Small Loop Antenna Array for HF Reception

Matt Roberts - matt-at-kk5jy-dot-net

Published: 2016-03-24

Updated: 2017-01-16

While working with small transmitting loop antennas, I noticed that there has been considerable discussion (and disagreement) regarding receive-only loop antennas, and low-gain receive antennas in general.  The same inductive feed loops used in transmitting loop antennas are sometimes used in larger sizes, in isolation, as receiving loops.  This seems to be most common for direction-finding applications, but also for dedicated receive antennas on lower frequencies, such as 160m or 80m.  Nonresonant lossy loop antennas have some nice response characteristics that make them ideal when used for reception of skywave signals, and some very successful high-performance receive loops are now being offered as commercial products.

There is also considerable research in the field of phased arrays of nonresonant vertical antennas, for use as HF low-band receiving antennas.  Antennas such as the four-square and two-element end-fire arrays have all been used with short, lossy vertical elements, to provide highly-directional, low-noise receive antennas for the lower HF and MF bands.  Beverage arrays, another vertically-polarized antenna, have also been used in broadside arrays to provide similar performance where space for such devices is available.

While doing EZ-NEC design for a two-element array over vertical monopoles, I decided to substitute a small, vertically-polarized loop element in place of the vertical element.  As it turns out, this substitution provides significant benefits over an array of vertical whip antennas with the same spacing.  Both antennas can be made broadband by keeping the elements reactive and lossy, but the loops have additional benefits:
  1. The elements are self-contained, and do not require radials.
  2. The resulting pattern has a ~50% narrower azimuth-plane beamwidth.
  3. Small, directly-fed loops are DC-shorted feedpoint antennas, with associated immunity to static buildup and lightning.
The first point above is mostly mechanical.  It minimizes the real-estate needed for an array of any spacing.  The second point is more interesting, because it allows a two-element array of loops to provide a much more focused pattern than the same array would when using vertical elements.  The predicted beamwidth is on par with a one- or two-element beverage antenna many times longer than the loop array spacing.  This enables very small arrays to be extremely effective even for long wavelengths.

For the plots below, the following construction is used: The preamplifier is a critical component, not so much for recovering sufficient signal levels, but to isolate the antenna from the delay lines.  The delay lines cannot be allowed to run with any significant SWR, because doing so causes wave interference effects that destroy the pattern of the antenna.  Tests with single 30" loop elements show that these elements collect more than enough signal to set the noise floor in any decent receiver on 40m and 80m in a suburban environment.  So unless then antennas are located in a very quiet rural area, a unity-gain amplifier would be more than sufficient to isolate the antenna from the delay lines.

The EZ-NEC model of the loop pair is shown in Figure 1.  The main lobe of the antenna pattern projects in the direction of the positive X axis.

2:1 Signal Combiner
Figure 1: EZ-NEC 3D Wire Model

Plots are given for five different wavelengths.  For each wavelength three different plots are given: Each plot is computed with the antenna above fairly good ground (0.0303S/m, and a dielectric constant of 20).

The phase delay between the elements is adjusted for best low-angle front-to-back ratio for each band.  The performance for each band is remarkably similar.  Note that overhead response is considerable for all of the bands that support NVIS.  This allows for use of the antenna for communications both at DX angles and within the skip zone.

The bands plotted, associated optimal delay line lengths, and the matching physical delay line length for Belden 8241 are given by: Other cable types will require recalculating the desired delay using the velocity factor of that cable.

Note that any length of delay line can be used between the preamplifiers and the combiner or mixer; the critical component is the difference between the length of the lines between the preamplifiers and the combiner.  For lines that are too short for the 20' spacing, simply add the needed length to both cables, which will give the same effect.  There are other ways to accomplish this with shorter lines, but this is the most straightforward, and is only an issue with lines that are already short, such as the 20m example above.  Depending on the installation details, another option is to add 360° of cable to one of the delay lines, if this makes more sense mechanically.

