Phased and Parasitic Arrays Need Measured Currents and Patterns

Including the primer’s swapped phase values and undisclosed feed-network losses.

Phased and parasitic arrays are not controversial ideas. Both are established ways to produce useful directivity. A two-element array can place more field in one direction and less in another. A parasitic reflector can work without its own feed cable. A correctly designed phase network can reverse a beam without moving an antenna.

But none of those facts makes a cable length, element label or NEC polar plot a measurement of a physical array.

The radiation pattern is set by the complex current that actually flows in every element: its magnitude, phase and distribution. Those currents are affected by mutual coupling, the element impedances in the assembled array, the ground and radial systems, the feed cables, matching components, switching paths, chokes and nearby conductors. The forward gain available at the array input is then reduced by real feed-network and antenna losses.

Greg Mihran’s July 2026 antenna primer presents precise gains, front-to-back ratios and radiation sweeps for PERformer parasitic and phased arrays. It shows model plots and input-match traces. It does not disclose measured complex element currents, calibrated feed-network loss or comparative azimuth-pattern measurements for those arrays. That is the missing evidence.

What the primer presents

The array section begins with a two-element PERformer parasitic array. One element is driven. A second, nominally two percent longer element is used as a reflector at quarter-wave spacing. Slides 60, 65, 66 and 68 present low modelled SWR, roughly 1.9 to 4.1 dBi modelled gain across six bands and front-to-back ratios around 7 to 8 dB. The plotted “radiation sweep” is a computed result.

Slides 70–72 add a shorter director to form a three-element parasitic array. Slides 74–79 then describe two identical driven elements joined through a T or switching arrangement, with a feedline cable, a phase cable and an optional beta match. Modelled figure-eight and cardioid patterns are assigned gains and front-to-side or front-to-back ratios. Slide 149 advertises custom electrical-length cables for the phased array.

Later sections repeat the same evidential pattern for Challenger, Dominator, Marauder, Thruster and Hammer arrays: exact model gain, efficiency and rejection figures alongside photographs or SWR traces. The physical builds and impedance sweeps may be real. They do not turn the modelled pattern quantities into measurements.

Primer location	What is shown	What remains unestablished
Slides 60, 65–68	Two-element PERformer parasitic-array geometry, low SWR and computed gain, sweep and rejection values.	The induced reflector current, a measured azimuth pattern and realised gain at a defined array input.
Slides 70–72	A three-element model with a shorter director and longer reflector.	Measured parasitic currents and a measured pattern. The caption on slide 70 also calls the −6% element a reflector while the diagram correctly calls it a director.
Slide 74	A phased-array connection diagram, cable-length equation and a normalised phase/spacing definition.	Element driving-point impedances, terminal-current ratio, reference-plane calibration and network loss.
Slides 75–77	Model SWR, horizontal patterns and “radiation sweeps” for figure-eight and cardioid modes.	Measured current phasors and comparative field-strength patterns.
Slide 78	Two array cases with exact spacing, phase and gain/rejection values.	The displayed normalised phase percentages contradict the formula on slide 74 and are swapped between the panels.
Slide 79	Six-band tables and average “gain” statements.	Measured system gain. The slide also uses gain for both absolute values and improvement over one PERformer, two different quantities.
Slides 121, 127–129	Later parasitic designs with model gain, efficiency and front-to-back values, plus field SWR traces.	A measured pattern and a complete loss budget for each array.
Slide 149	Custom electrical-length cables offered for the phased PERformer array.	Measured insertion loss, electrical length versus frequency, amplitude balance, phase tolerance and power/temperature stability.

The fair conclusion is not that the arrays cannot work. It is that the numerical performance claimed for the physical arrays has not been demonstrated by the evidence presented.

The values on slide 78 fail the primer’s own arithmetic

This is not a disagreement over NEC settings or antenna theory. It can be checked directly from the two slides.

Slide 74 defines a normalised phase/spacing percentage as:

Slide 78’s left, bidirectional case shows:

• Δ = 0.500 λ
• Φ = 0° = 0.000 λ
• displayed Ψ = 71%

But the stated equation gives 0.000 / 0.500 × 100% = 0%.

The right, cardioid case shows:

• Δ = 0.203 λ
• Φ = 52° = 0.144 λ
• displayed Ψ = 0%

Here the equation gives 0.144 / 0.203 × 100% = 70.9%, or 71% when rounded.

