Owen Duffy wrote:

Roy,

Aren't limitations on segment size an issue, especially with small diameter
coils?

Although NEC has recommended limits on both minimum and maximum segment
length, my experience is that the minimum can often be disregarded with
no apparent degradation of the result. A good test is a small
single-turn loop, for which the theoretical result is well known. There
is a limit to NEC's ability to model a single turn loop for reasons
other than segment length, but the result holds very well for very short
segments. A more important criterion is the segment length to wire
diameter (or radius) ratio. Results begin getting poor when the segment
starts looking like a coin, that is, with a segment length considerably
smaller than the wire diameter. EZNEC spots these conditions and issues
appropriate warnings via its guideline check. Most errors caused by
segmentation problems show up in a poor average gain.

Can NEC model a typical HF helically loaded mobile whip reasonably, sy 16mm
diameter, 1.2m long with variable pitch progressing to close spaced turns
on the top 30% or so?

There's no provision in either NEC or EZNEC to automatically generate a
helix with variable pitch, but you could make a stepped approximation
with several tandem helices. However, if by "close spaced turns" you
mean wires which touch or nearly touch except for the insulation, this
violates the rule I stated earlier, that the distance between outsides
of wires should be at least one wire diameter, and preferably several.
So the coil as you describe it can't be accurately modeled. As long at
the rule is followed, though, NEC or EZNEC should be able to do a good
job of modeling a fairly typical whip loading coil.

I suspect NEC is not a magic bullet either.

Because EZNEC uses NEC for calculations, they both have nearly the same
limitations. I say "nearly" because EZNEC does have some added
protection against very small or large intermediate values in some
calculations, fixes for a few special cases, and minor changes from NEC
for various reasons. Most users are unlikely to see the effects of any
of these, however.

No modeling program, regardless of what's being modeled or how, is a
"magic bullet". Every last one has limitations and requires knowledge,
skill, and often a bit of art to utilize effectively and without a good
probability of serious error. I spend a lot of time and effort here, via
customer support, in the EZNEC manual, and in other venues, bringing
attention to EZNEC's limitations in the hope that it will save users
from getting results which are thought to be valid but aren't.

Side note: I had forgotten that at least one of the copies of NEC-2 I
have includes a GH 'card' provision, so the modification is apparently
quite common. It's not mentioned in the NEC-2 manual I have, however.

Roy Lewallen, W7EL