PERformer Arrays: Cable Phase Is Not Element-Current Phase

Greg Mihran, KJ6ER, presents an appealing proposition in the July 2026 edition of KJ6ER Antennas Primer 1: combine portable PERformer verticals and obtain useful gain and switchable directionality. I have no quarrel with that principle. Phased arrays work. Parasitic reflectors and directors work. My objection is to treating the model's precise gain and rejection as established performance of the physical installation.

The distinction matters to the person buying cables, putting up another mast or deciding which antenna to take into the field. A successful match does not tell that person whether the extra hardware produces the promised beam. In the array sections of this edition, the connection from model to measured element currents, complete feed loss and measured angular pattern is missing.

There is also a smaller, completely checkable problem: two percentages on slide 78 are exchanged relative to the equation on slide 74. We can correct that arithmetic without pretending it proves that the antennas do not work. The larger question is what current the real feed system delivers.

What the PERformer Comparison Actually Shows

The primer's parasitic arrangement uses a driven vertical and a reflector nominally 2% longer, separated by a quarter wavelength. Its phased arrangement instead drives both elements through a combined feed system. These are different ways of creating the current relationship that shapes a beam. Neither can be reduced to the label on an element or one length of coax.

Mihran correctly identifies mutual coupling as the mechanism in the parasitic explanation. That is exactly why the induced current, not simply the extra element's nominal length, is the quantity that must support the result. The construction photographs are useful too. My criticism is not that a model or a field build is worthless; it is that neither is a calibrated measurement of the other.

The Swapped Percentages Are an Arithmetic Error, Not an Antenna Verdict

Slide 74 defines its normalised phase-to-spacing quantity as Ψ = (Φ / Δ) × 100%, with phase-line electrical length Φ and element spacing Δ both expressed in wavelengths. Applying that definition to slide 78 gives:

The 0% and 71% values belong on the opposite panels. This calculation does not establish that the 0° and 52° phase labels are exchanged, that a particular beam must point the other way, or that the NEC run is invalid. It identifies the printed normalisation error and nothing more. Correcting those percentages still leaves us needing the relationship between the cable and the element currents.

Two other presentation details deserve clarification rather than speculation. Slide 70's lower caption calls the shorter −6% element a reflector, although the drawing labels it as a director. Slide 65's horizontal plot has a phased-array filename under a parasitic-array heading. A caption or filename can be wrong without the electromagnetic result being wrong. The drawing's roles should be labelled consistently, and the model files should identify which run produced each plot.

Cable Phase Is Not Element-Current Phase

The 0.144-wavelength phase cable in Mihran's example corresponds to about 51.8° of one-way propagation phase. That is useful design information. It is not automatically the phase between currents entering two mutually coupled antennas. Copying that cable number into an element-current label skips the coupled load that the cable actually feeds.

For a two-element driven system:

Z₁₁ and Z₂₂ are the self terms in the assembled array; Z₁₂ and Z₂₁ describe mutual coupling. The terminal currents result from solving the coupled network with the real feedlines, matching devices and terminations. An element that measures 50 Ω alone can present a very different driving-point impedance when its neighbor is excited.

Line attenuation, mismatch and multiple reflections further alter terminal magnitude and phase. Slide 79 spans 40 m through 6 m in seven band rows. That need not mean one unchanged cable is intended for all bands, but it does mean each band needs its actual cable arrangement and coupled-load solution. Slide 149's offer of custom electrical-length cables makes cable implementation a practical part of the proposition, not merely a symbol in a model.

Garth Swanson, G3NPC, provides a useful contrast in his 21 MHz four-square report in QEX: design with measured coupling and line parameters, then disclose the achieved current phasors. They were not perfectly ideal. Publishing that difference makes a design reproducible.

Terminal Currents Are Necessary, but Not the Whole Field

For identical elements in simple array-factor teaching, the field is often written schematically as a weighted sum. In a real mutually coupled array, a more careful form uses each element’s embedded pattern:

The complex coefficient a_n represents the achieved excitation under the chosen normalization, while F_n,embedded includes geometry, environment and mutual coupling. For wire antennas, the current distribution along each conductor matters—not only one terminal-current sample.

Measured terminal phasors are therefore powerful validation evidence, but they do not replace accurate element geometry, ground, support and feedline modelling. This matters for PERformer installations with elevated radials and a real support/feedline system: an identical current reading at one terminal does not guarantee the same current distribution everywhere or the same embedded field pattern.

