A Vertically
Polarized Delta Loop for 40m
By Andrew Roos, ZS1AN
Having
recently moved into a new house I found myself without a 40m antenna for use
during the IARU HF contest. Although I am still awaiting permission to erect a
tower, the garden of my new home has several moderately tall trees, so I
decided to erect a temporary wire antenna to go with the paralleled dipoles for
20, 15 and 10m that I had already put up.
One obvious
possibility was to put up another dipole for 40m. As a wise amateur once said,
it takes a very good antenna to outperform a dipole. However all horizontally
polarized antennas suffer from high radiation angles if the average height is
much less than half a wavelength above ground. Although not a problem for local
use, the higher angle of radiation compromises DX performance, a key
requirement for any international contest. Half a wavelength on 40m is about
21m, so I measured my trees and found that the most suitable was only 12.5m
(0.29 wavelengths) tall.
This led me
to look for suitable vertically polarized antennas. Although they suffer more
from ground losses than do horizontally polarized antennas, this deficiency is
redeemed by their ability to emit a reasonable low-angle signal even when
mounted close to ground.
The first
option was a simple quarter-wave vertical. The radiator would be about 10.6m
tall and could be suspended from the tree. However vertical antennas driven
against ground are terribly inefficient unless a decent ground screen of 8 to
32 radials can be installed on or just below ground. This wasn’t possible in
the time available. The need for a ground screen can be avoided by driving the
antenna against a “ground plane” of quarter-wave radials raised 1/10 wavelength
or more above ground, but it just wasn’t possible to install three or four 10m
radials in my limited garden. Not to mention the danger to visitors and pets if
the radials aren’t high enough to prevent inadvertent contact.
So I
finally decided on one of the one-wavelength loop family of antennas. This family
includes the quad and delta loops as well as the rectangle or “magnetic slot”.
Although most commonly erected as horizontally polarized antennas, by changing
the orientation or feed-point they can also be configured for vertical
polarization. In this configuration they have the combined virtues of good
low-angle radiation without excessive ground losses or the need for an
extensive ground screen. The performance of these antennas has been extensively
analysed by L. B. Cebik W4RNL in the excellent series of articles
“Self-Contained Vertically Polarized Wire Antennas: A Family Album” that can be
found on his website www.cebik.com.
The choice
of which member of the family to use was based on practical considerations rather
than performance. The available supports were the 12.5m tree, and my house
about 25m away, with a height at the top of the pitched roof of about 8m.
Fortunately the line from the tree to the house runs roughly West to East, so a
loop antenna erected in this plane would radiate most strongly in the
North/South axis, which tied in with my propagation predictions which showed
that Europe would be the best target area for 40m during the contest. However
because the house was significantly lower than the tree, I could not use an
antenna design which required a long vertical axis on both sides of the loop.
This immediately ruled out the magnetic slot and quad loop, as well as the
related (although open-ended) half-square. However it was shaped ideally for the
triangular delta loop, which could be suspended on its side between the house
and tree so that the long vertical axis was on the tree side, and the feed
point at the apex of the triangle on the house side.
17.8 m
Feed Point
9.4 m
17.8 m
Tree
House
Practical
construction was simple. I tied one end of a length of nylon ski-rope to a
decorative lantern on the east of the house (right-hand side in the diagram),
and tied a couple of heavy fishing weights to the other end. I then threw the
weights over the roof, having first ensured that the pets were safely shut
inside and the YL out shopping. (A cute kitten brained by plummeting fishing
weight is not a good way to endear our wonderful hobby to your life partner.
Life partner brained by fishing weight is even worse.) I then threw the
weighted end of the rope over a suitable branch in the tree and tied it to a
stake embedded in the ground beneath the tree. The rope served as a support for
the antenna itself, which was made of 1.5 mm diameter enameled copper wire. The
bottom of the vertical side was tied to the tree using another short length of
ski-rope to maintain the delta shape. (The ski rope is shown dotted in the
diagram above, while the wire is shown as a solid line).
The antenna
dimensions are approximately 17.8 m on each of the long sides, and 9.4 m on the
vertical axis. These dimensions give a good match to 50 ohms with an SWR of
less than 1.5:1 throughout the 40 m band. As with all antennas, you should
initially cut it long and then progressively shorten it until resonance is
achieved. The feed point is at the apex of the triangle, the far right hand
side as depicted above. Although it is a balanced antenna and should ideally be
fed with a balun, you can feed directly from 50-ohm coax at the risk of some
pattern distortion. If you do feed it without a balun then it does not matter
which leg of the loop is connected to the shield and which leg is connected to
the inner conductor.
I
constructed an EZ-NEC model that predicts a maximum gain of 1.6 dBi
perpendicular to the plane of the loop at an elevation angle of 22°, and a gain of –3.0 dBi at an elevation angle
of 6°, which is the angle suggested by
Moxon as being representative of DX performance. While by no means spectacular,
this should nevertheless provide better DX performance than a flat-top dipole
at the same maximum height of 12.2 m above ground, which has a maximum gain of
5.8 dBi at 51° elevation (resulting in lots of
local QRM) but is over 4 dB worse at 6° with a gain of –7.1° dBi. Both antennas were modeled
over “Medium” ground using the Sommerfeld-Norton model.