Antenna Modelling with EZNEC — Complete Ham Radio Guide
A complete practical guide to using EZNEC for amateur radio antenna modelling. Covers the NEC2 simulation engine that underlies EZNEC, installing and configuring the software, building wire antenna models from scratch, setting up ground and frequency parameters, running simulations, interpreting gain and radiation pattern plots, and using modelling results to optimise real antenna designs before cutting a single piece of wire.
NEC2 — the simulation engine
EZNEC is a Windows-based graphical front-end for the NEC2 antenna simulation engine, developed by Roy Lewallen W7EL. NEC2 — the Numerical Electromagnetics Code, version 2 — is a method-of-moments electromagnetic simulation program originally developed at Lawrence Livermore National Laboratory and released into the public domain. It models antennas as collections of thin wire segments, solves for the current distribution along those segments when driven at a specified frequency, and computes the resulting radiation pattern, gain, feedpoint impedance, and other antenna parameters.
NEC2 is accurate for thin-wire structures — dipoles, Yagis, loops, verticals, and most wire antennas that amateur radio operators build. Its accuracy depends on the antenna being modelled as a collection of straight wire segments that are thin relative to their length and to the operating wavelength. For antennas made from thick tubing, large flat surfaces, or complex three-dimensional structures, NEC2 has known limitations. For the vast majority of amateur HF and VHF antenna designs, it produces results that closely match measured performance.
EZNEC versions and availability
EZNEC has been distributed in several versions. EZNEC Demo is the free version, limited to 20 wires and 500 segments — sufficient for most simple HF antenna models including dipoles, Yagis up to three or four elements, verticals with radials, and loops. EZNEC+ extends the segment limit and adds features. EZNEC Pro/2 and Pro/4 are full professional versions with NEC4 engine support and higher segment limits for complex models.
As of 2023 Roy Lewallen W7EL retired EZNEC from commercial sale and released the full EZNEC+ version as freeware, available from his website at eznec.com. This means the previously paid version is now available at no cost, giving access to higher segment limits and additional features without purchasing a licence. The Demo version remains available for those who prefer a simpler introduction. Both run on Windows; Linux and macOS users can run EZNEC under Wine with minor configuration.
What EZNEC tells you and what it cannot
EZNEC computes the following outputs for any antenna model at any specified frequency: gain in dBi and dBd at any elevation and azimuth angle, the full three-dimensional radiation pattern as a 2D plot in any plane, feedpoint impedance as resistance and reactance, SWR relative to a specified reference impedance, current distribution along all wire segments, and near-field information. These outputs allow the modeller to compare antenna designs, optimise element lengths and spacing, understand pattern shapes, and predict feedpoint impedance before building.
EZNEC cannot model antenna performance accurately when the antenna is near lossy structures that are not modelled — building walls, trees, and vehicles that are not included in the wire model are invisible to the simulation. It cannot model the effect of insulation on wire elements, which causes a small but real frequency shift compared to bare wire. It cannot model distributed transmission line effects in traps or coils with the same precision as a dedicated trap modelling tool. And it cannot account for manufacturing tolerances, connector losses, or feedline radiation. The results are always best-case theoretical performance of the modelled structure, not a prediction of what you will measure in your specific yard with your specific ground conductivity.
Ground models — the most important setting
The ground model is the single most influential setting in an EZNEC model for HF antenna work. EZNEC offers three ground options: free space (no ground), perfect ground (infinite conductivity), and real ground (specified by conductivity and dielectric constant). Free space is useful for comparing antennas in isolation and for VHF/UHF work where ground effects are minimal. Perfect ground is unrealistic but produces the optimistic results often quoted in antenna advertisements. Real ground using the Sommerfeld-Norton calculation is the most accurate and the appropriate choice for all practical HF antenna evaluation.
The real ground parameters that matter most are conductivity (in Siemens per metre) and relative dielectric constant. Typical suburban ground has conductivity around 0.005 S/m and dielectric constant around 13. Poor rocky or sandy ground might be 0.001 S/m and 7. Rich agricultural ground might reach 0.030 S/m and 20. These differences significantly affect the modelled gain of low antennas — a dipole at 10 metres height on 40m can show 2 to 3 dB difference in low-angle gain between poor and good ground in the model, which is a real effect that matters for DX work.
