One, or two?

A recent article online raised an interesting question about the use of one radial, rather than eight for a 1/4 wave vertical antenna and whether the modelled slight directionality of the resultant pattern for the former would be seen in practice. The author concludes that the real antenna performed in accordance with the model output.

Firstly, it's good to see the model being questioned and not taken as gospel; they are by no means infallible.

I thought it might be an idea to try a side-by-side comparison at 14MHz using WSPR, rather than 'real' QSOs with CW and a very subjective assessment of signal strength by ear or quivering S-meter. The enormous advantage of WSPR is that it is independent of human bias - provided the experimental set-up carefully eliminates other sources of error.

With the fields being rested a little at the moment, I had several acres of free space, devoid of charging cattle and munching sheep, to set-up a peaceful test last evening.

Antennas in position.

 

The test was to compare my standard elevated 1/4 wave vertical (MW1CFN) with sloping elevated radials against an identical vertical but using just one sloping radial (MW6PYS). Both were made of the exact-same wire, mounted at the same height about ground (1.5m), 32 metres apart, had the same slope and final height above ground for the radials, were fed by the same coax type and length (RG8-X, Nevada, 1.5m) and SO239/PL259 connectors. 

Each was fed by separate but identical WSPRlite units sending 200mW 40% of the time. Identical fibreglass poles supported the wires and the ground conditions were a gently sloping short grass field of very moist, rich soil; the water table is about 4 feet beneath the surface at the antennas' location (a well and/or surface breakthrough allows this to be known throughout the year).

The horizon is flat and open, with no buildings or other clutter, apart from a very thin dotting of trees and bushes at a distance. Mostly collapsed wire fencing lies at about 40m from each antenna and seen not to interact to any measurable degree.

Finally, each in-situ antenna was analysed to ensure matching of each was as close as possible to one another. Matching at the 14MHz WSPR frequency for the 2-radial antenna was 1.12:1 (RL 24.67dB) and for the 1-radial antenna, 1.19:1 (RL 21.08dB). For a nominal 200mW output, this translates into 199.32W from the 2-radial antenna and 198.44mW from the 1-radial antenna, a 0.88mW difference. More time would allow for minor trimming of the 1-radial antenna and identical matching, but this small difference remains practically inconsequential.

Unfortunately, the WSPRlites can't be set to 100% duty, which means that two independent units will not very often be transmitting at the same time. But statistics lets us out of that problem.

How did things pan out?  Well, I changed the radial 'pointing' direction from NW to NE during the test, to see whether the claimed directivity of the pattern was seen in respect of the G-to-VK path and, conversely, in the G-to-US path. 

Looking at results to VK7JJ, there's no evidence of directivity. Given the outcome in the final third section of the experiment, when the radial was to the US but the general signal seems unaffected to VK7, it seems the absence of spots for the single radial antenna in the first third and halfway through the second third of the experiment is most probably explained by it having a higher departure angle compared to the two radial antenna, where lower angles were favourable earlier on, when the path was briefly becoming active. I've checked that MW6PYS was transmitting during the 16:45-17:35UT period, and can confirm it was sending every 4-8 minutes or so. Certainly, when the radial was changed to the US direction, there was no sudden drop in the signal to VK7, which in fact increased slightly for a time.

The overall result is a median SNR from the 2-radial antenna of -19.5dB (range of 14dB to -23dB) against a median of -21dB (range of -18 to -26dB). I'd conclude that this is a small but sometimes greater and significant advantage in favour of the 2-radial antenna, especially noting the earlier period of the plot.

RX at VK7JJ

 

Let's check this against another VK station - VK5ARG, a notably good and long-term receiver. Here again, we see no evidence of a meaningful directivity when the radial is aimed at the shortpath great circle direction to VK. There is one point at which the single radial antenna significantly rises above the two radial version and when both transmitters were sending at the same time. But it's an outlier; the overall output is clearly one of perhaps surprisingly little difference between the two antennas, the range of signals being exactly the same for both (-12db to -24dB) and the median received signal at VK5ARG being only 1dB lower for the single radial antenna). I conclude that, for this DX path, at this time, there is no significant difference between the antennas. This is perhaps quite surprising and useful; deploying one wire near the ground is more convenient than two.

RX at VK5ARG

Note from the original article, though, that the author is, like many in north America, interested in north American QSOs. This selects for fairly modest distances of, at most, 3300km, with the majority of QSOs being 1500km or less. I've selected OH8GKP as a modest-distance receiver (2100km) that happens to lie on much the same great circle path to VK. Does a higher pattern and, specifically, a higher peak gain elevation explain the claimed directionality of the 1-radial vertical? Let's see (and don't ask me why Libre Office randomly changes the plot line colours!):

RX at OH8GKP

Well, that's a big 'nope'!  Only changing propagation is leading to the ups and downs (and long absence) of spots in this case.

OK, how about a transatlantic path, over to the eastern side of the US and WA3TTS. Will this kind of distance (~5500km) yield anything useful?

RX at WA3TTS

A cursory look might lead to excitement; the signal seems to drop when the radial is switched away from the US and then rises again when the radial comes back to the US. But the devil is in the detail: the signal has already risen before the radial direction is changed to the US and after a very brief period of broadly matching the 2-radial antenna, drops in strength. This is again probably the result of propagation along different initial elevation angles, although the sustained stronger signal for the 1-radial antenna between ~17:50 and 18:04UT means this is an advantage for that antenna. The overall result is that the median SNR for the 2-radial antenna to this station was -19dB, with a range of -18db to -24dB and for the 1-radial antenna, a median of -20dB with a range of -15dB to -25dB. I'd have to conclude there is no practical difference between the two antennas, though it's again interesting that the single radial antenna performs essentially equally well on this particular path.

A final look, then at N2HQI, which is a shorter period of study as the path only opened to that station a bit later in the experiment:

RX at N2HQI

 Pretty certain there's no directionality of the pattern here, the signal from the 1-radial antenna being heard only briefly. The 1-radial antenna is clearly at a substantial disadvantage in this case.

Overall, then, my conclusion based on this carefully-conducted but rather too brief test is that the 1-radial antenna performs surprisingly well against a two-radial antenna, but that the two radial antenna remains a better choice. For simplicity, space and public safety, and because I operate in moist or wet places most of the time, two radials have never left me wondering if I should use three or more. But these results suggest it's worth trying. That experiment, and the large amount of time analysing and presenting the data it takes, will have to wait for another day...