Hi Rick,
My point was we should be more careful in saying something is "accurate", or
that some entirely unrelated trend "verifies" a entirely different
measurement parameter.
It certainly is not that the NEC engine is flawed, but rather that there are
two problems:
1.) Soil is not homogeneous. The model assumes it is.
2.) Users almost always have no idea what the field strength is, what the
average soil conditions are, and often don't even add other structures to
the model.
> When we straightened out all this mess, we were able to generate NEC-3
model
> electric field magnitudes that were within 1.5 dB (some
1.5dB very good. In an off-list discussion with Pete I pointed out that BC
proof measurements often have 3dB of scatter from multiple effects
(including some or all of the effects you pointed out).
There were two primary areas of concern:
1.) To the best of my knowledge, there is no verification of accuracy of
buried wires in this application at lower HF or MF. Even if there was, it is
virtually impossible that anyone would have uniform homogeneous soil or have
any idea at all how to measure the soil characteristics.
2.) Most NEC interfaces we use are at best unverified, and at worse
seriously flawed when it comes to calculating things like low angle field
strength. Yet we regularly "pick" an antenna based on the most unreliable
part of the model. In example, we might choose an antenna with a TOA of 20
degrees over one with a 40 degree TOA when the 40 degree TOA antenna could
actually provide more absolute FS at the desired antenna.
There is a predictable EFFECT that as ground conductivity improves FS at low
angles from a vertically polarized source improves. It is a significant
stretch of the imagination to assume we know how to design a system that
would be a few dB better than another based on a model (unverified) when the
very parameter changed is the least accurate part of the entire modeling
process, and we don't have any idea of what data to input in the first
place!
The person quoted (W4RNL) doing the modeling very clearly warns it is only a
comparison of one model to another model and the models do not in any form
represent real conditions. He, in essence, warns of all the things we are
discussing.
73 Tom
|