G3PQA wrote:
> I would appreciate advice from Group which earth is recommended for
> best accuracy when modelling an inverted L. Real-mininec-earth or real
> high-accuracy-earth? Am getting around 2dB difference between the two
> methods (+1 and +3dBi).
> Also, should a loss resistance be put in series with real high
> accuracy model radials to simulate earth impedance?
>
>
Modeling shunt fed towers (if the shunt wire is included) and inverted
Ls are two very difficult antennas to model in EZNEC (if you want an
accurate answer) because you run up against rule limitations.
With an inverted L for 160, invariably you end up with a low horizontal
wire (compared to the wavelength). That is a rule breaker when you use
a Mininec ground. If you do this, it usually results in a gain that is
too high and a feedpoint impedance that is too low (usually but not
always). The longer the horizontal wire, the higher the error. So you
have to use a real ground (Sommerfield-Norton). That means you have to
model the radials. Modeling radials is also tricky. Since you have to
put them above ground with NEC2, they will show resonance properties
(unlike radials in the ground). That means that if you model the exact
radial system you have and that radial system only has a few radials, or
radials that are not a quarter wavelength, you are likely to get some
really strange results. The height of these radials above ground will
also affect the antenna gain and feedpoint impedance, so you have to
determine the proper height for the radials. I also think EZNEC under
estimates near field ground loss for low wires. That is an opinion
shared by several persons whom I would consider to be experts.
I will give you my method of reducing some of these errors (even though
most people won't do it because it is too much trouble). My solution is
to model a very good radial system that has insignificant loss. That
requires many radials and some of different lengths to eliminate the
resonant effects. You can determine when you have arrived at a good
radial model by using a vertical and comparing the response to a Mininec
ground (which has no near field ground loss). You will discover EZNEC
arrives at an almost lossless radial system long before most tables for
radial ground loss do. NEC2 thinks that 16 low radials is within 0.03
dB of perfect, however I use more than that to eliminate the
resonances. Then I go to that table of equivalent ground loss
resistance for the radial system I want to model. I then insert this
resistance at the base of the vertical. There is another problem here.
Those tables were generated for quarter wavelength verticals in the
broadcast band. That may be close enough for a resonant inverted L, but
it may not be accurate for an inverted L longer than a quarter
wavelength or for frequencies far removed from the broadcast band. You
can then determine the proper height for the radial system by using a
vertical and comparing the response between the radial system at various
heights and a Mininec ground. Then go back to the L model after you
have determined a proper height for the radials.
So you see it isn't easy when you try to use work-arounds for NEC2
limitations. NEC4 also has some limitations but most people don't have
that. Using the method above I have been able to accurately calculate
the feedpoint impedance for inverted Ls, but I have no way to validate
the gain. It does yield gain numbers that are CLOSER to measured data
generated by others, than when not doing these work-arounds. That's
mainly because you will be replacing EZNEC's calculated near field
ground loss with measured ground loss. So I really don't know how
accurate the gain numbers are, only that they are closer to real.
Be sure and always do the test for simulation errors using the "Average
Gain" number. There are a few other minor problems, but the stuff above
should keep you busy for a while.
I don't think you will find any references to the techniques listed
above since I made it up. If you do find any please let me know.
Comments appreciated.
Jerry, K4SAV
_______________________________________________
Topband mailing list
Topband@contesting.com
http://lists.contesting.com/mailman/listinfo/topband
|