What Rudy Severns Actually Proved About Elevated Radials

Four controlled radials versus 64—and why the same research led Severns to recommend 10–12 for reliable real-world performance.

There is a genuine and important result in Rudy Severns’ elevated-radial work: on his 40-metre test range, four carefully arranged radials elevated 48 inches produced essentially the same measured signal as 64 quarter-wave radials lying on the ground.

Greg Mihran, KJ6ER, reproduces that comparison on slide 19 of his July 2026 antenna primer. So far, so good.

But the primer then crosses a line that Severns’ experiment does not cross. Slides 17, 24 and 25 claim that two elevated radials have approximately 4 Ω loss, deliver 90% radiation efficiency, gain 4.8 dB over two surface radials and outperform the number of surface radials that would otherwise be required for comparable efficiency.

Four has quietly become two. A controlled relative field-strength result has become an absolute efficiency figure. One installation has become a universal rule.

None of those conversions follows from Severns’ experiment.

The experiment everyone quotes

The detailed source is Severns’ 2009 Ground Systems, Part 3. It was not a casual comparison between two convenient antennas. Severns constructed a test in which he could change the radial system while keeping the rest of the installation as controlled as practical.

The reference level was the signal from four quarter-wave radials on the surface. Relative to that reference, 64 surface radials measured approximately +5.8 dB. Four radials elevated 48 inches measured approximately +5.9 dB.

The difference was 0.1 dB—too small to be practically meaningful in this experiment. That is the basis for the often-repeated statement that four elevated radials were equivalent to 64 surface radials.

The measurement was relative transmission magnitude, |S21|. It did not directly measure the absolute radiation efficiency of either ground system. Matching an efficient reference and matching a mediocre reference would both produce a small relative difference, so the comparison cannot be converted into an absolute 90% figure without additional evidence.

It is a strong result. It is also a bounded result.

What “equivalent” did—and did not—mean

The two ground systems produced nearly the same received signal in the controlled measurement path. They were not electrically identical.

Severns measured the following feedpoint impedances:

The impedances were plainly different even though the relative signals were almost the same. That alone should stop us from treating a feedpoint-resistance number as a direct efficiency meter.

Nor does a single receive path prove that the complete three-dimensional patterns were identical. The appropriate conclusion is narrower: under the controlled excitation, measured path and conditions of this test site, four well-balanced elevated radials performed about as well as the 64-radial surface reference.

In experimental engineering, the conditions are part of the result. Remove them and the result becomes a slogan.

How the primer changes the claim

Slide 19 of the primer is a reasonable summary of Severns’ measured four-versus-64 comparison. The problem appears when that result is placed beside very different claims elsewhere in the primer.

The numbers may look related because they all concern radials. Scientifically, they are different propositions requiring different evidence.

Why Severns revisited his own result

Severns did something good engineers do: he continued testing the limits of a successful result.

In his 2012 elevated-ground-system investigation, Part 1, he warned that the earlier four-radial result must not be applied uncritically. Four radials can work extremely well under favourable conditions, but a sparse elevated system is susceptible to modest asymmetry, nearby conductors and variations in the soil below it.

His blunt practical message was that we cannot simply install any four wires and assume we have recreated the controlled experiment.

The reason is fundamental: an elevated quarter-wave vertical with tuned radials is a coupled resonant structure. The vertical, every radial, the earth, the feedline and nearby conductors can participate in the current distribution. A small change in one element can move it closer to or farther from resonance and redistribute current through the whole system.

This is why Severns began strongly recommending at least 10–12 elevated radials.

That recommendation does not contradict the four-radial result. It explains the difference between best-case efficiency and repeatable field performance.

1. Current imbalance: the central weakness of sparse systems

In a perfectly symmetric four-radial system, each radial should carry roughly one quarter of the total radial current. Severns’ 2009 currents were close to that ideal.

Real installations are rarely so kind. One radial may run over wetter soil. Another may pass a fence, vehicle, metal mast or buried service. The wires may differ slightly in length or height. Their end surroundings may not be equivalent. Because the radials are resonant conductors, a small electrical difference can create a large current difference.