Antenna Element Scaling and Frequency Calculator
Scale a known antenna design to a new frequency, or calculate element lengths for standard antenna types at any frequency. Useful for translating EZNEC models between bands.
Modelling a 20m Half-Wave Dipole in EZNEC
A step-by-step walkthrough of building, configuring, and running a complete antenna model. Use this as a template for any wire antenna.
Open EZNEC and set frequency and units
Launch EZNEC. From the main screen, click Frequency and enter 14.200 MHz — the centre of the 20m phone band. Click Units and select Metres if not already selected. EZNEC can work in feet or metres; metres are generally more convenient for HF antenna dimensions. These two settings define the simulation frequency and the coordinate system for all wire positions you will enter.
Set the ground type
Click Ground Type on the main screen. Select Real/MININEC for a first model — this uses a simplified but fast ground calculation suitable for horizontal wire antennas at modest heights. For verticals and for more accurate results on low horizontal antennas, select Real/High Accuracy (Sommerfeld-Norton). Enter ground conductivity 0.005 S/m and dielectric constant 13 for typical suburban ground. These values can be varied later to explore how ground quality affects performance.
Add the antenna wires
Click Wires on the main screen to open the wire table. Click Add Wire. A dipole consists of two wires meeting at the feedpoint. Enter Wire 1 coordinates: End 1 at (−5.03, 0, 10) and End 2 at (0, 0, 10). This places the first arm of the dipole from the feedpoint at the origin to 5.03 metres to the left at 10 metres height above ground. Enter 11 for the number of segments — use an odd number so a segment boundary falls at the exact centre of each wire for accurate feedpoint placement. Click Add Wire again and enter Wire 2: End 1 at (0, 0, 10) and End 2 at (5.03, 0, 10), again with 11 segments. The two wires share the point (0, 0, 10) as the feedpoint junction.
Enter wire diameter
In the wire table, enter the conductor diameter for each wire. For a wire dipole made from 1.5mm diameter copper wire, enter 0.0015 metres (1.5mm). For a dipole made from 4mm diameter copper pipe, enter 0.004 metres. The conductor diameter affects the resonant length — thicker conductors resonate slightly shorter than thin wire. This is why EZNEC models often produce slightly different element lengths than the free-space formula, and why specifying the actual conductor diameter you plan to use gives the most accurate results.
Place the source (feedpoint)
Click Sources on the main screen. Click Add Source. Set the wire number to 1 (the first dipole arm) and the position to End 2 — the end of Wire 1 that meets Wire 2 at the feedpoint. Set source type to Voltage and leave magnitude at 1V and phase at 0 degrees. This places the excitation at the junction of the two dipole arms, which is the correct feedpoint position. EZNEC will compute the current flowing from the source, from which it derives feedpoint impedance and all other parameters.
Run the simulation and read the output
Click FF Plot (far field plot) or press F7 to run the simulation and display the radiation pattern. EZNEC will show an elevation pattern plot — gain versus elevation angle — and an azimuth pattern plot. Note the maximum gain value displayed in dBi and the elevation angle of maximum radiation. For a 20m dipole at 10 metres height the peak gain should be approximately 7 to 8 dBi at an elevation angle of around 25 to 35 degrees, depending on ground type. The feedpoint impedance appears in the Source Data box — for a resonant dipole expect approximately 70 to 80 ohms resistive with near-zero reactance.
Trim to resonance and explore height effects
If the reactance at the source is not near zero, the dipole is not at resonance. Positive reactance means the antenna is too long — reduce both arm lengths by equal amounts. Negative reactance means too short — increase both arms. Typical adjustments are 2 to 5 cm per arm. After trimming to resonance at 14.200 MHz, vary the antenna height — change the Z coordinate of both wires — and rerun the simulation to see how height affects gain and radiation angle. This is one of the most instructive EZNEC exercises: watching the low-angle gain increase as height rises from 5 to 15 metres shows clearly why antenna height matters for DX.
Gain and radiation patterns
EZNEC reports gain in dBi by default. To convert to dBd subtract 2.15. The elevation pattern plot shows gain versus elevation angle above the horizon — the angle of peak gain tells you what propagation paths the antenna favours. A peak at 20 to 30 degrees suits DX propagation on the higher HF bands. A peak at 60 to 80 degrees suits NVIS regional propagation on 40m and 80m. The azimuth pattern at the elevation angle of peak gain shows the horizontal directivity — a dipole shows a figure-eight pattern with nulls off the wire ends, a Yagi shows a cardioid forward lobe.
