Diversity Receive on HF: Why One Antenna Is Often Not Enough

I do not expect one receiving antenna to win every minute on every HF path. A low horizontal antenna can be excellent for regional high-angle work while a vertical, loop or directional array gives better copy on another bearing or in another noise field. The useful word is not more; it is different.

What Diversity Can—and Cannot—Do

Diversity does not create transmitter power, antenna gain or information that never reached the station. It reduces the probability that every receive path is simultaneously trapped in the same fade or noise maximum. The two channels may differ through antenna position, pattern, arrival angle, polarization, frequency or time.

Independence is a matter of correlation, not antenna names. Two nearby antennas can still be useful if their patterns, polarizations or local-noise pickup differ. Two widely separated antennas can disappoint if the same dominant signal and interference reach both in almost the same way. ITU work on HF diversity makes the same point: improvement falls as channel correlation rises, and any diversity claim must state the channel count, non-diversity reference and performance metric.

Selection and Combining Are Different Jobs

The simplest station uses an A/B switch and chooses the channel with better copy. Selection combining automates that decision. Equal-gain combining aligns phases and gives the branches equal weight. Maximum-ratio combining weights them according to their useful signal and noise. These methods do not promise the same result.

Coherent combining requires phase- and frequency-coherent receive channels, known delays and a valid signal model. Summing two arbitrary receiver audio outputs can reinforce wanted audio, reinforce noise, or cancel part of either. Even selection needs care: the strongest S-meter reading is not necessarily the best SNR, especially when AGC, bandwidth, preamplifier gain or ADC scaling differs.

For an operator starting with one receiver, fast manual switching is already valuable. It teaches which antenna rejects the neighbour's power supply, which one favours the wanted bearing and whether a deep fade moves between channels. That evidence should come before a complex combiner.

Why HF Polarization Refuses a Fixed Label

A horizontal dipole and a vertical monopole describe the antennas at the station; they do not force the returning skywave to remain purely horizontal or vertical. The ionosphere is an anisotropic, magnetized plasma. A transmitted field can excite ordinary and extraordinary characteristic waves that follow different phase paths. Their recombination can rotate the orientation of a linear field, while an individual downward characteristic wave on a steep NVIS path can approach circular polarization under the right geometry.

That distinction matters. “Faraday rotation” is not a blanket explanation for every circular signal, and the received polarization is not fixed by frequency alone. It depends on frequency, electron density, geomagnetic field, propagation direction, path and mode composition. A practical receiver therefore measures what arrived instead of assigning one permanent polarization-loss number.

ITU-R F.763 documents the operational result for HF data systems: a minimum on one orthogonal antenna element can coincide with a maximum on the other, and a diversity combiner can exploit that partial decorrelation. It does not say that every pair of crossed wires is automatically a calibrated polarization-diversity antenna.

What Witvliet's NVIS Measurements Actually Show

Ben Witvliet and colleagues did more than repeat the idea that low-band polarization changes. Their NVIS experiment separated the ordinary and extraordinary characteristic waves with oppositely circular receive channels and reported at least 13 dB isolation in that measured arrangement. Their central result is useful because diversity improves when the streams are less correlated.

That is evidence for a measurement method and a propagation mechanism—not a universal “circular always wins” number. The result belongs to the tested path, frequencies, antenna system, calibration and definition of isolation. A station with poor quadrature, unequal elements, cable error, local coupling or the wrong hand convention will not inherit it.

The other station does not need to transmit a circularly polarized antenna for opposite-hand reception to be worth testing. A linearly launched wave can excite both magneto-ionic characteristic waves. But whether the two receive outputs are usefully independent at a particular moment must still be observed.

Crossed Elements Need More Than a 90° Label

Two orthogonal antennas plus a nominal quarter-cycle phase shift are the familiar sketch. The installed system also needs comparable patterns over the arrival sector, matched amplitude at a declared reference plane, correct quadrature, stable cable delay and verified port isolation. Buildings, feedlines and the ground can disturb all of them.

Circular hand labels are viewing-direction dependent. Document the convention and verify the complete signal chain with a known field or a calibrated model. A label such as LHCP or RHCP without its viewing convention and reference plane is not enough for reproducible work.

When the elements and channels are imperfect, the outputs are generally elliptical mixtures rather than pure opposite hands. They may still provide useful diversity. The engineering question is not whether the labels are beautiful; it is whether wanted-signal fades and noise are sufficiently decorrelated without introducing overload, loss or calibration drift.

Vertical, Horizontal and Directional Views

A low horizontal antenna often provides a strong high-angle view on the lower bands. A vertical or directional receive antenna may emphasise lower arrival angles, another azimuth or a different local-noise field. Neither statement is a guarantee. Height in wavelengths, ground, pattern, feedline common mode and nearby conductors decide the installed response.

On 80 and 40 metres I want a regional high-angle reference and at least one genuinely different view. That might be a crossed pair, a loop, a vertical, a terminated wire or a directional array. On 160 metres the station's external and local noise can dominate the decision. On higher HF bands, directivity and bearing often become more useful sources of diversity than a low-band NVIS arrangement.

At our Field Day-style operation we used circular receive on 40 and 80 metres during two consecutive events, and those bands produced our highest contact totals. That remains an operational observation, not a controlled attribution: band conditions, operators, transmit antennas and activity changed too. It is enough to justify keeping multiple receive views available; it is not enough to publish a universal contact or SNR advantage.

Six Metres Is a Different Borderland

Six metres sits at the HF/VHF boundary. During sporadic-E, the existence, geometry and duration of the opening can dominate the contact opportunity. ITU-R P.534 treats sporadic-E statistically through the layer's critical frequency and path geometry; it does not make all antennas equivalent.

A resonant dipole, halo, vertical, Moxon or Yagi can all put an operator on the band. Horizontal polarization remains the usual weak-signal convention and vertical the usual FM convention, but an ionospheric path can change the field. “Use any antenna and listen” is good advice for catching an opening. “Every antenna performs the same” is not.

A Comparison That Survives Fading

• Declare the question. Are you measuring signal level, noise, SNR, readability, outage probability, error rate or correlation?
• Describe every branch. Record antenna geometry, position, orientation, feedline, filters, gain, bandwidth and reference plane.
• Calibrate the channels. Inject a common signal and measure amplitude, phase, delay and frequency coherence where the combining method needs them.
• Freeze receiver behaviour. Use identical bandwidth, AGC, attenuation, preamplifier and ADC settings, or record the differences explicitly.
• Compare simultaneously or quickly. A slow A/B sequence can mistake propagation for antenna performance.
• Record signal and noise separately. A quieter channel can give better copy with a lower S-meter reading.
• Measure correlation over time. One dramatic fade is an example, not a diversity distribution.
• Repeat the baseline. Use A/B/B/A or simultaneous recordings and return to the initial state.
• Check strong-signal behaviour. Extra branches, preamps and combiners must not create compression or intermodulation.

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