HF Fractal Horizontal U Antenna: Why EFOC Beats EFHW

This article started with a very practical inspiration: Jean-Claude Ducasse, F1QM, who shared his fractal-like horizontal “U” end-fed half-wave (EFHW) installation. You can see his antenna here on QRZ: fractal horizontal U EFHW by F1QM .

When you model this kind of antenna in EZNEC or any NEC-based tool, it quickly becomes obvious: the geometry of the wire does most of the interesting work in the far-field pattern. Whether you label the feed box “EFHW 49:1” or “EFOC 4:1” doesn’t magically create a new lobe.

But once you move from the simulator to a real garden with real coax and real cores, something else becomes clear: this kind of fractal horizontal U is usually much better behaved as an EFOC / OCF feed than as a classic 80–10 m EFHW on a 49:1.

Geometry vs. label: EFHW and EFOC share the same basic pattern

In NEC, when you draw a thin wire in free space (or over a simple ground model) and apply a source, the far-field pattern is dominated by two things:

• The shape and routing of the wire (geometry)
• The resulting current distribution along that wire

If you keep the wire identical and only move the source from “right at the end” (EFHW) to “somewhere 20–30 % from an end” (EFOC / OCF), you are still exciting essentially the same resonance structure. In most cases:

• The feedpoint impedance changes substantially.
• The absolute current level for a given transmitter power changes.
• The shape of the far-field pattern only changes slightly, especially on the lower bands.

In an ideal NEC model without feedline or common-mode currents, a well-designed EFHW and EFOC using the same wire geometry will produce very similar patterns. The label on the box does not create gain.

Where things diverge strongly is in the real-world implementation: how hard you drive the transformer, how the current maxima line up with the “busy” part of the fractal U, and how ugly some of the upper-band impedances become. That’s exactly where the fractal horizontal U starts to favor the EFOC approach.

What a 49:1 EFHW transformer really faces on a multi-band fractal U

The usual 80–10 m EFHW recipe uses a 49:1 transformer designed for about 2.4–2.5 kΩ at the secondary. A half-wave wire near resonance indeed often sits in the 2–5 kΩ ballpark, so on the primary band this is not a crazy idea.

The problems appear when you:

• Stretch that same wire into a fractal U around the garden, and
• Try to make it work across 80, 40, 20, 17, 15, 12, 10 m from a single 49:1 box.

On an 80 m EFHW:

• 40, 20 and 10 m are near harmonics of 80 m.
• 17 and 12 m are not harmonically related in the same clean way.

Add the fractal routing and varying height, and the feedpoint impedance on those upper bands can land almost anywhere: a few kΩ, tens of kΩ, capacitive, inductive… often all of that across the span of a single band.

That’s why a statement like “17–10 m will be compromised due to the nature of the 49:1” is a good rule of thumb for a generic 80–10 m EFHW on a complex geometry. 10 m can sometimes be tuned into an “ok-ish” SWR window by trimming length very carefully, but that doesn’t magically fix the stress and loss inside the transformer.

The 17–10 m behavior discussed here is based on measurements without the usual shunt capacitor across the secondary. That capacitor is indeed a shunt element – it mostly makes the mismatch look better in SWR, without really curing the underlying impedance problem or core stress.

A good 49:1 on a simple, well-behaved dual-band wire – think 160/80, 80/40 or 40/20 m – can be perfectly usable. The trouble comes when you transplant that same approach onto a geometrically complex, long, multi-wavelength structure.

Why the EFOC feed is naturally better suited to a fractal horizontal U

The same horizontal U that stresses a 49:1 can become a lot more “civilized” if you:

• Move the feedpoint to about 20–30 % from one end, and
• Use a 4:1 current unun plus an appropriate common-mode choke in the system.

What changes?

• You tend to land in a multi-band impedance range of roughly 150–300 Ω on many HF bands, instead of several kΩ.
• The transformer ratio (4:1) means fewer turns, so:
  • Lower copper loss
  • Lower leakage inductance
  • Lower stray capacitance
• Voltage stress on the core and winding drops significantly.

The result is not “perfect 1:1 SWR everywhere” — that’s marketing fantasy — but rather:

• Impedances that are less extreme and much more often in a tuner-friendly range.
• A transformer that is much more efficient, especially on the upper HF bands.

Both EFHW and EFOC versions still need sensible common-mode control: in practice the choke may sit some distance away along the coax and can even be buried. The real gain with the EFOC layout on this fractal U is different:

• The feedpoint split (typically around 27 % / 73 %) puts the strongest current region in the longer ~73 % leg of the wire.
• That longer leg contains most of the fractal “structure”, so the interesting pattern you modeled in NEC is carried mainly by that part of the wire.
• On 17–10 m, the combination of current distribution and feedpoint impedance becomes more reasonable for both the transformer and the radiation pattern.

In other words: the EFOC does not magically remove coax radiation, but it does place the feedpoint where the λ-related current maxima and the fractal geometry work together more naturally across multiple bands.

60 m and 10 m: why they usually behave better on an EFOC

In many “80–10 m EFHW with 49:1” products, 60 m is either omitted or added as a sort of “bonus band” that works only under kind conditions. On a fractal horizontal U, it often:

• Lands in a very awkward combination of R and X at the end-feedpoint, and
• Pushes the 49:1 far outside its most efficient operating zone.

With an EFOC feed:

• The same wire often presents a much more moderate, tunable impedance on 60 m.
• The 4:1 transformer sees less extreme reactances and can work more linearly.

On 10 m, both EFHW and EFOC versions of a long fractal U will produce patterns full of lobes and nulls. That’s simply what happens when your wire is several wavelengths long. The EFOC doesn’t “fix” the lobes — it just makes the feed system and current distribution play nicer with that geometry, while keeping the transformer in a saner operating region.

Practical design recipe for a fractal horizontal U EFOC

If you like the pattern your fractal U produces in EZNEC, here is a practical way to turn it into a robust real-world installation:

• Keep the same wire geometry that modeled well (horizontal U / fractal around the garden).
• Choose a feedpoint roughly 20–30 % from one end. This is where many good 80–10 m OCF recipes live.
• Use a current 4:1 transformer rated for your power level.
• Include a serious common-mode choke in the feed system (even if it sits some distance away or is buried).
• Accept that you will still need a tuner, but it now has a much easier job.

You can then design with all the usual HF bands in mind — 80, 60, 40, 20, 17, 15, 12, 10 m — and let the tuner clean up the remaining mismatches instead of trying to make a single 49:1 solve everything at once.

In that sense, the conclusion is simple: the fractal horizontal U is not “better” because it’s EFHW or EFOC. It’s better because the geometry is good, and the EFOC feed system is a better technical match for that geometry and current distribution across many bands.

Interested in more technical content? Subscribe to our updates for deep-dive RF articles and lab notes .

Questions or experiences to share about EFHW vs. EFOC on complex wire geometries? Feel free to contact the RF.Guru team .