Severns’ 2012 Part 2 presents field examples in which two nominally similar radials divided current 1.00:0.80 at one frequency but 1.00:0.10 only 80 kHz higher. Other two-radial examples showed even larger differences.

That is not a small bookkeeping error. If one radial carries most of the return current, the antenna is no longer the symmetric structure assumed by the simple model.

Consequences can include:

• greater electric-field penetration into lossy soil
• a shifted feedpoint impedance and resonant frequency
• azimuth-pattern distortion or tilt
• greater voltage on one part of the radial system
• more incentive for current to find another return path

With 10–12 radials, one imperfect radial is a smaller fraction of the complete system. The system averages local differences instead of allowing one or two conductors to dominate.

2. Soil asymmetry: elevation does not remove the earth

Elevating tuned radials can greatly reduce near-field loss compared with placing a few wires directly on or in the soil. It does not make the earth disappear.

The elevated conductors remain electromagnetically coupled to the ground beneath them. Soil conductivity, dielectric constant, moisture and uneven terrain affect their capacitance, resonance and loss. If those properties differ from one side of the antenna to the other, the radial currents can differ too.

This is also why slides 18, 23 and 25 of the primer are too simplistic when they treat the radial system and soil as parallel capacitor plates and state that loss falls reciprocally with height through C = ε0A/d.

A radial field is not the uniform field between two large parallel plates. The wires are thin, their fields fringe, the soil is lossy, radial currents are non-uniform, and the vertical and feedline are coupled into the same structure. The parallel-plate equation may suggest a trend; it cannot calculate ground-loss resistance or prove an inverse law for the complete antenna.

3. Nearby objects: the missing conductors in the diagram

A model of a vertical and two perfect wires in an empty world can be beautifully symmetric. A portable or garden installation is not.

Metal supports, guy hardware, coax, control cables, fences, gutters, vehicles, masts, buried conductors and even another antenna can couple unequally to the radial system. The fewer radials there are, the more influence each unmodelled conductor can have on current division and pattern.

Severns therefore advised keeping the antenna and radials clear of conductive objects and modelling special situations rather than assuming the ideal case survives every installation.

More radials do not magically abolish coupling. They make the intended return structure less dependent on one or two sensitive conductors.

4. Feedline coupling: the coax can become an extra radial

The outside of a coax shield is available to carry common-mode current. If the intended radial currents do not sum to the return current demanded at the feedpoint, current will seek another path. The feedline, mast, control cable or station earth can provide it.

Then the measured system is no longer merely a vertical plus two or four radials. The feedline is part of the antenna.

This can change the apparent match, pattern and received signal. It can also make a sparse-radial system appear to work better than the radials alone would allow, because an uncontrolled extra conductor is doing part of the work.

Severns used a common-mode choke in the four-versus-64 experiment. In his later practical recommendations, he called for a base choke or balun presenting more than about 2 kΩ to unwanted shield current. Greg’s slide 20 correctly says to use an RF choke, but that advice does not validate the 4 Ω or 90% figures on the neighbouring slides. The choke performance and resulting shield current still need to be disclosed or measured.

5. Narrow bandwidth and tuning sensitivity

A few elevated quarter-wave radials can form a high-Q system. It may be very efficient at the frequency where every conductor is carefully adjusted, yet change rapidly as frequency, height or surroundings change.

Severns noted that the important advantage of adding elevated radials was not necessarily a dramatic increase in the best-case efficiency. Additional radials reduced current asymmetry and produced a less fragile system with better impedance bandwidth.

This point is often missed. A recommendation for 10–12 radials is not an admission that four cannot be efficient. It is a recommendation to make good performance less dependent on laboratory-like symmetry and exact tuning.

Why two radials deserve more scrutiny, not less

Two opposite quarter-wave radials can certainly form a working antenna. In a favourable, balanced installation they may be quite efficient. Nothing in Severns’ work justifies declaring that such antennas cannot work.

But two is the sparsest possible nominally balanced elevated return system. There is no redundancy and almost no spatial averaging. Any difference between the two wires is immediately a large percentage of the whole radial system. If one radial current falls, the other radial, the soil and the feedline must take up the imbalance.

