Modeling Non-Resonant Traps in NEC: Measure the Network First

I do not start every antenna in a solver. I start with the current path, the available wire, the intended bands and a defensible first estimate. Then I use simulation to expose interactions that the sketch cannot show, measure the real load, update the model and build a prototype. That loop is more useful than either blind optimisation or trial and error without a model.

“Non-Resonant Trap” Describes a Design Intent

A conventional parallel-resonant trap is often designed to present a high impedance near one chosen frequency. A network used away from a sharp parallel resonance can also reshape current over several bands. RF.Guru uses non-resonant trap for that current-shaping role, not as a promise that one component is broadband, lossless or opaque.

The wire beyond the load still couples through the finite series impedance and through the electromagnetic field around the complete antenna. Current may be reduced, redistributed or phase-shifted; it does not encounter an ideal open circuit. The result depends on the load position, the electrical length and orientation of every conductor, the feed and return structure, nearby objects and ground.

This is why I look at conductor current rather than declaring an “effective endpoint” from inductance alone. On one band the outer section may carry little current. On another it may carry enough current to change feed impedance or create an important lobe. The model must show both magnitude and phase, and the prototype must confirm the prediction.

An Ideal Inductor Is Only the First Model

For an ideal series inductor, the load is Z = j2πfL. Add a series resistance and an ideal series RLC model becomes:

That equation is useful when the physical network behaves like the assumed topology over the frequencies of interest. A real coil also has conductor and connection loss, distributed capacitance, frequency-dependent current distribution and sometimes a magnetic core whose complex permeability, temperature and drive level matter. Near self-resonance, a single ideal inductance can be a poor description.

There is therefore no universal 5–15 µH range, resistance limit or frequency at which an NRT becomes “invisible.” Choose the network from the target current distribution and acceptable loss and stress, then characterise the built part across the actual frequency and operating range.

What NEC-2 Loading Cards Actually Represent

The original NEC-2 LD card supports several distinct load models. The load is attached to selected wire segments; it does not require a fictitious tiny-radius wire. The type number matters:

A fixed LD 4 value does not automatically acquire the measured frequency dependence during a sweep. For a measured NRT, I generate or rerun each frequency point with the corresponding R + jX. If the software supports an external network, fitted rational function or circuit co-simulation, measured network data can be used through that route—after confirming how the program defines ports, reference planes and current direction.

The NEC-2 LD-card reference gives the exact field definitions and units. The original NEC-2 User's Guide also documents segmentation, excitation, ground and output controls that surround the load model.

MMANA-GAL Can Model Lumped Loads

MMANA-GAL is based on MININEC, not a simplified NEC-2 engine. Its own documentation describes lumped LC loads, fixed R + jX loads and more complicated loads expressed as rational functions of the Laplace variable. That makes it capable of useful NRT studies when the selected load representation matches the real network.

The limitation is not that MMANA-GAL cannot place an inductor. The limitation is the same one found in any model: an ideal or fixed load is only as valid as the data behind it. A single fixed R + jX value reused across several bands is not the measured Z(f). A fitted network that ignores loss or self-resonance can also predict the wrong current and stress.

Use the official MMANA-GAL Basic help to confirm its load definitions and examples. Then verify the generated current distribution and feed impedance against a second implementation or measured case before trusting a design conclusion.

Measure the Network at Its Terminals

The trap model needs a declared reference plane. Calibrate the VNA or impedance analyser as close as practical to the two terminals that will join the antenna wire. Keep the fixture short and repeatable, and de-embed it only when its behaviour is characterised well enough to justify the correction.

A one-port series fixture can estimate the impedance of a two-terminal network, but fixture inductance, shunt capacitance and connector repeatability become important as impedance rises. A two-port shunt-through or series-through method may provide a better measurement range in some cases. Select the fixture from the expected impedance and validate it with known standards.

