A Deep Null Is Not the Whole Receive Story
A Deep Null Is Not the Whole Receive Story
A sharp rear notch looks decisive on a polar plot. On the low bands, however, the useful question is not how impressive one null appears. It is whether the complete installed array improves reception, remains stable and rejects the noise that is actually present.
I understand the attraction of the rear-null contest. A deep notch promises a clean answer to QRM, and classic low-band phasing work made null steering an essential station tool. The mistake is turning one notch into the score for the whole antenna.
The working rule: use a deep steerable null against a known directional interferer. For general low-band reception, judge the array with three-dimensional directivity, null location and width, calibration stability, overload margin and controlled A/B/A signal-to-noise measurements.
Why the Rear Null Became the Headline
Low-band operators often fight a strong local source on a repeatable bearing: a power line, switch-mode supply, industrial controller or another station. When two or more receive channels are combined with the right amplitude and phase, that source can be placed near a cancellation direction. The audible result can be spectacular.
That practical success sits comfortably beside the phased-array tradition represented by John Devoldere, ON4UN, in Low-Band DXing. It does not mean that ON4UN's work reduces receiving performance to one rear-null figure. Pattern bandwidth, element systems, phasing methods and installation behaviour all matter. Joeri's point is narrower and more useful: do not let the most dramatic number on one cut hide the rest of the receiving system.
A null is a cancellation condition. It belongs to a stated frequency, azimuth, elevation, polarization, steering state and reference plane. Move any of those conditions—or let channel amplitude and phase drift—and the minimum can move or fill in.
Null Depth, Front-to-Back Ratio and RDF Answer Different Questions
Null depth compares the response at a minimum with a declared reference, often the forward response or pattern peak. Front-to-back ratio compares two declared directions, commonly the forward direction and the direction 180° behind it. Neither value describes how much response remains over all the other directions.
Receiving directivity factor (RDF), as used in low-band practice, is the receive-pattern directivity expressed in decibels. For a declared forward direction Ω0 and power response G(Ω), a useful form is:
RDF(Ω0) = 10 log10[G(Ω0)/Gavg] dB
Gavg = (1/4π) ∫4π G(Ω)dΩ
If the declared forward direction is the pattern maximum, the result is peak directivity in decibels. The average is over the full sphere, so a single horizontal azimuth cut cannot establish RDF. Ground, elevation pattern and polarization are part of the result.
RDF is valuable because it summarizes how strongly a pattern favours one direction over an angularly uniform field. It is not a universal promise of site SNR. Real noise can be directional, vertically and horizontally structured, correlated between elements or generated inside the station. Receiver noise and wanted-signal arrival angle also matter.
A Beautiful Null Can Still Be Fragile
Deep cancellation demands close complex equality. Consider two contributions intended to cancel. If one channel has a small fractional amplitude error ε and a small phase error δ in radians, the residual relative to one contribution is approximately:
|1 − (1 + ε)ejδ| ≈ √(ε² + δ²)
That first-order relation explains why the forward lobe may look acceptable while a very deep notch disappears. Cable length, loss, group delay, termination impedance, temperature, component tolerance, mutual coupling and ground changes all alter the complex channel response. The deeper the claimed null, the more the measurement must document calibration and repeatability.
Position matters as much as depth. A 35 dB minimum at the wrong elevation does not cancel a 20 dB local interferer arriving by another path. A narrow notch can also remove a wanted signal when ionospheric arrival direction changes. Record null angle, width and stability instead of publishing only the minimum number.
When a Deep Steerable Null Is Exactly the Right Tool
A null earns its keep when the interference is dominated by a stable directional source. Confirm that condition rather than assuming it:
- Rotate or switch the array through known states and record whether the interference follows one bearing.
- Repeat the observation at different times and frequencies to see whether the source direction and spectrum remain stable.
- Use fixed receiver settings and enough attenuation to rule out overload, intermodulation or AGC effects.
- Compare the wanted signal and interferer separately; a lower S-meter reading is not automatically better SNR.
When those checks identify one persistent bearing, steerable cancellation can produce a larger practical benefit than a modest change in overall RDF. That is not a contradiction. It is the difference between solving a specific interference geometry and optimizing a general receive pattern.
