Models are very good in relative comparisons, with some very important caveats.
1) Hold everything possible constant between the two models, EXCEPT
the singular issue(s) you are trying to trend. Don't try to vary
"dirt." If you don't already have burn scars and embarrassing gaffes
about dirt, you likely won't have the experience or the burn scar gut
motivation to guide you through this mine field.
2) Dirt is not handled well by any model, and each model has
techniques for getting reasonable results in spite of dirt. If the
comparison between two models is distance from dirt, or varying the
kind of dirt, then other than just noting a general "trend", one must
take any absolute result with a fair measure of suspicion. The way to
avoid dirt as an issue is to pick dirt characteristics and technique
for representing dirt, and then hold those CONSTANT everywhere in the
suite of comparisons. The comparisons will be good, but the absolute
values are suspect. EZNEC, in particular, has help sections on how to
deal with dirt, which are required reading if you are trying to do
something serious that will cost you time and money.
3) Every model has combinations of model input that will cause the
model to compute incorrectly. Even the mighty NEC-4 has these. If
you have one of these "gotcha's" in your data, it can queer the
outcome, even between two models with very little difference between
them. You need to know these for your modeling program and rigorously
avoid them.
4) Get a measured actual data point from a test case for low band
wire antenna stuff. Account for the difference between reality and
your model of that reality before you venture on. If you can't
reconcile the two, it's time to bulk up your understanding of what is
going on and get some experienced help to go forward. In my last
design for immediate construction and contest use, the data points vs.
modeled proved that height and droop angle of the elements were
critical, and the antenna would fail if height and droop angle were
not maintained to design specs for the life of the antenna. More
important, seeing the results the owner was convinced as well. This
resulted in some on the spot adjustments to design so those height and
droop specs COULD be maintained reasonably on his property.
5) If you are serious about an antenna in a certain place, be sure to
model ALL the conductors literally. Especially on 80 and 160.
Particularly if the antenna is supported by a tower. Everything
within a wavelength is in play to some degree. Comparing two antennas
where the real ones are on different sites with different sets of
miscellaneous conductors and dirt, probably can't be done adequately
with models.
The best thing to do with modeling is to get started and stick with
it. Personally I can't imagine doing antennas without the programs.
Like trying to drive a car blind. People have their favorite logical
simplification devices to describe what is going on in antennas. Most
of these break down if you use those mental devices to actually design
something. The models simply show you what is going on between wires
due to the physics. Even if you are right when reality and a model
differ, you need to know why and how.
Some here advocate the throw it up and be happy school of antenna
design, and it IS a hobby. Your money and your time, hope you enjoy
yourself, honestly. But if you've gotten beyond that, and want to
COMPETE with other hams, so pieces of dB's here and there add up to
your advantage, or you're just not working anyone and don't get it,
running blind without the model won't cut it. And there is lots of
help available, if you want it.
73, Guy.
On Thu, Nov 24, 2011 at 11:44 AM, Rik van Riel <riel@surriel.com> wrote:
> On 11/23/2011 05:49 PM, Gene Smar wrote:
>> Gents:
>>
>> A man much wiser than I once told me, "The difference between theory
>> and practice in theory is less than the difference between theory and
>> practice in practice."
> I suspect that while models may be somewhat inaccurate
> sometimes (especially due to lack of things like trees and
> houses in the models), the modeling software will still give
> a good picture of the relative quality between two antennas
> in the same installation.
>
> For example, the performance of antennas in my yard is
> likely to be off due to the presence of things like trees,
> my house, and the neighbors' houses.
>
> This could cause the model to be off by several dB.
>
> However, I suspect that the models of various antennas
> are likely to each be off by similar amounts in the same
> direction.
>
> In other words, if NEC tells me that one antenna has a
> few more dB gain than another, this is likely to be true,
> even if it gets the gain for both antennas slightly wrong.
>
> Am I wrong in my thinking?
>
> Has anyone observed something contrary?
> _______________________________________________
> UR RST IS ... ... ..9 QSB QSB - hw? BK
>
_______________________________________________
UR RST IS ... ... ..9 QSB QSB - hw? BK
|