The two displayed Ψ values are therefore on the opposite panels. The underlying 0° and 52° labels are not what this arithmetic proves to be swapped; it is specifically the 0% and 71% normalised values.

That correction does not by itself establish whether 52° is the right excitation for the claimed cardioid. That requires a defined sign convention, identification of which element is delayed, the coupled element impedances and the resulting terminal currents. It does show that the published summary was not checked against its own formula.

There is another unresolved audit-trail warning on slide 65. The vertical-plane output is labelled as a parasitic-array model, while the filename visible on the horizontal-plane output includes PERFORMERArrayPhased. This may be nothing more than a copied filename. It may also be a plot from a different run. Without the model package and a figure-to-file revision map, the reader cannot determine which.

A phase-delay cable is not an element-current phasor

A length of coax has a propagation phase. If its electrical length is 0.144 wavelengths, the corresponding phase travel is about 51.8 degrees. That is useful design information. It is not automatically the phase difference between the currents flowing into two coupled antennas.

For a two-element driven array, the terminal relationship is:

Z11 and Z22 are the elements’ self terms in the assembled system. Z12 and Z21 describe mutual coupling. The element currents are obtained by solving the coupled system. They are not obtained by copying the source voltages or cable phase labels into a pattern formula.

This is why an element that measured 50 Ω alone may present a very different driving-point impedance when its neighbour is excited. The neighbour’s current induces voltage back into it. The transmission line then transforms that coupled load, while line attenuation and mismatch change the magnitude and phase at the element terminal.

The far field can be written schematically as:

The QEX report on G3NPC’s 21 MHz four-square demonstrates the correct chain of reasoning. Swanson measured isolated and mutual impedances, independently characterised cable impedance, length, velocity factor and attenuation, calculated the feeder system around the coupled loads, and then measured the actual element-current magnitudes and phases. Even with deliberate design, the measured currents did not perfectly equal the intended set.

Phase must have a sign, reference and direction

“52 degrees” is incomplete by itself. A reproducible phased-array statement must define:

• the time convention, such as ejωt or e−jωt
• which element is the reference
• whether positive phase means lead or lag
• which physical path contains the delay
• the reference plane at which phase is specified
• the current-reference direction at each element terminal
• whether the value describes source voltage, line propagation or element current

Change the sign convention and the number changes sign without changing the antenna. Move the delay to the other element and the preferred direction should reverse. Reverse one current-probe reference and its displayed phase shifts by 180 degrees. These are ordinary bookkeeping issues, but without the bookkeeping the claim cannot be independently reproduced.

A fixed physical cable also produces a phase delay that changes with frequency. A six-band array therefore needs the actual cable implementation and switching arrangement for each band, or measured network data showing how the required relationship is achieved across the stated frequencies.

Parasitic elements still require current evidence

A parasitic array avoids a driven-element phase network at the reflector or director. It does not avoid the current problem.

For a simplified two-element system with element 2 passive, its applied terminal voltage is zero:

The NBS/NIST Technical Note 1082 develops this same driven/reflected-current relationship and obtains the total field by combining the driven and reflector fields. The passive element current depends on mutual impedance and on that element’s own complex impedance. Its length is only one influence on those terms.

A slightly longer element often behaves as a reflector and a slightly shorter one as a director in a conventional Yagi-like arrangement. That is a sound design heuristic. It is not a substitute for reporting the induced-current ratio or measuring the resulting pattern.

For the PERformer concept, the current can also be affected by:

• radial length, droop, height and current balance
• soil conductivity and permittivity beneath each element
• the exact open, shorted or matched condition at the parasitic feedpoint
• contact and conductor loss at that termination
• mast, support, coax and common-mode current
• small length and spacing errors
• nearby vehicles, fences, trees, wiring and operators

A model may predict all of this well if the relevant geometry and losses are included. The physical build still needs a measurement if the model result is presented as installed performance.

Deep nulls are the most fragile part of the claim

Forward gain and front-to-back ratio are not equally sensitive to error. A deep modelled null is produced by near-cancellation. If two contributions intended to cancel have a small amplitude or phase error, the residual may be much larger than the ideal residual. The forward lobe may change only modestly while a 22 dB null becomes noticeably shallower or shifts in azimuth.

