Comparing Antennas: Measure the Installation, Not the Label

Joeri’s point is deliberately blunt: antennas are systems. A named radiator is only one part of the result. Height, ground, feed line, matching, common-mode current, nearby structures, polarization, noise and the propagation path all decide what arrives at the other end.

Two Honest Comparisons Answer Different Questions

A component comparison tries to isolate the radiator. It holds geometry, height, surroundings and reference planes as closely as the designs allow, then measures impedance, loss, current distribution, gain and pattern. This is the right approach when the question is what one physical design changes.

An installed-system comparison asks which complete setup works better for a defined operating objective at a real site. The antennas may need different heights, radial systems, matching networks and feed lines. Do not force those variables to be identical; document them and include their losses and current paths in the result.

The Installed Pattern Chooses the Winner

Free-space labels such as vertical, dipole, end-fed or beam do not reveal the installed three-dimensional pattern. Ground conductivity and permittivity, radial or counterpoise geometry, height, bends, loading, nearby buildings, trees, masts and feed-line exterior current can move lobes and nulls.

A vertical may favour a lower elevation region than a low horizontal wire in one installation. That does not give either class a universal DX or regional advantage. The useful elevation and azimuth depend on the path, frequency, ionospheric state, terrain and the station at the other end. A strong lobe is valuable only when it illuminates the required path.

Polarization belongs in the same model. On HF sky-wave paths it can change through propagation, while local ground-wave, near-field coupling and man-made noise can retain stronger orientation effects. A signal difference cannot be assigned to polarization without ruling out pattern, loss, fading and noise changes.

S-Points Are Reports, Not Calibrated Evidence

The IARU Region 1 technical recommendation defines a nominal six-decibel step per S-unit and reference levels for S9. That is a recommendation for meter calibration, not proof that two radios, modes, bandwidths, AGC states or display algorithms agree.

A report of “two S-points” is therefore incomplete without the receiver, settings, signal level, timing and calibration. Even a calibrated signal-power difference does not by itself establish a better communications result, because the noise may have changed by the same amount.

For receive comparisons, record both signal and noise in the same bandwidth and use SNR. Also watch overload, intermodulation and gain compression: an antenna can deliver more signal and still make a receiver less usable if it also raises noise or strong out-of-band signals beyond the available headroom.

Align Power and Reference Planes

On transmit, equal transmitter settings do not guarantee equal power accepted at the two antenna feed systems. Feed-line attenuation, tuner and transformer loss, mismatch and transmitter foldback can differ. Measure power or de-embed the networks to a declared reference plane, and keep waveform and duty cycle the same.

Using the same length and type of coax can be convenient but is not a universal requirement. Different antenna impedances may need different lines and interfaces. The defensible comparison either aligns the reference plane at each antenna input or explicitly includes every line and matching loss as part of the installed-system result.

Measure exterior feed-line current as well. If the feed line, mast, station wiring or earth network joins the radiator, the pattern belongs to that complete current structure. A change in choke placement can then alter the result even when the named antenna has not changed.

Propagation Moves While You Measure

HF paths fade and evolve. A slow manual cable swap can turn time variation into an apparent antenna advantage. Prefer simultaneous receiver channels with known gain and phase behaviour, or a fast, repeatable switch feeding the same receiver. Restore the first antenna after the second—A/B/A—so drift becomes visible.

WSPR and Reverse Beacon Network reports can provide many paths and samples, but they are not automatically controlled antenna ranges. Compare paired reports from the same receiving station and close time interval. Keep transmit frequency, power, timing and modulation fixed, and analyse results by band, azimuth and distance. Different receiver calibrations and reporting populations should not be mixed into one absolute-gain claim.

Repeat on more than one day. Report the distribution—median, spread and sample count—not only the strongest spot. A network result can show a repeatable installed-path advantage; it does not directly produce a laboratory gain, efficiency or pattern value.

A Practical A/B/A Method

1. Write the decision first. Name the band, paths, distance range, direction, mode and whether transmit, receive or both matter.
2. Inventory each installation. Record radiator geometry, height, ground/radials, feed line, tuner, transformer, choke, surroundings and weather.
3. Declare the reference planes. State where impedance, power, current, signal and noise are measured.
4. Stabilize the equipment. Fix receiver bandwidth, AGC or gain, attenuation, preamplifier, frequency and transmitter waveform/duty cycle.
5. Switch quickly and restore. Use A/B/A or simultaneous calibrated channels. Reject a run if the restored A result has moved beyond the planned tolerance.
6. Measure more than signal. Record signal, noise, SNR, accepted power, feed-line current and any overload or thermal state.
7. Sample the required paths. Group paired results by band, azimuth, distance and time rather than averaging unlike circuits.
8. Repeat the experiment. Preserve raw records, state uncertainty and separate observations from the mechanism used to explain them.

What Models and Field Measurements Contribute

A model can show why one geometry is expected to favour certain azimuths or elevations, but only for the conductors, ground and surroundings represented. Perform convergence and sensitivity checks, and add the feed line or common-mode path when it carries appreciable current.

Near-field strength around the antenna is not a direct gain measurement. IEEE 149 treats gain and pattern measurement as range problems with defined geometry, instrumentation and uncertainty. Local probes remain useful for finding current-path changes, leakage and coupling, provided they are not relabelled as far-field gain.

The best conclusion joins both: model the mechanism, measure the installed system, and keep the claim no broader than the tested bands, paths, site and conditions.

Primary Engineering Sources

• IEEE 149-2021, Recommended Practice for Antenna Measurements: pattern, gain, test-range and instrumentation practice.
• IEEE 145-2025, Standard for Definitions of Terms for Antennas: gain, efficiency, polarization and pattern terminology.
• Recommendation ITU-R P.372-17, Radio Noise: external-noise sources, variability and the feeder/screening boundary in receive systems.
• IARU Region 1 Technical Recommendation for S-Meter Readings: the nominal six-decibel S-unit and reference-level convention.

Joeri’s Bottom Line

An antenna does not win because its name carries a familiar pattern, because one station reported more S-points or because one afternoon produced more spots. The winner is the installation that repeatedly serves the chosen paths with useful signal-to-noise ratio, acceptable loss, safe stress and a controlled current path.

Compare apples with apples when you want to isolate a design variable. Compare complete fruit baskets when you want to choose a station. Just say which experiment you ran.

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