Why Polar Plots Lie—Unless the Model Earns Trust

Every ham loves a good polar plot. A beautiful lobe here, a crisp null there—it looks scientific, precise and trustworthy. My objection is not to modelling. It is to publishing the answer while hiding the question that the model actually solved.

A plot is a prediction, not a property deed

A radiation pattern belongs to a declared antenna configuration at a declared frequency and polarization. Change the element geometry, height, conductor properties, ground, feed structure, return path, loading, nearby objects or observation cut and the answer can change. A free-space model is not false when it is labelled as free space; it simply does not predict a wire installed over a roof.

Likewise, “average ground” is not a physical field behind every house. It is a pair of electrical parameters used by a chosen ground model. That abstraction may be adequate for a sensitivity study and inadequate for a shoreline, layered soil, reinforced concrete, wet woodland or a metal roof. The plot earns trust only inside the problem it represents.

Nearby objects are inputs, not automatic invalidation

The old warning that anything inside one wavelength makes a plot meaningless is too blunt. Coupling depends on electrical size, material, orientation, separation, current path and frequency. A short dry branch and a resonant metal gutter are not equivalent. An omitted object matters when adding a defensible model of it changes the quantity you care about by more than your decision tolerance.

That gives us a better test: add the mast, feedline, support wire, roof edge or soil variation one at a time, then compare pattern, impedance and conductor current. If the result is stable, the simple model may be enough. If a null moves 20 degrees or fills by 15 dB, the elegant baseline was never the installed answer.

Current matters, but one current maximum is not the pattern

Yes, current distribution is central. The far field is formed by the vector contribution of the complete time-varying current distribution, including amplitude, phase, direction and geometry. Maximising current at one point does not prove high efficiency, useful directivity or the wanted elevation angle. A lossy conductor can carry current; exterior coax current can also radiate; two segments can reinforce in one direction and cancel in another.

Use current plots to understand which conductors joined the antenna and where loss or unintended radiation may occur. Then calculate the field. “Current, not SWR” is a valuable counterweight to match-only thinking, but it is not permission to discard pattern analysis.

Do not mix directivity, gain and realised gain

A normalized polar plot can hide every absolute loss because its strongest direction is scaled to the same outer ring. Directivity describes how radiation is distributed in angle. Gain also includes radiation efficiency. Realised gain additionally includes mismatch at the stated port under the applicable convention. Installed receive performance adds polarization alignment, local noise, receiver state and the complete feed system.

That is why a “5 dBi” caption needs more than a pretty curve. State whether it is directivity, gain or realised gain; the reference antenna; the polarization component; the angular cut; the ground and loss model; and whether the feedline, matching network and common-mode path are inside the calculation.

Make the numerical model earn its digits

NEC and other full-wave solvers are powerful because they solve a defined electromagnetic model. Their precision is not evidence that the model is complete. For a thin-wire method-of-moments model, segment length, radius, junctions, source placement, loading, material loss and ground treatment all need to remain within the solver’s valid region.

• Declare the coordinate system. State azimuth, elevation, polarization, frequency and whether the cut is total field or one component.
• Model the complete intended RF path. Include the feed and return conductors that can carry radiating current, or explain why their omission is bounded.
• Use measured or defensible material inputs. Document conductor diameter and conductivity, ground conductivity and relative permittivity, loads and connection loss.
• Run convergence checks. Refine segmentation and the geometry representation until the reported impedance, current and pattern quantities stop moving materially.
• Run sensitivity checks. Vary uncertain height, ground, loss and nearby conductors across plausible limits instead of publishing one lucky curve.
• Check conservation and plausibility. Accepted power, loss, radiated power, efficiency, pattern integration and reported gain must tell one consistent story.

A backyard pattern needs a measurement plan

A casual signal report is useful experience, not a calibrated far-field pattern. A defensible test fixes or records source power, distance, height, polarization, receiver gain state, cable routing and environment. Rotate the antenna or source through repeatable angular positions, measure both principal cuts where practical, and repeat the first point at the end to reveal drift.

Residual multipath, alignment, position, mismatch, receiver linearity, noise and common-mode current all contribute uncertainty. IEEE 149 treats range design, instrumentation and pattern measurement as part of the result. NIST’s antenna-metrology work likewise treats uncertainty as a budget, not a footnote. For a home installation, an A/B/A field test may not produce an accredited pattern, but it can still falsify a fragile model when the controlled observations disagree beyond the stated uncertainty.

What antenna builders should really focus on

Keep the polar plot. Just keep its assumptions beside it. Use the model to choose measurements, identify sensitive geometry and compare controlled changes. Use current and loss analysis to explain the mechanism. Then validate the decision at the installed antenna rather than arguing over the smoothness of a simulated line.

A polar plot does not lie by itself. We make it lie when we remove the model, scale, polarization and uncertainty that give the curve meaning.

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