Buck wrote:

Thanks for the help, Wes. I will be looking for that additional
thread. One of these days I'll get rich and famous and I can budget
the cost of EZNEC and get the full copy. 20 elements is just not
quite what I need for an antenna I am toying with in my mind.

BTW, how closely does the program get the measurements for antennas?
If I plot a dipole and EZNEC says the length should be 45 feet, will I
need to trim the antenna to a different length or will it be within an
inch or two?


I sure encourage Wes and anyone else to post his experience with
agreement or disagreement between modeling and measurement. But I've
gotten a pretty good overview from the comments I've gotten in 15 years
of selling EZNEC and its predecessor. These include reports from a
spectrum of users ranging from amateurs who've measured what they can
with the tools they have available, to professional users who have
exceptionally good measurement equipment and access to high quality test
ranges.

With quite a handful of caveats, below, the program gives very accurate
results(*). Here are the caveats.

First, the antenna has to be just like the model. By "just like", I mean
that anything materially affecting the antenna has to be in the model.
Radiating feedlines are one of the most common elements overlooked by
amateurs. I've also seen cases where the test meter itself is large and
close enough to the antenna to change its impedance very significantly,
and correspondence wasn't achieved until the meter was added to the
model. Also, most people don't appreciate how much effect another
antenna (or other resonant conductor) can have, even when surprisingly
far away.

Next are a handful of modeling traps. These include but aren't limited
to making sure the model is within modeling guidelines (segmentation,
choice of ground type, etc.), that the NEC engine isn't having any
particular difficulty (average gain close to 1), and the model doesn't
contain any features which are known to cause an error (e.g., connected
wires of different diameters, particularly in a parsitic element).
Serious geometry errors such as connecting a wire end to a segment
midpoint on another wire appeared quite frequently in files sent to me
by users, which prompted me to expend the considerable effort to develop
a comprehensive geometry check for the EZNEC v. 4.0 release. But many
errors can still be made. It takes a fair amount of care, knowledge, and
experience to gain a high degree of confidence that all these conditions
are met.

Then, even when the model is correct, most people don't know just how
difficult it is to make good measurements. Measurements made through
feedline are subject to a host of errors including failure to account
for the effect of even a small amount of loss, and line impedance that's
different from its specification -- a very common situation.
Measurements made without feedline often suffer from the effects of
common mode current and the close proximity of the person making the
measurements. And so forth.

All that being said, it's really not hard to make a very decent model of
a simple antenna -- or even a fairly complex one. But there are too many
things I can't know about your modeling ability, how well you've
suppressed feedline current in the real antenna, and how you'll be
deciding that the antenna "works" to know if it'll come out within an
inch or two.

Based on past experience, some number of the readers will interpret what
I've written to mean that it's impossible to make a good model, or that
modeling doesn't give accurate results. Neither is true at all. But a
careless model and casual measurement probably won't match very well,
either.

So my advice is to cut the antenna a bit long. Only after *you* have
built and measured a few can you gain (or not gain) confidence in the
ability of *your* models to match *your* measurements.

(*) I can't and don't take credit for the accuracy of the program.
That's solely due to its NEC calculating engine which was developed by
others. I will take blame for any EZNEC results which differ from those
given by NEC. Those would be bugs, and I'd appreciate very much if
anyone seeing a difference would contact me so I can track it down and
fix it.

Roy Lewallen, W7EL