORY
Some insight into the behaviour of this antenna may be gained by extending the analysis commonly made of the impedance transformation from the bare dipole to the folded dipole. We recall that the simple folded dipole has impedance 4 times that of the bare half-wave dipole. 4*73 ohms is 292 ohms, which is close to the characteristic impedance of flat twin ribbon feeder.

We recall that this 4-fold transformation comes about because the currents in the driven arm and adjacent arm are the same, and this halves the current from the feed for the same total dipole current. Thus for the same radiated power, the radiation resistance has to increase fourfold because IIR = (I/2)(I/2)4R. Further details may be found on http://www.ee.surrey.ac.uk/Personal/D.Jefferies/dipimp.html

In the ADR case, the adjacent arm is split into two parts, each carrying I/4. Our transformation factor 4 now becomes 8/3 and the expected radiation resistance is 8/3 times 73 ohms = 195 ohms, which is very close to what is measured through the balun.

We can estimate the characteristic impedance of the transmission lines formed by the radiating wires of the ADR. For the close spaced loop this is about 300 ohms. For the wide spaced loop, it is somewhat larger, possibly up to 480 ohms. Starting from a short circuit at the ends of the radiators and transforming around the SMITH chart until we reach reactances of +/- j200 ohms, we expect the half power bandwidth to lie between 48% and 72%. This is also close to what is measured.

DISCUSSION
In his earlier work for the UK Defense Research agency, DJJ had suggested that antenna structures containing coupled arrays of identical dipoles could be made broadband in a similar way to the emergence of energy bands in semiconductors such as silicon, when many identical electron transitions of a collection of Si atoms are coupled together in a solid due to the close proximity of the atoms. This coupling broadens the resonance in a way similar to the coupling of identical-tuned circuits in bandpass filters.

In the case of the antenna, we might expect that the total bandwidth of the Handelsman ADR would approach three times that of a single dipole, as there are three radiating wires which are coupled together strongly by the wires connecting their ends as well as through the near fields.

In fact, the fractional 3dB bandwidth of around 60% is in line with this estimate. It is also worth remarking here that the same arguments apply to Handelsman's prismatic polyhedral antennas in which the bandwidth goes up as the number of elements is increased. There seems to be an important design principle for wideband antennas that has been uncovered by these simulations and confirmed by the measurements. It may also be that near-field coupling alone is not sufficient to provide this bandwidth enhancement, but that some kind of physical transmission line coupling is also needed.

In any case, our experiments on the 1.8GHz ADR show that the class of designs occupies an important niche in broadband antenna technology having approximately omni-directional patterns and corresponding low gains.   Uses for this class of designs may be found in mobile communications applications at these frequencies. It may be that this is an alternative to the more usual methods that employ highly resonant and sensitive micro-strip patch antennas on dielectric substrates. Being very well-behaved, it is likely that the ADR structure may be incorporated into linear and broadside array structures for enhanced gain when this is needed without losing the broadband characteristics.   Also, as Dan has shown, there is a niche for these antennas in HF applications on the amateur bands.

Dr. David J. Jefferies
School of Electronic Engineering, Information Technology and Mathematics
University of Surrey
Guildford GU2 7XH
Surrey, England
D.Jefferies email
Click Here for the Authors' Biography

~ antenneX ~ October 2001 Online Issue #54 ~

Send mail to webmaster@antennex.com with questions or comments.
Copyright © 1988-2004 All rights reserved worldwide - antenneX©
Last modified: December 27, 2003