Jim Lux wrote:
Roy Lewallen wrote:

Absolutely true. And it can't handle things like patch antennas or
antennas printed on a PCB.

NEC does OK at microstrip patches with air dielectric (or foam with very
low permittivity). I've used it to model an array of 9 patches and the
port to port coupling calculated by NEC and measured by a VNA were
pretty close (within measurement uncertainty).

Yes, I meant patch antennas with common dielectrics, which are far more
common. It's of course the dielectric that NEC can't account for. I've
designed quite a few antennas on PCB material, but use a fudge factor
based on comparison between measured and EZNEC results of a simple
antenna near the same frequency. This gets me pretty close, but even
this approach wouldn't be adequate if field coupling through the
dielectric is significant.

It's pretty darn slow at this, though (lots and lots of wires in each
patch), I used lumped loads for the matching network model (capacitive
probe feed)

For large models, calculation time goes up as the cube of the number of
segments, so big models can get slow all right. However, EZNEC has
undergone a pretty dramatic speed improvement over time as various code
substitutions and updated compilers have been used, and it's much, much
faster than older NEC compilations. And some versions of NEC have been
similarly updated, so people using different NEC compilations can
experience pretty different calculation speeds.

Roy Lewallen, W7EL