That makes the following claim especially demanding:

Rloss ≈ 4 Ω → η = 37 / (37 + 4) ≈ 90%

The equation is not a measurement. It only calculates a ratio after both resistance values have been assumed. To turn it into evidence, the author must establish the actual radiated and dissipated powers of the complete system, including soil, conductors, matching network, feedline and common-mode behaviour.

The primer supplies none of the information needed to reproduce the claimed 4 Ω:

• no model file or complete geometry
• no disclosed NEC engine and ground method tied to the claimed result
• no soil conductivity or dielectric constant
• no radial wire size, insulation or loss model
• no individual radial-current values
• no common-mode choke model or measured shield current
• no radiated-power and loss-power integration
• no controlled field comparison with uncertainty

Without those details, “4 Ω” is not an experimentally established property of two elevated radials. It is an undocumented model result presented as a general physical specification.

Pattern change is not automatically efficiency gain

A sparse, imbalanced radial system can skew the radiation pattern. A receiver in a favoured direction may report a stronger signal even if total radiated power has not increased by the same amount.

That is why one-path field strength cannot by itself establish total efficiency. A robust comparison needs either radiated-power integration in a validated model or measurements that sample the pattern sufficiently to distinguish increased total radiation from redistributed radiation.

Likewise, the +4.8 dB figure on slide 24 is not independent evidence. It is simply 10 log(90/30). If the assumed efficiencies are unsupported, the calculated decibel difference is unsupported too.

How to test a two-radial claim properly

A claim this specific is testable. A useful experiment would:

1. Document frequency, radiator, radial lengths and angles, wire, height, soil and nearby conductors.
2. Use a characterised feedpoint choke and measure common-mode current on the coax below it.
3. Measure the magnitude and phase of current in each radial across the operating bandwidth.
4. Compare two, four, 8 and 12 elevated radials while keeping accepted feedpoint power constant.
5. Re-tune each configuration without hiding matching-network loss.
6. Measure several azimuths, or otherwise determine whether a signal change is efficiency or pattern redistribution.
7. Repeat the sequence and publish the uncertainty, not only the best run.
8. Release the complete model so others can reproduce the power budget.

Such a test may show that a particular two-radial system performs very well. That would be useful evidence. It still would not create a universal 4 Ω law for every pair of elevated radials.

What 10–12 radials buys you

This is engineering margin. Four radials can win the controlled demonstration; 10–12 make it more likely that the amateur who copies the design will obtain comparable performance in a different garden, field or band.

Takeaways you can trust

• Severns really did measure four elevated radials at 48 inches performing within about 0.1 dB of 64 surface radials in his 7.2 MHz test.
• The four elevated radial currents were deliberately close to equal, and a common-mode choke was part of the test system.
• The result was specific to the geometry, frequency, soil, current balance and measurement path.
• It was not a measurement of two elevated radials, 4 Ω ground loss, 90% efficiency or equivalence to 120 surface radials.
• A sparse elevated system is a coupled resonant structure, not two lossless return wires isolated from the earth and feedline.
• Current imbalance, uneven soil, nearby conductors and common-mode feedline current can change its impedance, loss and pattern.
• Severns recommended 10–12 radials primarily for robustness, symmetry and repeatability—not because four controlled radials were shown to be poor.
• Two radials can work well. Their efficiency must be demonstrated for the actual installation, not inherited from a different four-radial experiment.

In Summary

The honest reading of Rudy Severns is more interesting than the slogan.

Four elevated radials can be remarkably effective. In a carefully balanced 40-metre system, they matched the measured signal from 64 surface radials. That is the experimental achievement.

The same research programme also showed why that result is easy to misuse. Sparse elevated systems are sensitive to current imbalance, soil asymmetry, nearby conductors, frequency and feedline coupling. Those are not minor construction details. They determine whether the real antenna is the symmetric antenna in the diagram.

Severns therefore moved from showing what four radials can do to recommending 10–12 radials for what real installations can do reliably.

Want more technical RF content? Subscribe for new deep dives and lab notes.

Have a question or field observation? Contact RF.Guru.