Record complex data across all relevant bands, with enough frequency resolution to expose self-resonance or sharp changes. Repeat after remounting the device. For a wound component, record physical orientation, lead length, nearby conductors and temperature. Small layout changes can be part of the RF network.

The Keysight Impedance Measurement Handbook explains fixture selection, compensation, residual errors and the impedance ranges of common measurement methods.

The Complete Antenna Still Decides

A correct load does not rescue an incomplete antenna model. Include the actual conductor geometry, wire diameter and material, feedpoint and return path, source model, mast or feedline conductors that carry current, ground model and important nearby structures. Check segmentation and convergence rather than trusting the first smooth plot.

For every band, inspect:

• complex feedpoint impedance at the declared model plane;
• current magnitude and phase immediately before and after each load;
• current on outer wire sections, feedline, support or return conductors;
• power dissipated in the load and other lossy elements;
• voltage and current stress across the physical network;
• far-field pattern, elevation cuts and realised gain in the required directions; and
• sensitivity to plausible changes in geometry, ground and component data.

A deep current reduction at one segment can be useful, but it does not by itself prove efficiency or a desirable pattern. Likewise, a low SWR can coexist with loss in the load, feedline or ground. Keep accepted power, dissipated power and radiated pattern as separate results.

Heuristic Design, Simulation and Measurement Form One Loop

Simple wavelength estimates remain useful for a first wire length, but formulas with one shortening constant do not guarantee resonance, impedance or pattern after bends, loading, ground and nearby conductors are added. A 5/8-wave label also does not guarantee a low-angle pattern without the complete geometry and ground conditions.

A Reproducible NRT Modelling Sequence

1. Define the purpose. State which conductor current should change, on which bands, and what pattern or matching consequence is expected.
2. Draw the complete current path. Include wire sections, feed and return conductors, support conductors and relevant ground or structures.
3. Start with an ideal load. Sweep plausible topology, value and position to understand sensitivity—not to claim final performance.
4. Build and measure the network. Save complex impedance versus frequency at its installation terminals, including repeatability and fixture uncertainty.
5. Insert measured values. Use the corresponding R + jX at every frequency or a validated fitted/network model over the required band.
6. Check numerical validity. Refine segmentation, inspect convergence, confirm load placement and compare with another solver or known case where practical.
7. Inspect current and power. Look on both sides of the load and account for dissipation rather than reading only SWR.
8. Inspect the pattern. Compare elevation and azimuth cuts, realised gain and unwanted lobes for the intended paths.
9. Test sensitivity. Vary plausible component, height, ground and geometry values to find fragile conclusions.
10. Prototype and restore. Measure impedance, current or field response with A/B/A changes, then feed the discrepancy back into the model.

Primary Software and Measurement References

• NEC-2 LD-card documentation—the original definitions for series, parallel, distributed, fixed-impedance and conductivity loads.
• Burke and Poggio, Numerical Electromagnetics Code (NEC)—Method of Moments, Part III: User's Guide—the original NEC-2 model-input and output reference.
• MMANA-GAL Basic documentation—MININEC basis, lumped-load options and model examples.
• Keysight, Impedance Measurement Handbook—measurement methods, fixtures, compensation and residual-error boundaries.
• Lawrence Livermore National Laboratory, NEC-5 Validation Manual—validation cases and comparison against analytical or measured results.
• IEEE 149-2021, Recommended Practice for Antenna Measurements—standardised antenna measurement terminology and methods.

Joeri's Bottom Line

You can begin an NRT antenna with a current-path sketch, a sensible load estimate and ordinary antenna physics. That is engineering, not a rejection of simulation. The next step is to ask the solver where current actually flows and what pattern the full structure produces.

Then replace the ideal load with the network you measured. If the simulated current, feed impedance and prototype disagree, do not tune the story—find the missing conductor, fixture error, parasitic or environmental term. The useful result is the design loop: estimate, model, measure, update and verify.

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