The delay line length becomes particularly critical as the wavelength increases, because the 20' spacing is a progressively smaller fraction of one wavelength.  That said, constructing an accurate delay line with the needed resolution is just as easy with long wavelengths, because each degree of length is larger at longer wavelengths.  Longer wavelengths also require longer delay lines, with 160m requiring more than 160' of length differential between the feedlines.  Compensating for this is the fact that good quality 75-ohm cable has very small loss at such low frequencies.

The azimuth-plane beamwidth is consistently close to 90°, especially at the lower angles, and regardless of band.  This is possible because the loop antennas each have nulls to the side, whereas vertical monopole elements have a single null overhead.  The side-facing nulls help to "squeeze" the overall pattern to provide much better directivity than with an equivalent array of monopoles.

Example elevation plots are shown below.  Note that the so-called "take-off angle", or angle of maximum response, is lower for longer wavelengths.  More importantly, the lower -3dB angle also drops for the lower bands, giving strong response at the low DX angles, well into the single digits.

20m Elevation Plane
Figure 2: 20m Elevation Plane

30m Elevation Plane
Figure 3: 30m Elevation Plane

40m Elevation Plane
Figure 4: 40m Elevation Plane

80m Elevation Plane
Figure 5: 80m Elevation Plane

160m Elevation Plane
Figure 6: 160m Elevation Plane

The magnitude of the rearward-facing minor lobe is a function of the magnitude of the delay.  By making small adjustments to the delay, the minor lobe can be made even smaller than shown here.  The trade-off is in the rearward-facing higher-angle response of the antenna.  The delay lines shown here were selected so that the 180° response was roughly the same magnitude as the two quartering sidelobes at low angles.

Here are the azimuth plots for the 15° elevation plane.  Note that the 3dB azimuth beamwidth is close to 90 degrees on all bands, and even less on the lower bands.  As a result, the antenna has excellent directivity for DX work, regardless of the wavelength.

20m Azimuth Plane at 15 degrees elevation
Figure 7: 20m Azimuth Plane (15° elevation)

30m Azimuth Plane at 15 degrees elevation
Figure 8: 30m Azimuth Plane (15° elevation)

40m Azimuth Plane at 15 degrees elevation
Figure 9: 40m Azimuth Plane (15° elevation)

80m Azimuth Plane at 15 degrees elevation
Figure 10: 80m Azimuth Plane (15° elevation)

160m Azimuth Plane at 15 degrees elevation
Figure 11: 160m Azimuth Plane (15° elevation)

For comparison, here is the azimuth plane at the elevation angle of maximum response.  The shorter wavelengths have wider bandwidths in this plane, because the plane is at a relatively high angle.  Compare this to the azimuth plots at 15° elevation.  At the lower elevations, the directivity remains intact, even on 20m.  This directivity is important for noise and interference rejection.

20m Azimuth Plane
Figure 12: 20m Azimuth Plane (Peak Elevation Angle)

30m Azimuth Plane
Figure 13: 30m Azimuth Plane (Peak Elevation Angle)

40m Azimuth Plane
Figure 14: 40m Azimuth Plane (Peak Elevation Angle)

80m Azimuth Plane
Figure 15: 80m Azimuth Plane (Peak Elevation Angle)

160m Azimuth Plane
Figure 16: 160m Azimuth Plane (Peak Elevation Angle)

The azimuth response on shorter wavelengths can be narrowed by moving the antenna elements closer together.  The azimuth response is partly a function of the element spacing as measured in wavelengths.  On 160m, the 20' spacing is less than 4% of one wavelength, while on 20m, the 20' spacing is more than a quarter wavelength.  A 10' spacing on 20m would narrow the azimuth response by several degrees.  The plots shown above assume a constant 20' spacing, where only the delay lines are changed in order to change bands.  The resulting responses, while excellent, are somewhat of a compromise between good element spacing and delay line criticality across the four bands of interest.  Again, the focus of the 20' spacing is to optimize pattern shape for the longer wavelengths.

