----- Original Message -----
From: "Tom Rauch" <firstname.lastname@example.org>
To: "W7TMT" <email@example.com>; <firstname.lastname@example.org>
Sent: Saturday, October 23, 2004 5:53 AM
Subject: Re: [TowerTalk] Ground radials/elevated radials
> First, absolutely the least reliable model is one used when
> a wire is parallel with and close to earth. That's pretty
> well established.
I think you might find some dispute there, as far as the modeling code
itself goes. I would state it as:
The model results will be unreliable without a good description of the
earth. A uniform slab of earth (the usual NEC model) is probably not very
realistic for most ham installations.
The problem is the same as if you tried to model a dipole hanging from a
tower, and didn't bother to model the tower and guy wires.
> Second, most of the work available to us simply involves
> models. Almost no one actually built an antenna and measured
> FS with only a radial change. People use a model that is
> known to not be very reliable (with 5dB or more), and
> publish results to the tenth of a dB.
The important thing in modeling (in a practical sense) is not whether a
particular antenna matches a particular model to a gnat's eyelash. It's
whether a change in that particular antenna results in a corresponding
change in the model results. This allows the modeling process to be useful
for assessing potential changes, even it's not accurate in the absolute
However, one has to be careful about 0.1 dB precision. That would imply
that you've entered the geometry and the physical parameters to a
corresponding precision. Just for a rough guideline, a pair of dipoles, one
driven, one shorted, a half wavelength apart will show a gain variation (in
the far field) of 0.1 dB with a 1 degree change in relative orientation or a
change in spacing of 1%.
> > > 1) In your opinion where is the cutoff between "elevated
> > radials" and a "ground plane" system. How high off the
> > ground does the system need to be before it becomes a GP
> > rather than an "elevated radial" system?
"Many wavelengths" would be the trite and useless, but true answer. Another
common modeling guideline is that you need to model anything in the reactive
near field (2 pi wavelengths, by convention), however the fidelity of the
modeling can vary somewhat.
If you've got a wire within a half wavelength of the ground, it certainly
interacts with the ground. If you're modeling, an important question would
be how close to you have to be before some assumption in the modeling
process, i.e. uniform slab, or reflection coefficient approximation (non
Sommerfeld ground in NEC) is invalid for your situation. NEC, by default,
uses the reflection coefficient approximation for segments farther than 1
wavelength away, so that's probably a decent threshold for where the
"details" of the ground stop being important.
This is why it's important to use the Sommerfeld ground in your NEC models.
I would guess that most ham antennas are close enough to the ground that
it's significant (unless you've modeling your 10m Yagi on a 200 foot tower).
It used to be too time consuming to use the Sommerfeld ground, but with
modern PCs, the extra time is insignificant, and most of the NEC shell
programs around handle the icky details of firing up SOMNEC for each
You also need to actually measure your ground properties! Particularly the
epsilon. It doesn't have such a huge effect in the far field (i.e. where
the lobes are, etc. in the HFTA sense), but it has a big effect when the
ground is in the reactive near field.
> It won't be a cutoff, it'll always be a very gradual
> transition. The wires wouldn't really be divorced from earth
> below the antenna until the earth is much further away than
> the area of the groundplane. It's a geometry thing that
> depends on wavelength and occupied area.
I don't know that occupied area enters into it. Either the current in a
wire is influenced by the ground or it's not.
See: http://www.mscomputer.com for "Self Supporting Towers", "Wireless Weather
Stations", and lot's more. Call Toll Free, 1-800-333-9041 with any questions
and ask for Sherman, W2FLA.
TowerTalk mailing list