There is a great difference with a model that allows to design an
antenna, a model for predicting propagation or a model that tries to
show as an antenna will perform over a media.
When is sure that one can define a parameter, i.e. a certain height
from ground or a certain number and lenght of radials, then it's
perfectly possible to trust results, knowing that those results are
true in that specifical condition.
In other words, I can design with optimal results and precision a
complex antenna in free space or an horizontal array like a yagi over a
real ground if this ground is far enough to allow me to neglect its
effect within inherent antenna performances in term of elements
On the other end I must be aware that when parameters are simplyfied
and assumed instead of beeing real and over full control, ground
reflections and morfology, what I get is a result that is valid within
the assumption I made but not necessarily with the real world, if this
differs from the model.
> >The REASON we have models is the actual system is too complex to
> >consider. While the models are certainly very good in some aspects,
> >they always miss some things.
> This is certainly true, but there is another good reason, and that is
> the sheer magnitude of the task required to accurately measure a pheno
> is impractical.
> I remember that back in the 60's and 70's VOA used to employ people wh
> job involved monitoring VOA transmitters' signals at various locations
> around the globe. Expensive, if you're going to compile a large-
> data set to be more than anecdotal. They adopted VOACAP as their stan
> for propagation analysis and transmitter siting in the 80's, presumabl
> after satisfying themselves that its results were sufficiently consist
> with the empirical data.
> I've always wondered, though, how they made their decisions on their b
> installations, since these would be near or below the threshhold of
> accuracy of VOACAP.
> One final comment, on the utility to giving NEC model results to two
> decimal places. For those of us who are still learning how to use the
> programs effectively, it can be useful to have these numbers, even tho
> we know the experimental accuracy may be in the +/- 1 dB range. For
> example, one important source of error in these models is not using en
> segments to get convergence. If I look at Bill's results, and they ar
> the same as mine, then I want to know why, and in the process I can le
> 73, Pete N4ZR