Phase Needs a Convention

“52 degrees” is incomplete unless the documentation states:

• the e^jωt or e^−jωt time convention;
• the reference element and whether positive means lead or lag;
• which physical path contains the delay;
• the voltage/current reference planes;
• the current-reference direction at each terminal; and
• whether the value describes source voltage, line propagation or element current.

Changing the time convention reverses a phase sign without changing the antenna. In a symmetric two-element system, transferring the corresponding excitation delay to the other element reverses the preferred direction; an asymmetric installation needs its full network solution. Reversing a probe reference adds 180°. For the primer's 52° example, the reader needs both the convention and the achieved current phasors, not just an unsigned cable label.

Parasitic Elements Still Need Current Evidence

The parasitic PERformer saves a separate drive branch, but it does not escape the current question. For a passive second element with zero applied terminal voltage, and its termination included in Z₂₂:

NBS/NIST Technical Note 1082 develops this driven/reflected-current relationship and combines the fields to obtain the array response. The passive current depends on mutual impedance and the passive element’s complex self impedance. Length is only one influence.

If a separate load Z_L terminates the parasitic port, the denominator becomes Z₂₂ + Z_L. Open, shorted and loaded are not interchangeable states. That is why the PERformer reflector's termination and radial connection belong in the documented geometry.

A slightly longer element often acts as a reflector and a shorter one as a director in a conventional Yagi-like geometry. Mihran's +2%/−6% dimensions are sensible starting variables, not a measurement of the resulting current. Radial geometry, soil, termination, mast, support, coax exterior and nearby conductors also enter the answer. The useful advantage of a parasitic arrangement is simpler feeding; the compromise is that its excitation follows the coupled geometry and load instead of being independently driven.

Deep Nulls Are the Fragile Part

The phased summary's 22 dB average front-to-back figure deserves particular care. A deep rear null depends on near-cancellation: a small amplitude or phase error can leave much more residual field behind the array while the forward lobe changes only modestly. This is why a few good reports in front of a PERformer array cannot establish its claimed rejection behind it.

Front-to-back and front-to-side figures therefore need:

• current magnitude and phase accuracy;
• element, radial and feed-path symmetry;
• controlled common-mode current;
• defined azimuth, elevation and polarization;
• adequate receiver dynamic range and noise-floor margin; and
• a measured angular sweep rather than only “front” and “back.”

A normalized NEC cut can predict where an ideal null should occur. It cannot establish that the built array achieved the cancellation.

The Feed Network Has a Power and Phase Budget

Slide 74 already draws the components that make this a system question: the T/switch arrangement, feed lines, phase cable and optional beta match. A practical build adds connectors and chokes. Their loss and phase behaviour sit between the main input and the elements. A model excited directly at ideal element ports does not automatically include that complete path.

When the factors are separable and have not already been counted, a main-input result can be written:

In decibels, the multiplicative efficiencies become additive loss terms. The active IEEE 145-2025 antenna terminology standard is the current formal reference, but the article must still declare whether its “gain” excludes mismatch, includes the feed network or is merely a relative improvement over another antenna.

Swanson reported about 2 dB feed loss, roughly 63% feed efficiency and about 42% effective array efficiency. These are not PERformer loss figures; the comparison is about disclosing the loss. For otherwise equal systems, an extra 2 dB in the feed consumes 2 dB of a modelled directional advantage.

Reference Plane First, Gain Second

Possible power reference planes include the transmitter output, main array input, divider outputs, individual element terminals and the sum of accepted element powers. They produce different loss budgets.

• Directivity is pattern concentration relative to total radiated power.
• Gain includes antenna radiation loss relative to accepted power under the declared antenna boundary.
• Realized gain also includes input mismatch at the stated port.
• Complete-system realized gain from the main connector must include feed-network loss between that connector and the elements.

Slide 79 illustrates why this distinction is not pedantry. Its tables contain model gain values; the larger averages below them describe improvement over a single PERformer. They should not be read as the same number. Absolute gain in dBi and a comparison in dB both need a labelled reference, stated frequency/direction and consistent input-power boundary. The additional feed network belongs in the array side of an installed comparison.

SWR Does Not Measure a Beam

An analyser measures reflection at its calibration plane. It does not reveal how current divides between elements, whether the required current phase exists, or where the array radiates.

Slide 75's plots identify a 50 Ω model source. They are useful impedance results, but they do not measure the physical common input of the complete switch, phase lines and coupled antennas. Nor do they establish that the current relationship responsible for the model's beam was achieved.

