Build a W8JK Bidirectional Array
The W8JK — named for its inventor, John Kraus, W8JK — is one of the more surprising results in classic antenna engineering: two half-wave elements spaced much closer together than a normal array, fed 180° out of phase, deliver real broadside gain from a footprint far smaller than the spacing a conventional array would need. The result is a compact, rotatable "flat-top" wire beam that fits on a boom a fraction of the length a Yagi of similar gain would require. This guide covers the complete build from element and phasing-line dimensions through feedpoint assembly and tuning.
The Counterintuitive Discovery
Conventional array theory suggests that two elements need meaningful separation (a quarter-wave or more) before they contribute useful gain to each other. Kraus's research in the 1930s found that if two close-spaced collinear half-wave elements are instead fed 180° out of phase, the array produces gain comparable to — and in some spacing ranges better than — much more widely spaced in-phase arrays, all while occupying dramatically less physical space. This is the entire premise of the W8JK: get array gain without needing array-sized real estate.
Why the Pattern Is Bidirectional, Not End-Fire
Even though the elements are fed out of phase like an end-fire array, the close spacing changes the geometry of how the fields combine. The result is a figure-eight pattern with two main lobes broadside to the line connecting the two elements (off to the sides, not off the ends) — the same general shape as a single dipole's pattern, but with meaningfully more gain in the favored directions and a useful null off the ends. Rotating the whole array changes which two directions are favored, exactly like rotating a Yagi.
The Phasing Line — A Crossed Feeder, Not a Matching Stub
Because the elements are so close together, achieving the 180° phase reversal doesn't require a quarter-wave phasing stub the way many other phased arrays do. Instead, the two-wire feeder connecting the centers of the two elements is simply transposed (crossed) once between them — this single crossing flips the phase of the second element relative to the first. The main feedline then taps into this phasing line, typically at its center, and runs to the shack as balanced open-wire line.
- Spacing vs. gain tradeoff: wider spacing (up to about 0.5λ) generally increases gain somewhat, at the cost of a longer boom and a longer phasing line to build precisely.
- Practical builder's choice: 0.125λ spacing is a well-documented, forgiving starting point that keeps the array compact while still delivering solid gain over a single dipole.
Feedpoint Impedance and Matching
The feedpoint impedance at the center of the phasing line depends on spacing and element length, typically landing in the range of a few hundred ohms — well suited to balanced open-wire or ladder-line feed into an antenna tuner, rather than a direct 50Ω coax match. Most builders feed the W8JK the same way as a simple center-fed doublet: open-wire line to a balanced tuner or a balun-plus-tuner combination at the shack entry.
| Band | Element length (ft) | Spacing at 0.125λ (ft) | Notes |
|---|---|---|---|
| 40m (7.15 MHz) | 65.5 ft | 17.2 ft | A large array — needs two well-separated masts or a long boom |
| 20m (14.15 MHz) | 33.1 ft | 8.7 ft | The classic rotatable W8JK band — fits on a modest boom |
| 17m (18.1 MHz) | 25.9 ft | 6.8 ft | |
| 15m (21.2 MHz) | 22.1 ft | 5.8 ft | |
| 10m (28.4 MHz) | 16.5 ft | 4.3 ft | Compact enough for a small rotatable boom |
W8JK Array Calculator
Materials for a complete W8JK array
Building the W8JK Bidirectional Array
This guide builds a 0.125λ-spaced array fed with 450-ohm ladder line to a balanced tuner. Adjust the spacing input above if you want a different tradeoff between gain and boom length.
Cut Both Elements
Cut two identical half-wave elements to the length from the dimensions table, each about 3% long to allow trimming to final resonance.
Terminate Element Ends
Attach an end insulator to both ends of each element, with support rope running to your spreader frame or boom-end supports.
Build the Phasing Line
Cut a section of ladder line or open-wire line equal to the element spacing from the calculator, connecting the center of element 1 to the center of element 2.
Establish the Feedpoint Tap
At the center of the phasing line, install a spreader block and connect your main feedline (open-wire or ladder line running to the shack) to this point.
Mount on the Spreader Frame or Boom
Assemble both elements and the phasing line onto your support structure, maintaining the correct spacing precisely — the array's performance depends on accurate, consistent spacing along its whole length.
Raise and Route the Feedline
Raise the completed array to at least 0.5λ above ground for a useful low-angle pattern, and route the balanced feedline down to the shack, keeping it away from metal structures.
Tune and Verify
Connect the feedline to your balanced antenna tuner and find a low-SWR match at your target frequency. Because this is a fed-with-a-tuner design rather than a resonant 50-ohm match, "tuning" here means finding the tuner setting that presents an acceptable SWR to the radio, not trimming for a direct resonance.
| Symptom | Most likely cause | Diagnosis | Fix |
|---|---|---|---|
| No gain over a plain dipole, pattern seems omnidirectional | Phasing line not actually crossed, elements fed in phase | Check the phasing-line connections at both element centers | Re-wire the transposition — one crossing between the two elements |
| Tuner can't find a low SWR anywhere | Feedline or phasing-line connection fault | Check continuity through the phasing line and feedline | Re-solder connections; verify no shorts at the spreader blocks |
| Pattern nulls are in the wrong place relative to expectations | Uneven element spacing along the array | Measure spacing at multiple points along the boom/frame | Re-true the spreader frame; the spacing must be consistent, not just correct at one point |
| Gain seems underwhelming even with correct wiring | Insufficient height above ground, or spacing too tight for the target gain | Check installed height in wavelengths; compare spacing to the table | Raise the array; consider widening spacing toward 0.25λ for more gain at the cost of boom length |
| SWR shifts significantly with weather/humidity | Open-wire line spacing affected by ice or wet conditions | Check line spacing and insulation after weather events | Use commercial ladder line with a stable dielectric, or accept re-tuning after major weather changes |
Why feed the elements out of phase instead of in phase?
At the close spacings a W8JK uses, out-of-phase feed produces noticeably more gain than in-phase feed would at the same spacing — this was Kraus's key finding and the entire reason the design works the way it does.
Can I rotate this antenna like a Yagi?
Yes — many builders mount the W8JK on a rotatable boom exactly like a Yagi, taking advantage of its bidirectional gain in whatever two directions the boom is pointed.
How does spacing affect performance?
Gain generally increases as spacing increases from very tight (0.05λ) up to about 0.5λ, after which additional lobes start to appear and the simple bidirectional pattern breaks down. Most practical builds use something in the 0.1-0.25λ range as a good compromise between gain and physical size.
Do I need a balun with this antenna?
Since the W8JK is fed with balanced open-wire or ladder line into a balanced tuner, a balun isn't strictly required in that feed path — but if you feed it with coax and an unun into an unbalanced tuner instead, ensure the coax has good common-mode isolation to keep the balanced antenna from imposing common-mode current on the shield.
How does this compare to a Yagi of similar gain?
A 2-element W8JK typically gives somewhat less gain than a comparable 2 or 3-element Yagi, but requires far less boom length for its spacing and no gamma match or driven-element isolation — a genuinely different set of tradeoffs, not a strict upgrade or downgrade.
Can I build a multi-band W8JK?
Some builders do, feeding it with open-wire line and a tuner across several bands, accepting that the spacing (optimized for one band) will behave somewhat differently at other frequencies. A single-band design cut and spaced for one specific band will always give more predictable results.