The 3D pattern view in EZNEC combines both elevation and azimuth into a surface plot that immediately reveals the full radiation characteristic of the antenna. For a dipole the 3D pattern is a toroid — a doughnut shape around the wire axis. For a Yagi it is a teardrop pointing in the direction of the directors. Viewing this 3D plot while changing antenna parameters — height, element length, element spacing — gives an intuitive feel for how antenna geometry controls radiation that no amount of reading can fully replace.
Feedpoint impedance and SWR
The Source Data output in EZNEC shows the feedpoint impedance as resistance (R) and reactance (X) in ohms. At resonance X is zero. The SWR relative to 50 ohms is calculated automatically and displayed alongside the impedance. A resonant half-wave dipole in free space shows approximately 73 ohms resistive — SWR 1.46:1 to 50 ohms — which is acceptable without a matching network. The same dipole at certain heights above ground will show higher or lower impedance depending on the ground reflection contribution.
Use the frequency sweep to generate an SWR versus frequency plot across the full band. This reveals the bandwidth of the design — how quickly SWR rises as frequency moves away from resonance — and shows any secondary resonances on harmonic bands. A multiband antenna like an EFHW can be verified to have acceptable SWR on all intended harmonic bands in a single frequency sweep from 7 to 30 MHz.
Current distribution
The current distribution display — accessible from the View menu — shows the magnitude of RF current flowing along each wire segment as a coloured overlay on the antenna structure. This display is invaluable for diagnosing modelling errors and for understanding antenna physics. A correct dipole model shows maximum current at the centre (feedpoint) and zero current at the tips. A correct Yagi model shows the driven element carrying maximum current with the parasitic elements carrying induced current of lower magnitude.
Common modelling errors visible in the current display include: a source placed at the wrong position (the current maximum is not at the source), unconnected wires that should be joined (current stops abruptly at the wire end instead of flowing through), and wires that are electrically connected but not geometrically coincident (current shows a step discontinuity). Checking the current distribution plot before trusting the gain and impedance output is good modelling practice for any new antenna design.
Optimising a Yagi in EZNEC
EZNEC does not include an automatic optimiser in the standard version — element adjustment is done manually by changing dimensions and rerunning. The workflow for optimising a three-element Yagi is to start from the calculator-derived dimensions above, run the simulation, note the gain and F/B ratio, adjust reflector length by 1 to 2 cm in the direction that improves F/B, rerun, and repeat until F/B is maximised. Then adjust director length to maximise forward gain. Finally adjust element spacing to arrive at a feed impedance close to the target — typically 25 ohms for a three-element Yagi suitable for a hairpin match.
This manual optimisation process builds genuine understanding of how each element interacts with the others. A builder who has manually optimised a three-element Yagi model understands intuitively why the reflector controls F/B, why the director controls gain, and why feed impedance depends on the driven element's position relative to both. This understanding transfers directly to troubleshooting real antennas that do not perform as expected.
| Antenna | Wires needed | Segments (approx) | Source placement | Expected gain (real gnd) | Common errors |
|---|---|---|---|---|---|
| Half-wave dipole | 2 | 20–40 | Junction of two arms | 7–8 dBi at optimal height | Source off-centre; wires not meeting at feedpoint |
| Inverted-V dipole | 2 | 20–40 | Junction at apex | 6–7 dBi | Wrong apex angle; arms not symmetric in height |
| 2-element Yagi | 2 | 40–60 | Centre of driven element | 5–6 dBi at height | Reflector sourced instead of DE; elements not colinear |
| 3-element Yagi | 3 | 60–90 | Centre of driven element | 8–10 dBi at height | Wrong element order; spacing errors |
| Quarter-wave vertical | 5+ (1 + radials) | 50–100 | Base of vertical element | 4–5 dBi with radials | Radials not connected to base; wrong ground model |
| Full-wave loop | 4 (square) or 3 (delta) | 60–100 | One side of loop | 6–7 dBi broadside | Loop not closed; feedpoint at wrong position |
| EFHW | 1 | 20–40 | End of wire (near UNUN) | 6–8 dBi depending on height | Feedpoint at wrong end; counterpoise not modelled |
| Fan dipole (3 band) | 6 | 60–80 | Common junction | Similar to single dipole per band | Arms crossing; interaction between bands not checked |
How accurate is EZNEC compared to a real antenna measurement?
