The W8NX “dominant element” 5-band dipole

Choosing an HF antenna was something that occupied my time through summer 2011.

Though I have a fairly large suburban lot, the number of antennas that could go up is limited by not wanting to have the space above my house full of clutter. I tossed around several ideas, including buying a commercial vertical or a multiband “fan” dipole. Two designs found via Internet browsing caught my fancy: the W8NX family of multiband dipoles, and the VE3HKY “semi-vertical” antenna. More on VE3HKY in a future post.

In the end, I settled on the W8NX design because technically it’s pretty nifty. The source for the five-band design is in the March-April 2004 issue of QEX, in an article titled “Dominant-Element-Principle Loaded Dipoles.”

Al Buxton W8NX is a retired electrical engineering professor who for many years was an active contributor to ARRL antenna publications. He definitely knows his stuff. The family of antennas he invented includes one that’s been in several editions of the ARRL Antenna Book. It’s an 80-40-17-10 or 160-80-40-15 meter dipole that uses one pair of traps. The length is slightly shorter than a half wavelength on the lowest band.

How does Buxton get four bands out of one pair of traps? Well, first of all, they’re not traps, and they’re mostly not used at resonance. They embody two very nifty tricks used to tune the antenna to multiple bands.

(The tricks are explained after the cut in a l-o-n-g post.)

A half wave dipole resonates at its fundamental frequency and odd harmonics of the fundamental. At the design frequency, roughly
L (ft) = 468 / F (MHz),
the antenna shows an impedance of Z = R + j0, where the radiation resistance is 72 ohms in free space at the fundamental, lower in configurations like the inverted vee, and higher on the odd harmonics. The reactance at the resonant frequencies is zero, neither capacitive nor inductive.

Now, to radiate, an antenna does not need to be an exact physical half-wavelength, nor does it need to be resonant. A dipole driven a bit above its resonant frequency has positive (inductive) reactance, and just below resonance, negative (capacitive) reactance. A parallel LC circuit, on the other hand, has negative reactance above resonance, positive reactance below, and at the resonant frequency, is effectively an open circuit.

Now, what happens if we want one antenna for 80 and 40 meters? Traditionally, one would put a resonant trap at 10 meters from the center insulator, λ/4 on the higher band, tune it to the middle of the 40 meter band, and adjust the wire beyond the trap to make the total length 20 meters per half antenna. This works okay but the SWR changes rapidly with frequency on 40 meters. What happens if we tune that same parallel LC circuit to, say, 6 MHz? Now this “load,” to use Buxton’s term, can be placed somewhere on each 80 meter half-dipole so it cancels out the reactance of the wire at BOTH 3.8 and 7.2 MHz. In Buxton’s words, we have doubled the fundamental resonance. This is the first nifty trick. Where do the other bands come from? They are the 3rd and 5th harmonics of the uncompensated antenna. Far away from resonance, the load reactance has little effect on the electrical length of the antenna. For 80-40-17-10 coverage, this means the length of the antenna is roughly the same as a dipole tuned to 6 MHz.

Finding the right load reactances in the ham bands, and the right wire lengths, is all done through optimization software.

W8NX describes antennas covering all the way up to nine bands in his QEX article. He runs an eight-band version (no 60M coverage) at his QTH. Most of these designs have a lowest band of 160 meters, but since I don’t have the room for a 200-foot antenna, I chose the five-band version, covering 80-40-20-15-10. At 111 feet long, it’s a reasonable size. Here’s a schematic:

Just to explain briefly, the fundamentals are on 80 and 40, the third harmonic resonance on 20, the fifth harmonic on 15, and the seventh on 10 meters. The stub near the center insulator is there to tweak the 10 meter resonance to the SSB part of the band. The left-hand half of the antenna is the mirror image of the right-hand side. The entire length of the antenna radiates on each band, and the patterns are supposedly what you’d expect at each harmonic of a standard dipole. I used Buxton’s design software to pick slightly different frequencies, and to use 12 AWG wire instead of his 14 AWG.

The loads are wound out of coax on a PVC pipe form, where the L comes from coupling between the shield on adjacent turns, and the C comes from the capacitance of center conductor to shield. So, for a given length of coax, the capacitance is fixed, and the inductance depends on the spacing of the turns, somewhat tweakable. Since the loads are not operated at resonance, a Q in the range of 100 to 200 doesn’t matter too much.

Here’s the second nifty trick: it’s not always possible to get the desired reactances in the ham bands from a length of coax connected at the ends of the coil. So Buxton adds a tap on the shield on the “output” side, i.e. farther away from the center. This leaves the resonant frequency of the load unchanged, but changes the L-to-C ratio into something more suited to the canceling reactances. Here’s a diagram.

Let’s put this all together for the four-band antenna. At each desired resonant frequency, each half of the antenna ideally meets the constraint:
X (outer wire length) + X (L || C) + X (inner wire length) = j0.
That is, the components of X cancel. The optimization is for four equations of four variables: that is, calculating reactances at four frequencies where the variables are the two wire lengths, the load inductance and the load capacitance.

The stub in the five-band version is located at a high-voltage point and offers a parallel capacitance to ground. Instead of six degrees of freedom, fixing the location of the stub on the inner wire gives five equations in five variables.

This design has only two disadvantages that I can see. first, in the harmonic modes, 20 through 10 meters, the antenna has a multi-lobe pattern just like a conventional dipole.

The second disadvantage is that the W8NX design does nothing to change the radiation resistance of the antenna compared to a conventional dipole. So the resonant impedance is 72 ohms at the two fundamental frequencies, rising to above 100 ohms for the higher harmonics. Buxton waves his hands at this and suggests using RG-59 coax. In a 50-ohm system, it means the minimum VSWR will be around 1.5. This is acceptable to me, though.

Readers wanting to find out more will have to locate the March/April 2004 issue of QEX on paper or CD. To my knowledge, W8NX’s article has not been published online. ARRL members can get three antenna articles by W8NX from the QST Archive on the ARRL website. These have specific details but not the design method. Some of Al Buxton’s work also appears in recent editions of the ARRL Antenna Book and a recent volume of the ARRL Antenna Compendium.

The optimizing software is available on the ARRL website, QEX Files section, as 0403BUXTON.ZIP. It requires GWBASIC to run, which is easy to find online. Run it from a DOS window on Windows XP or earlier. For some reason, Windows 7 doesn’t like GWBASIC, even in compatibility mode.

To end the post, here is the output of the antenna design program DEP5B.BAS for my design.

And here is the load design calculated by SINGCOAX.BAS. Primitive software, but the theory behind it looks sound.