That makes a high front-to-back figure especially dependent on:

• current amplitude balance
• current phase accuracy
• element symmetry
• feed-path equality
• local ground and environmental symmetry
• the azimuth and elevation at which “back” is evaluated

A normalised NEC cut can show where the ideal null should occur. It cannot show that the built array achieved the current cancellation. A single signal report in the favoured direction cannot do that either. The array needs a pattern sweep or a set of controlled measurements around it.

The feed network has a power budget

Slide 74’s diagram contains more than an abstract phase angle. It shows a T connector or switching box, feedline cable, phase cable and optional beta match. A practical system also contains connectors, chokes and terminations. Each item may introduce attenuation, mismatch, phase error or common-mode current.

An ideal equal power division is not itself a 3 dB dissipative loss. Half the power goes to each of two outputs, but the total remains available to the array. Real dissipation is the extra power converted to heat in the cables, transformer, splitter, relays, connectors and matching components. Mismatch can also prevent part of the available power from being accepted.

A useful system accounting, with all terms referred to a declared input plane, is:

A NEC result excited directly at ideal element ports may be a legitimate antenna-model benchmark. A network model using ideal, non-radiating and lossless lines may also be useful. Neither includes real network loss unless that loss has been entered. The primer does not provide enough source data to determine which network effects were included in the displayed array results.

Swanson’s four-square is instructive again. He measured about 1.2 dB attenuation in each matching transformer and about 2 dB from the power-splitter input to the element inputs for the complete feed path. That particular feed system was therefore about 63% efficient. Combined with the estimated element efficiency, the reported complete-array radiation efficiency at the feedpoint was about 42%. These numbers do not predict the PERformer array’s loss. They prove why feed-network loss must be measured rather than omitted from the accounting.

Reference plane first, number second

“Gain” is incomplete until the input reference is stated. Possible reference planes include:

• the transmitter output
• the main array input before the splitter or switch
• the network outputs
• each element terminal
• the sum of power accepted by the individual elements

Directivity is a pattern-shape quantity. Power gain includes antenna dissipation. Realised gain also includes mismatch at the defined input. Complete-system gain from a main coax connector must include the network between that connector and the elements.

Slide 79 makes this distinction harder to follow by placing absolute model values in its tables and then describing an average “gain” of 4.3 dB or 3.9 dB over a single PERformer. Improvement over a reference antenna and absolute gain in dBi are not the same number. Both can be reported, but each needs its own label, reference plane and power normalisation.

Low SWR proves neither phasing nor directionality

An analyser can measure the complex reflection coefficient at the calibration plane. That establishes an input-impedance fact. It does not reveal how current divides between elements, whether the intended current phase exists or where the array radiates.

The same caution applies to model SWR. The traces on slide 75 are headed SWR [50 ohm] [Src 1]. They show the impedance seen at a model source. They do not, by themselves, demonstrate the match at the common input of a physical T, switch, phase line and matching network.

Several very different arrays 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 therefore the wrong pattern
• an imbalanced system in which the feedline radiates
• a system whose network loss broadens the match
• a correctly matched splitter feeding unequal or mutually coupled loads

Low SWR is desirable. It is evidence of low reflection at one port. It is not a beam-direction instrument, a current-phase meter or a gain measurement.

What current measurements should be published?

For a phased array, report the complex current at every element terminal across every claimed operating band. A useful table includes:

• frequency and mode or beam direction
• current magnitude normalised to a named reference element
• current phase relative to that same element
• the sign and probe-current convention
• measured driving-point impedance of each coupled element
• accepted power at the common input and at each element
• uncertainty, repeatability and calibration method

For a parasitic array, measure or otherwise validate the induced current on each reflector or director relative to the driven element. If direct current measurement is impractical, disclose a model that predicts it and validate the consequence with a measured pattern. A current plot from an undisclosed model is not as strong as a calibrated current measurement, but it at least exposes the mechanism being claimed.

Also measure common-mode current on the outside of the feedlines. If it is material, the feedline has become another array element. That may change the match and pattern while making the nominal two-element explanation incomplete.

