Antenna modeling
| Antenna modeling | |
|---|---|
| Purpose | Predicting an antenna's pattern, gain and feed-point impedance by computer |
| Method | Method of moments, with wires divided into segments |
| Main engines | NEC-2 (public domain), NEC-4 (licensed), MININEC |
| Popular programs | EZNEC, 4nec2, MMANA-GAL, xnec2c |
| In practice | |
| Cost | Capable free programs are available |
| Main inputs | Wire geometry, sources, loads and ground |
| Validity check | Average gain test |
| Main limits | Real ground, nearby objects and feed-line currents |
Antenna modeling is the use of computer software to predict how an antenna will behave before, or instead of, building it. The builder describes the antenna as a set of straight wires, says where the transmitter connects, and the program calculates the current on every part of the antenna. From those currents it works out the radiation pattern, the gain, the front-to-back ratio and the feed-point impedance and SWR across a range of frequencies. Modeling has changed amateur antenna design: dimensions that once took weeks of cut-and-try can be refined in an evening, and claims about an antenna can be checked against a model instead of an advertisement.
How it works
Most amateur modeling programs use the method of moments. Each wire is divided into short segments, and the program solves for the current on each segment, taking into account how every segment affects every other one. A source marks the segment where power is applied, and loads can represent resistors, coils, capacitors or traps. The program can also include a ground under the antenna, described by its conductivity and dielectric constant.
Segmentation matters. Too few segments give inaccurate results; a common guideline is at least ten segments per half-wavelength of wire, and more where wires join or currents change quickly. Segments should not be much shorter than a few times the wire's diameter, and segments meeting at a junction should be similar in length.
Software
The best-known engine is the Numerical Electromagnetics Code (NEC), developed at Lawrence Livermore National Laboratory for the US Navy. NEC-2, released in 1981, is in the public domain and sits underneath most amateur programs. NEC-4 handles some difficult cases better, including wires buried in or very close to the ground and elements made of tubing of different diameters, but it is distributed under license. MININEC, a simpler method originally written for early personal computers, is used in some programs.
- EZNEC, written by Roy Lewallen, W7EL, was for many years the standard commercial program for hams. It uses NEC-2, with versions that can use NEC-4, and in 2022 it was made free to download.
- 4nec2, by Arie Voors, is a free Windows program built on NEC-2, with tools for sweeping frequency and optimizing dimensions automatically.
- MMANA-GAL is a free program based on MININEC. It is easy to learn and popular for Yagis and wire antennas well above ground.
- xnec2c is a NEC-2 program for Linux.
What a model tells you
A model shows the azimuth pattern, seen from above, and the elevation pattern, which shows the angles at which the antenna radiates toward the horizon or the sky. It gives gain in dBi, the feed-point impedance and SWR across a band, and the current along each wire. That makes it possible to compare antennas fairly, see the effect of height above ground, design matching networks and optimize a Yagi for gain or front-to-back ratio. Modeling can also support an RF exposure evaluation by estimating field strengths near the antenna.
Limits
A model is only as good as its description of the real installation.
- Ground. Real soil varies with depth, moisture and location, and its true values are rarely known. MININEC programs calculate impedance as if over a perfect ground, so they are less reliable for low antennas, and NEC-2 cannot properly model buried radials.
- Nearby objects. Houses, gutters, wiring, towers, trees and other antennas all affect a real antenna, and they are seldom in the model.
- Feed lines. Unless the feed line is modeled, the program assumes no current flows on the outside of the coax, the problem a common-mode choke is meant to solve.
- Components. Losses in coils, traps, ferrite and baluns can be approximated with loads but not fully modeled.
A sensible approach is to start with a simple antenna whose behavior is known, such as a dipole, and compare the model with measurements from an antenna analyzer once the real antenna is up. Differences usually point to something in the surroundings that the model leaves out.