Fall Antenna Work Starts

With the contractors gone I have to catch up to 2 weeks of lost time. That is, time not spent on antenna work. Although work on the antennas wasn't possible I was still able to do computer modelling with EZNEC. I have many ideas brewing in my head, some for implementation this year and some that must wait until next year at the earliest.

In 30 minutes Sunday morning I pulled down the TH1vn tri-band dipole and the 1.5-bands inverted vee. The inverted vees  are down for at least a few days as I put my new plan into play. The TH1vn is now back on the tower but at a lower height where I can easily work on the feed system over the next week or two.

First I'll talk about the TH1vn. As mentioned in the original articles I wrote about this antenna, it is the driven element of my ancient TH6DXX yagi. When I put it up a few months ago I mentioned some odd behaviour on 15 meters. This eventually was discovered to be an intermittent connection...somewhere. I suspected one of the traps for 15 meters, a guess I based on the large effect on that band but lesser effects on 10 and 20.

Doing a search on the web turned up some scary stories of the difficulties of working on these traps, especially those for the driven element. Once I had them in my hands I discovered they came apart quite easily, and that they are straightforward to service. I found no obvious faults. I torqued all the screws, rebuilt the element clamps with stainless steel bolts (the old, rusty ones fell victim to my bolt cutters), and put it back on the tower. It now works as it should.

If the dipole stays problem-free for a few days I will again bring it down for my next experiment, which is to add 17 meters to it. I'll talk more about this in the future after I've made the modification. The modelling for the added band was straight-forward. However I need to consider a lightweight, low-cost and robust mechanical design. The antenna must last for at least the winter. Next year (2014) I may subject it to more modifications for additional performance.

Once the dipole is operational on 20, 17, 15 and 10, it will be time to put up a delta loop for 40 meters. If you've followed along you'll know that I decided on the delta loop as the best of several options for decent DX performance on that band.

Now on to my experiment with the multi-band inverted vee. I chose to do some research and model designs with EZNEC before tackling "cut-and-try" again. Like most hams I used to mostly rely on cut-and-try with dipoles and fan dipoles even though there were instances of deleterious interactions. Since the elements for each band would need to run in the same vertical plane -- the case where interactions are most intense -- I wanted to avoid a lot of physical labour, and frustration.

One thing I did know is that when the elements of a fan dipole (or inverted vee) are maximally separated the interactions are small enough that cut-and-try is a reasonable approach. Some adjustment is usually needed in any case, since even with modelling software there are inevitable interactions with ground, building wiring and so forth. When in the same vertical plane, tuning can be highly sensitive to small changes in wire position.

To this end I performed a couple of modelling experiments. First up is a 30 plus 20 meters fan dipole with a common feed point. In this configuration the 4 wires move out radially from the feed point, in accord with my chosen mechanical feed point design. The model uses dipoles in free space to aoid additional variables. Those can be added to model once a suitable design is selected.

I rotated the 20 meters dipole in 10° increments from 90° (dipoles are orthogonal) to 10°. In the orthogonal start position I "trimmed" the dipole so that the resonant frequencies were 10.125 and 14.100 MHz. The table shows what happens as the 20 meters dipole is rotated.

30 & 20 Meters Fan Dipole – Free Space – Radial fan-out
Angle	30 Meters		20 Meters
	Fr (MHz)	R (Ω)	Fr (MHz)	R (Ω)
90°	10.125	75	14.100	71
80°	10.100	70	14.130	73
70°	10.075	65	14.170	74
60°	10.050	62	14.220	73
50°	10.025	59	14.270	70
40°	10.000	56	14.360	64
30°	9.980	54	14.460	58
20°	9.970	51	14.610	50
10°	9.950	47	14.870	42

Although the affect on the lower frequency antenna is modest, on the higher band the interaction becomes severe as the angle between wires is lowered. This helps to explain the tuning sensitivity. Worse, since the 20 meters dipole would have to be far longer than a naive calculation would give, which make the "cut" part of cut-and-try difficult to put into practice!

There is a better way to construct a fan dipole, one where much of the sensitivity is eliminated. The big hint to me was the long-known technique of using ladder line as a two-band dipole. I do not mean a ladder-line feed system, but ladder-line for the antenna elements. The feed line is coax (with a common mode choke).

You choose a length of ladder line for the lower of the two bands (dipole formula). Then cut the line on one of the leads so that it is the proper length for a dipole on the higher of the two bands. Do this for each leg of the dipole. Tie the two sides of the ladder line together at the feed point, and connect those two to the coax feed line.

The model of the feed point looks something like the adjacent diagram. This is the second model I used for my parallel-wire fan dipole. I then modeled it in EZNEC.

As a starting point I used a 5 cm separation between elements and tuned the antenna for resonance as in the preceding case. Then the separation was increased in 5 cm steps up to 30 cm. To do this I left the horizontal lengths unchanged, only increasing the length of the vertical connecting wires (wires #4 and #5 in the diagram). As before I used free space. The 30 meters dipole is on top, connected to the feed line.

30 & 20 Meters Fan Dipole – Free Space – Parallel fan-out
Separation (cm)	30 Meters		20 Meters
	Fr (MHz)	R (Ω)	Fr (MHz)	R (Ω)
5	10.125	74	14.100	53
10	10.125	76	14.050	59
15	10.125	76	13.980	61
20	10.125	76	13.910	62
25	10.125	75	13.830	63
30	10.125	75	13.760	63

This is much better. Notice how the 30 meters dipole is unaffected by the separation, including any affect from the connecting wires #4 and #5. The impact on the 20 meters dipole is incremental, roughly in proportion to the length of the connecting wires. (Recall that the source is connected to the 30 meters dipole.)

For the modelled feed configuration and range a quick calculation shows that the equivalent length of each 20 meters dipole leg is 0.4x the length of wires #4 and #5 (the separation distance). For example, at a separation of 30 cm the equivalent lengthening of each of the 20 meters legs is 12 cm. I checked this by subtracting 12 cm from each leg of the 20 meters dipole and the antenna model resonated at 14.075. However, the impedance remained at 64 Ω. This is a curious benefit since we do not want the impedance to drop too low or the real antenna, when built, might otherwise have a higher SWR.

You should expect that the length ratio will be slightly different if wires #4 and #5 are a different distance apart. In my model the value is 8 cm (3").

The reason for the lower impedance on 20 meters is that current is induced on the other dipole. This also adds some gain to the antenna, a small 0.8 dBd. The reverse is not true, so the impedance on 30 meters is near to the nominal value of 73Ω in free space and the gain is close to 0 dBd.

This is the model I plan to build for the next iteration of Site-B inverted vees.