What a credible pattern test looks like

A useful comparative field test does not have to be an anechoic-chamber campaign, but it does need disciplined controls:

1. Define the frequency, array geometry, soil and surrounding environment.
2. Place the receiver far enough away for the intended comparison and document distance, height, polarisation and angular geometry.
3. Keep transmitter power stable and correct every reading to equal accepted power at the declared array input.
4. Measure enough azimuths to resolve the forward lobe, sides and rear null; do not sample only “front” and “back”.
5. Use A–B–A or repeated beam switching to reveal drift and propagation changes.
6. Record the noise floor and system dynamic range so a deep null is not merely a receiver limit.
7. Repeat the sweep and publish raw readings, normalisation, uncertainty and weather/site notes.
8. Use a calibrated reference antenna if absolute gain is claimed.

Swanson established stations every 22.5 degrees around a 38-metre radius, used a field-strength meter with 90 dB dynamic range, measured all four switchable beam headings and obtained repeatable patterns. The measured currents were then put back into the model, producing good pattern agreement. Crucially, the estimated 7–8 dBi forward gain was still identified as needing confirmation against a standard antenna. Pattern shape and absolute gain were not conflated.

A reproducibility package for an array claim

Claimed quantity	Minimum evidence
Cable delay	Physical length, velocity factor, frequency, measured electrical length and stated reference planes.
Element-current phase	Complex current measurements at the coupled element terminals, with sign and current-reference conventions.
Parasitic reflector/director action	Induced-current ratio or a fully disclosed model, plus a comparative measured pattern.
Feed loss	Calibrated insertion-loss or power measurements for the complete path, including cables, matching, splitter and switch state.
Front-to-back or front-to-side ratio	Measured field pattern at stated frequency, elevation geometry, polarisation, accepted power and dynamic range.
Absolute gain	Comparison with a calibrated reference antenna or a traceable gain-measurement method.
NEC prediction	Source files, solver/version, ground and material losses, feed model, currents, power budget and convergence study.
Multi-band performance	The complete preceding package at each band, including the band-specific feed configuration.

The ARRL Antenna Book support material places phased-array feed-system design in its multielement-array resources and supplies dedicated transmission-line design software and phased-array model files. That is another reminder that a feed system is part of the coupled design, not an afterthought represented by one standalone phase number.

What can fairly be concluded from the primer?

The presented models suggest that the proposed geometries are worth testing. The shown SWR traces suggest that particular builds could be matched over the displayed frequencies. The photographs show practical construction and operation. None of that is worthless.

But the following stronger statements are not established by the primer:

• that the phased build achieved the model’s element-current magnitude and phase
• that its complete feed network preserved the stated gain after loss
• that the physical array produced the displayed front-to-back or front-to-side ratio
• that a parasitic element’s nominal length produced the modelled induced current over every band
• that low SWR validates any of those pattern quantities

Those are measurements still to be made, not facts created by additional decimal places.

Takeaways you can trust

• Phased and parasitic arrays can both produce useful directionality.
• The array pattern is governed by actual complex element currents.
• Mutual coupling means cable phase and source phase are not automatically element-current phase.
• A parasitic element’s length influences its induced current but does not measure it.
• The 0% and 71% normalised values on primer slide 78 are swapped under the primer’s own equation.
• An ideal 3 dB split is not heat loss; real splitter, switch, transformer, cable and matching loss must be measured.
• Low SWR validates input match, not gain, rejection, current balance or beam direction.
• Deep nulls are particularly sensitive to amplitude and phase errors.
• A measured azimuth pattern tests directionality; a calibrated reference is needed for absolute gain.
• A model becomes persuasive when its inputs are disclosed and its predictions survive measurement.

In Summary

The primer’s array models may represent useful design hypotheses. The problem is the leap from nominal geometry and cable arithmetic to precise claims about a physical antenna.

Slide 78 contains a checkable internal error: by slide 74’s equation, the bidirectional panel should show 0%, while the 0.203-wavelength, 52-degree panel should show approximately 71%. More fundamentally, neither value is a measurement of terminal-current phase.

For phased arrays, publish the complex current at every driven element, the complete feed-network loss and a comparative field pattern. For parasitic arrays, establish the induced current and measure the pattern. Define the reference plane before quoting gain. Separate model output, impedance measurement and field measurement.

That does not make antenna experimentation less enjoyable. It makes a successful experiment scientifically useful.

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

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