[TowerTalk] Dayton 2004 Antenna Forum Papers now on the Web!

[Richard (Rick) Karlquist (N6RK)]

 One of the least accurate uses of any
> model is simulating loss with a wire close to earth.
> 
> An early comparison of NEC-2 models to actual FS
> measurements of a low dipole someplace under .01WL high (as
> I recall). Haggn-Barker did the study in Thailand, and VE2CV
> Belrose did the modeling showed about 5dB error. My Beverage
> antennas also show significantly more measured current taper
> with distance compared to a model.

Gerry Burke (of NEC fame) told me that some people had tried
to measure ground characteristics by running transmission lines
at low heights above the ground.  I don't know the exact
details of these tests, but they sounded like they were similar
to measuring Beverage current taper.  As with your beverage,
the current taper was surprisingly high.  What was really interesting
is that there was no set of numbers that could be plugged into
NEC that would produce a model in agreement with the measurements.
What NEC shows for ground loss is reaches a peak value (as
the parameters are varied) which is less than the measured value.
Away from this peak value, as you move towards a pure dielectric
or a pure conductor, the loss predicted is less than the peak, which
make sense.

My strategy is to monitor the drive impedance at the base of
the vertical and stop adding radials when the impedance nears
its asymptote.  In that way, I have correctly modeled my
own ground, whatever it's characteristics.  For my ground,
60 radials is about the useful limit.  As always, YMMV. 

Rick N6RK