The wires are all visible in the picture. That you can only see 3 of them is not a mistake. Some explanation is in order.
The wire going down to the left is half of an inverted vee for 20 meters. It is tied to the edge of the roof with a length of nylon rope. The other half of the vee is tied in the same fashion to the top of the tower at Site-C. That's the tower supporting the TH1vn tri-band dipole. The apex of the vee is ~14.1 meters above grade, with south (tower) end ~12 meters up and the north end ~10.2 meters up.
Total length of the antenna is 10.2 meters of insulated #14 stranded copper wire. Its resonant frequency is close to 14.0 MHz with an SWR of 1.2, and it stays below 2 across the band. The average height of the antenna current is approximately 13 meters above ground. This is 2.5 meters (or close to 25%) higher than the dipole. This matters since it is the current that determines the far-field pattern.
The third wire is 4.05 meters of the same wire. Its low point is also 12 meters. It is tied to tower as well, but lower down. Ignore this wire for now since it has negligible effect on 20 meters.
The orientation of the antenna is not ideal. I did what came easy also put the wires as high as possible. The antenna favours east and west directions, and so is -3 db towards Europe and west Asia, and worse towards central Asia. Nominal (modelled) gain in the major lobe is 7.8 dbi at a heading of 90° and elevation of 25°. The pattern is slightly asymmetrical, as expected from their positions. The null to the south is deeper than I'd like: -4 dbi. It is more omnidirectional than a dipole though not by much.
My purpose in putting up this antenna was to test whether a little more height and bending down the legs would outperform the TH1vn tri-band dipole at a height of 10.5 meters. In particular:
- Would the 25% increase in height have a noticable impact on low-angle DX performance. The TH1vn models with a maximum gain of 7.2 dbi at an elevation of 30°, and only negligibly lower at 25°. The antenna favours South Europe, at a compass direction of 70°.
- Would it be more omnidirectional. Wire antennas don't rotate, and I want to cover as many directions as possible, even at the expense of stateside QRM.
To be brief, the inverted vee works well but is not better than the dipole at the lower height. The EZNEC models appears to reflect my experience with the antennas. In one way this is unfortunate. On the other hand it does mean the modelling can be relied upon, including for future antenna designs at my location.
The highlights of the antenna comparison are as follows:
- On the longest paths (4J, FK8, VK, ZL, JA, BY, etc.) the antennas are roughly equivalent, though not in all cases. The vagaries of multi-path and other propagation effects are noticable even though both antennas are (mostly) the same polarization and in the same vicinity.
- The off-peak orientation of the vee does discriminate against much of Europe, as the model shows. Europe is typically 1 S-unit stronger on the dipole. This is a shame since that is the direction for the bulk of DX paths. However, for some reason northern Europe (TF, OH) slightly favours the inverted vee.
- On polar (north) paths the dipole wins. This includes UA0, BY, VU.
- There is some variation on southern paths, perhaps due to the different positions and depths of the side nulls. Since it can vary by 2 S-units it is a good idea to try both antennas and choose the best one for each station.
- Since the antenna is not up a multiple of ½-wavelenght there is lobe that points straight up. This was confirmed by the relative strength of many mid-distance W and VE stations. This is not desirable.
Now on to that third leg I mentioned earlier, the final half of this 1.5 inverted vee antenna. I had this odd notion that it might be possible to add a band with only one more wire. If the length is suitably chosen it should form an off-centre fed inverted vee in combination with one, or even both, of the legs of the 20 meters inverted vee.
The reasoning is that any combination of wire legs that has a high impedance on the target band would carry little in the way of antenna current. I decided to try this since it is easier than building a full fan dipole, and I was having some difficulty managing modelled interactions between the wires going to the tower. The closer the wires in a fan dipole come to each other the more they interact and affect the lengths of all antennas except the one for the lowest frequency band.
Since time was short I only modelled the SWR for the third wire, with the aim of making it resonant on 17 meters. I found that the closeness of the legs going to the tower had a large impact on the resonant frequency on 17, but almost no affect on 20. The spread of resonant frequencies went from 17.5 to 18.5 MHz over the range of the tower tie point for the 17 meters leg. You can see how close they are in the photo and model up above.
With that much sensitivity some experimentation is necessary. First, in a quest for height, I tied the rope off at 8 meters height, quite close to the tie point for the 20 meters leg -- the top of the tower is 8.8 meters above ground. That was too close. Then I tried going very low, at the 4.5 meters level. This swung too much in the other direction, much as the model showed. So I split the difference and had a 17 meters antennas that resonated at 18.050 MHz. I decided that was close enough since the band is so narrow. SWR is 1.2 at 18.068 MHz and 1.5 at 18.168 MHz.
I put it on the air and it worked out just fine. Not great, but ok. Stations went into the log. Then I got around to modelling the pattern. That was not fine.
The antennas showed a pretty typical pattern shape for an inverted vee. But something was amiss since the gain in the main lobe was only 3 dbi, several db lower than expected. I then modelled the currents and had a closer look at what was happening.
Imagine my surprise when the image at right appeared! The antenna current chose the adjacent 20 meters leg rather the far one, which I had incorrectly assumed would form the other half of the antenna. Moving the 17 meters legs up and down had little impact on the current in the far 20 meters leg.
This explains the low gain and (as I eventually noticed) the smaller than expected SWR bandwidth. The latter is unimportant on 17, while the former is a problem. I have not investigated further to understand what is happening. Regardless of the reason it seems my "innovation" is a sham. I may have to add bands in a more traditional fashion rather than taking shortcuts. Still, it is in a way quite interesting.
I'll end this article by describing the full extent of the EZNEC model. The model includes all the metal elements in the mast, going all the way to ground level, plus the coaxial cable. The coax is modelled as a wire (the insulated outer conductor of RG-213/U) that is not connected to any other at its ends. To keep things simple this assumes that the air-core coax chokes at the top and bottom are perfect on all bands of interest -- infinite impedance. House metal, in particular the aluminum eaves (my original antenna!) and soffits, are not modelled (at least not yet), nor is the Site-C tower and antennas.
The view of the complete model is shown at right. The perspective is that of an observer to the north-northwest and above the height of the mast. Hopefully it is recognizable.
Antenna currents are small on all these additional conductors on both 20 and 17 meters. I was concerned that the partly-vertical profile of the antenna legs could couple to the mast and coax. For the present I can ignore their small effect.