Several systems can present the same 50 Ω input:

• the intended low-loss array with the intended current ratio;
• a low-loss array with the wrong phase and pattern;
• an imbalanced array with radiating feedlines;
• a lossy network that broadens the match; or
• a matched divider feeding unequal coupled loads.

Low SWR is useful operational evidence. It is not a current-phase meter, direction indicator or gain measurement.

What Current Measurements Should Show

For each band, mode and beam direction, report:

• complex terminal current normalized to one named element;
• phase sign and probe-current direction;
• coupled driving-point impedance at each element;
• accepted power at the main input and element ports;
• complete feed-network loss and switch state;
• coax-exterior common-mode current;
• probe loading, calibration, de-embedding and uncertainty; and
• repeatability after reconnecting or switching the network.

For a parasitic array, measure or model the induced current relative to the driven element and validate its consequence with a pattern. If the coax exterior carries material current, it has become another radiating array conductor and must be included.

What a Credible Pattern Test Looks Like

1. Document frequency, geometry, soil, surroundings, polarization and all conductor heights.
2. Demonstrate adequate range distance or quantify the remaining phase-curvature error.
3. Normalize every reading to equal accepted power at the declared main input.
4. Measure enough azimuths to resolve lobes, sides and nulls.
5. Use repeated A–B–A or rapid beam switching to expose drift and propagation changes.
6. Record receiver dynamic range and noise floor so a “deep null” is not the instrument floor.
7. Publish raw readings, corrections, uncertainty and site notes.
8. Use a calibrated reference antenna or traceable range method for absolute gain.

Swanson sampled a 38 m-radius circle at 22.5° intervals, compared four beam settings and modelled the measured currents. He still identified forward gain as unmeasured. That is the useful example for PERformer: a measured pattern can validate directionality without establishing absolute gain.

Reproducibility Checklist

The ARRL Antenna Book support material includes multielement-array resources, transmission-line tools and phased-array model files. That is a useful reminder that the feed system and coupled antenna are one design—not an ideal pattern plus an afterthought cable.

What I Would Take from the PERformer Experiments

• The PERformer geometries are plausible array designs worth experimenting with; the July primer does not establish every printed number as installed performance.
• The 0% and 71% normalised phase/spacing values on slide 78 are swapped under slide 74's equation.
• The installed pattern depends on actual current distributions and embedded element patterns.
• Mutual coupling means cable phase is not automatically terminal-current phase.
• A parasitic element’s nominal length influences current but does not measure it.
• An ideal equal split is not a 3 dB heat loss; real excess loss must be measured.
• Deep nulls are especially sensitive to small phase, amplitude and symmetry errors.
• Low SWR validates input match—not current balance, beam direction, rejection or gain.
• A measured pattern validates directionality; a calibrated reference is needed for absolute gain.

The Beam Has to Survive the Feed System

My conclusion is not “PERformer arrays cannot work.” It is that the July 2026 primer supplies promising design cases, but not the current, loss and field evidence needed to treat their precise modelled performance as a demonstrated installation advantage. The exchanged percentages are a definite error; unknown feed loss is an unknown, not proof of a bad antenna.

I favour designing the feed system around the coupled loads and the current relationship the wanted pattern requires. That has a concrete engineering advantage over choosing a nominal cable delay first: it accounts for the voltage each neighbouring element induces and lets matching, phase and power delivery be solved together. A simpler parasitic array can be the more practical choice when its geometry delivers sufficient rejection without that extra feed network. Neither choice gets a free gain certificate from low SWR.

For the PERformer proposals, the convincing next result is therefore not another decimal place on the same polar plot. It is the documented current ratio and loss at each claimed configuration, connected to a repeatable field pattern. That would turn an attractive array idea into evidence a builder can use.

Primary Sources and Further Study

• Greg Mihran, KJ6ER — Antennas Primer 1, July 2026: the specific edition discussed; especially slides 60–79 and 149. Slide references here do not automatically apply to later editions.
• Garth Swanson, G3NPC — A 21 MHz Four Square Beam Antenna: coupled feed design, measured currents, loss and pattern.
• NBS Technical Note 1082: section 2.1 develops the driven/reflector current relationship.
• IEEE 145-2025 and IEEE 149-2021: antenna terminology and measurement practice.

Questions, antenna-factor records or height trials to share? Contact RF.Guru.