For simple wire antennas modelled with accurate wire dimensions, correct ground parameters, and appropriate segment density, EZNEC typically predicts resonant frequency within 1 to 2 percent, feedpoint impedance within 5 to 10 percent, and gain within 0.5 to 1 dB of measured values. The largest source of discrepancy is the ground model — real ground conductivity and dielectric constant vary across a property and are rarely known precisely. Modelling with several ground quality settings brackets the likely real-world performance.
Why does EZNEC show higher gain than I measure in practice?
Several factors reduce real antenna gain below modelled values: feedline loss, connector resistance, balun insertion loss, conductor surface oxidation, insulation effects on element resonant frequency, nearby metallic structures not included in the model, and ground parameters that differ from the model values. EZNEC models the antenna in isolation with ideal conductors — it does not model the complete station RF chain. Expecting modelled gain to match exactly is unrealistic; using the model for relative comparison between designs is the correct application.
What does the NEC2 segment length warning mean?
NEC2 requires that each wire segment be between 0.001 and 0.1 wavelengths long for accurate results. Segments shorter than 0.001 wavelengths cause numerical instability. Segments longer than 0.1 wavelengths reduce accuracy because the current variation along the segment is assumed linear — an approximation that breaks down for long segments. EZNEC displays a warning when segments violate these limits. Fix by increasing the number of segments on short wires or reducing the number of segments on long wires to bring all segments within the valid range.
Can EZNEC model traps and loading coils?
Yes, with some limitations. EZNEC can model a trap as a lumped RLC load placed at a specific segment — you specify the resistance, inductance, and capacitance of the trap at the operating frequency. The model will include the trap's effect on current distribution and resonance. The limitation is that the trap's frequency-dependent impedance must be entered for each simulation frequency — EZNEC does not automatically model the trap's full frequency response in a sweep. For detailed trap characterisation, measure the trap with a NanoVNA first and enter the measured impedance values into EZNEC.
How do I model an antenna over sloped ground?
EZNEC's standard ground model assumes flat horizontal ground at Z = 0. Sloped terrain is not directly modelled. The practical workaround is to model the antenna at its actual height above the ground directly below the feedpoint, using flat ground, and accept that the model is an approximation. For EZNEC Pro users, the terrain analysis can be extended using HFTA (HF Terrain Analysis) in conjunction with actual terrain data to model the effect of hills and valleys on the radiation pattern, particularly at low elevation angles relevant for DX.
Is EZNEC suitable for VHF and UHF antenna modelling?
Yes, with appropriate segmentation. At 144 MHz the wavelength is approximately 2 metres, so segments should be 2 to 20 cm long. A simple 2m dipole needs 20 to 30 segments per arm. A long Yagi for 2m EME with 20 elements could approach the Demo version's 500-segment limit, but most 2m practical antennas model comfortably within the Demo limits. The main difference from HF modelling is that element diameter has a proportionally larger effect at VHF due to the shorter wavelength, so entering the actual tubing diameter is even more important.
What is the difference between EZNEC and 4nec2?
Both are graphical front-ends for the NEC2 simulation engine and produce the same underlying simulation results. EZNEC has a cleaner, more polished Windows interface and is more beginner-friendly. 4nec2 is free, open-source, and includes a built-in optimiser that can automatically adjust element dimensions to meet specified performance targets — a significant advantage for Yagi design. 4nec2 also has more flexible scripting and variable support for parametric models. Serious antenna designers often use both — EZNEC for quick interactive modelling and 4nec2 for automated optimisation.
How do I save and share EZNEC models?
EZNEC saves models in its proprietary .EZ file format. Models can also be exported as NEC input files (.nec or .txt format) which are compatible with other NEC-based simulation tools including 4nec2, MMANA-GAL, and online NEC calculators. Many published antenna designs include NEC files that can be imported directly into EZNEC using File → NEC File → Import. The EZNEC antenna model library and community model archives on the internet contain hundreds of validated models for common amateur antenna designs that serve as useful starting points.