Distributed Noise Needs a Wider View
Many stations face several noise sources, reradiating conductors and time-varying electronics. Their combined field need not arrive uniformly, but it is rarely represented by one direction directly behind the array. In that environment, broad directivity and pattern stability often matter more than an extreme isolated null.
The combined voltage response of a calibrated array can be written compactly as:
B(Ω,f) = wHaemb(Ω,f)
The weight vector w contains the complex channel weights. The embedded-response vector aemb includes each installed element with the other ports in their operating terminations, plus coupling, ground, feedlines, supports, polarization and electronics to declared reference planes. An ideal array factor is a useful beginning, not the installed answer.
If Rn is the channel-noise covariance matrix, combined output noise is:
Pn,out = wHRnw
The weights that maximize modelled RDF need not maximize SNR at a particular site. Correlated external noise, independent receiver noise, element sensitivity and overload margin can change the best choice. This is why I keep modelled RDF, F/B, null depth and measured A/B/A SNR in separate columns.
Hybrids and Delay Lines Do Not Remove the Measurement
Coaxial delay-line phasing and hybrid-based networks can both be useful. A physical delay produces phase that changes with frequency. A quadrature or other hybrid has its own amplitude balance, phase balance, isolation, insertion loss and termination requirements across frequency. Neither architecture delivers broadband accuracy by name alone.
A hybrid may provide useful port isolation and a controlled combining relationship when it is correctly terminated and characterized. A delay line may provide the intended spatial compensation over a chosen window. In both cases, measure the complete channels—elements, feedlines, electronics, combiner and receiver interface—rather than assigning stability or RDF to one component.
Station Noise Can Bypass the Pattern
An array cannot reject interference that enters downstream of the spatial sampling. Common-mode current on feedlines, coupling into control cables, local digital noise, receiver overload and leakage between channels can all bypass or distort the intended pattern.
Check the complete installation with the array connected, terminated and powered as used. Measure channel current where practical, keep noisy equipment away from the elements, confirm receiver linearity with attenuation and restore the baseline after every diagnostic change. A stable pattern is a property of the installed system, not just its geometry.
A Better Acceptance Test
Before calling a receive-array state successful, record:
- Pattern definition: frequency, steering state, forward direction, polarization, elevation range and reference plane.
- Three-dimensional response: RDF or directivity, F/B, sidelobes, and the position, width and depth of relevant nulls.
- Channel calibration: amplitude, phase and group delay across the operating window, including temperature and reconnection checks.
- Installed interactions: element terminations, mutual coupling, supports, ground, feedline routing and common-mode current.
- Receiver headroom: overload, blocking and intermodulation checks with known attenuation.
- Field result: repeated A/B/A comparisons of wanted signal, noise and interference using fixed receiver settings.
A switch that produces the deepest trace minimum is not automatically the best listening state. The state that repeatedly recovers the wanted signal with usable margin is the one worth keeping.
Engineering references
- IEEE 145-2025 — Standard for Definitions of Terms for Antennas
- IEEE 149-2021 — Recommended Practice for Antenna Measurements
- Lawrence Livermore National Laboratory — Numerical Electromagnetic Code, NEC v5.0
- Recommendation ITU-R BS.705-2 — HF Antenna Characteristics and Diagrams
- John Devoldere, ON4UN — Low-Band DXing, phased receiving-antenna practice and system context
Stop chasing a null as a trophy. Use it deliberately when one interferer deserves cancellation. For the rest of the band, build the array for calibrated, stable directivity and prove the result with the signals and noise at your own site.
Mini-FAQ
- Is a deep rear null proof of a good receive array? No. It describes cancellation in a declared direction and state, not the full pattern, stability or site SNR.
- Is front-to-back ratio the same as RDF? No. F/B compares two directions, while RDF compares a declared forward response with the response averaged over the full sphere.
- When should I prioritize a null? Prioritize a steerable null when measurements show that one stable directional interferer dominates the listening problem.
- Why can a deep null disappear? Small channel-amplitude or phase errors, coupling, terminations, temperature and ground changes can disturb the cancellation.
- Does higher RDF guarantee better SNR? No. Site noise direction and correlation, receiver noise, element sensitivity and overload can change the installed result.
- Are hybrids automatically more stable than delay lines? No. Either approach must be characterized across frequency with its real terminations, loss, amplitude balance, phase balance and complete channels.