Single Loop Element
Figure 17: Loop Element

DXE RPA-1 Preamplifier
Figure 18: RPA-1 Preamplifier

2:1 Signal Combiner
Figure 19: 2:1 Signal Combiner

The 40m version of this antenna has been constructed, and shows significant directivity at low angles, but is also responsive to NVIS signals arriving at very high angles.  The noise response is greatly reduced, even with preamplifiers, yielding a pleasant and consistent signal-to-noise ratio.

Each element is a 30" square (10' of wire, total, per element), with a 1:1 choke balun at the feedpoint.  The choke is a QRP choke balun from Balun Designs.  The feedline is Belden 8241 (RG-59) 75-ohm coax.  I chose this cable because it is one of the few 75-ohm cables that has a solid dielectric, rather than foam.  The hope was to avoid water intrusion, although newer flooded cables have excellent water resistance when used with appropriate connectors.  Also, the 8241 has very detailed specifications available, whereas RG-6 or RG-59 from the local home improvement store may be hard to characterize.  Having accurate specifications with respect to the cable's velocity factor is very important for cutting accurate delay line lengths.

Each element has a mast-mounted preamplifier.  For the first draft, I chose the RPA-1 from DX Engineering, because it has excellent IP3 and low-signal characteristics.  There are, however, many different preamp kits available that can work just as well and are far less expensive.  DXE currently sells such kits, alongside their RPA-1.

The signal combiner is a DXE RSC-2.  This is a two-port passive splitter/combiner module, impedance-matched for 75-ohm cable.  The combiner needs to be impedance-matched to the delay lines and to the preamplifier output, to ensure low SWR within the delay lines.  This preserves the antenna pattern by eliminating reflections within the delay-producing feedlines.  An active combiner could also be used, and it could be something as simple as a single RF op-amp used as an additive mixer.  If an active device is used, the same rules apply -- the mixer inputs must be impedance-matched to the feedlines to avoid reflections.

I am using coaxial delay lines alone to set the delay value.  This involves cutting lines that have a specific electrical length.  Rather than cut a single cable for each band, I made a set of "jumpers," whose lengths are the set of a binary series: 6", 1', 2', 4', 8', 16', 32' and so on.  This allows any delay line length to be easily constructed with inexpensive barrel connectors, and adjustment of the line length can be done easily by adding or removing a segment.  A more flexible solution would use a phasing controller such as the DXE NCC-1.  This would allow instant reversal of the direction of the antenna, and more precise fine-tuning the pattern for best directivity and response.  Given that this is a $600 device, I am happy to use the less expensive option for now.  Reversing the direction of the antenna involves swapping which antenna is fed by the longer line.

Using EZ-NEC 6.0, I was able to determine the average gain and compare it to the maximum gain of the antenna.  Using the W8WWV formula for RDF, I calculated the RDF for 40m and 80m to be 9dB and 9.2dB, respectively.  This is on par with a well-designed beverage over 300' long.

Tests are ongoing, but the initial results are very promising.  The effective performance on 40m is good enough to be compared to a 2-element beam antenna.  The difference here is that this beam antenna can easily fit inside a house or attic space, and maintain its DX-friendly performance.  I have also experimented using a single element without a preamplifier, which can give nice bidirectional performance, and features a much simpler feedline system.

My main goal for these antennas is to use them for reception of the lower bands (40m, 80m and 160m), to obtain better S/N performance than can be achieved from typical resonant transmit antennas.  This should allow me to optimize the transmit and receive antennas independently, to get the best performance from each.  Since effective receiving antennas are the most difficult on the low bands, having the flexibility to pick the best antenna for TX and RX independently is a nice alternative to having to build a huge tower to support combined TX/RX antennas that can fill both roles well.  As the various suburban noises come and go, I expect to find that sometimes it will be nice to have a low-band receive antenna performance on higher bands, like 20m, as well.

Copyright (C) 2016-2017 by Matt Roberts, All Rights Reserved.