[Top] [All Lists]

[TowerTalk] Antenna Heights and EZNEC

To: <>
Subject: [TowerTalk] Antenna Heights and EZNEC
From: (David Robbins)
Date: Sun, 3 Feb 2002 16:02:22 -0000
> > Unfortunately in the real world you are not going to find a station
> > where you could take those measurements that is not encumbered with
> > stuff that would make your measurements questionable.  Things like
> > other towers, other antennas, power lines, buildings, hills, rivers,
> > the ocean, etc, etc, etc.  and even if you did find a reasonable
> > where someone had put a single 10m antenna at 200' with nothing
> > it how would you make the measurements?  Days of helicopter flights
> > and down and around and around?  Then prove the helicopter wasn't
> > disrupting the pattern you were trying to measure?
> >
> > Now, once you have made such a single set of measurements, what do
> > they prove?  You get a pattern for one antenna on one tower over one
> > specific topography.  The worst part is that the pattern you measure
> > would change when the ground froze or thawed, when another antenna
> > put on the tower, when antennas on the tower rotated relative to
> > other, when the neighbor moved his RV, etc, etc, etc, ad nauseum per
> > real world.
> So instead, we assume can use a model and get accurate results
> 100% of the time that apply to the same situation?
> Or am I missing something?
> 73, Tom W8JI

If you stay within the limitations of the model's methods, which for nec
have been pretty well documented, then you should be able to compare
models to each other.  The 'accuracy' of the comparison would depend on
how accurate your model inputs were... that is, if you model only the
antenna and not the tower you would be less 'accurate' than if you
included the tower.  Now how important that is to the decision you are
trying to make versus the added calculation time is something you would
have to decide based on your knowledge of the possible interactions and
the software.  

That is one great problem with the proliferation of relatively high
powered computers and the free availability of this very complex
software.  Very few users of this software really understand its
limitations, and I fear even fewer understand the complex interactions
between antennas and their surroundings well enough to know what is
important to model and what isn't.  As the developer of a program that
models lightning's effects on hv power lines (a task probably similar in
complexity to modeling antennas), I know that even if I could make the
program provide perfectly accurate results when compared to the real
world that once I give it to a user to analyze their power line their
results won't be perfectly accurate.  Why?  Because the user probably
won't have enough detailed information about their power line, the
ground under it, the tolerances of the installation, modifications from
the initial installation, etc... and most likely the user won't
understand the importance of certain inputs and will either ignore or
just approximate values that he thinks are ok.  In building antenna
models to try to model a real station there are many of the same

Lets say I am trying to model my 160m inverted L's.  I have 2 of them
100' high with a 30' top section 1/2 wave apart with 8 approx 1/4 wave
radials raised 10' off the ground under each of them.  Now, I can make a
model of that very quickly from that information and compare patterns
for them fed in phase and 180 degrees out of phase and get patterns that
look just like are seen in an antenna text book.  BUT, my antennas don't
perform like that.  I know the nulls off the side aren't as deep as
predicted... but why?  Well, it might be that there is a 120' tower
about 50' from one of the vertical sections and a 150' tower about 50'
from the other... it might also be that there are other guy wires, 2 or
3 other towers, a house, power lines, a pond, and all sorts of other
variables... I can spend weeks building that model, days of calculation
time, and find out that the side nulls aren't as deep as the bare bones
model predicts.  Or I can put it up based on the bare bones model and
see that it performs reasonably well and accept the difference from
prediction because I know that I didn't model all the possible
interactions.  If I am real curious I know I can add a couple of simple
towers to the model and see how that changes the pattern to verify what
my experience tells me, someone not as experienced may not know that a
particular tower height or guy wire length might be important.   Now
someone else may put up the same antenna based on that model and find it
works a quite differently because their surroundings are different, or
they aren't really 1/2 wave apart, or they couldn't raise the radials
off the ground, etc, and if they don't understand the implications of
the changes they have made they will say the software doesn't work,
where really they have gone outside the bounds of the model as defined.

This was also a problem with some of the early software that
automatically optimized yagi's (YagiMax was one of the first I tried
that did this).  If you blindly set up a yagi model and sent it off to
optimize it you would get a really SUPER result sometimes... and if you
blindly assumed the software was right and went to build it you would
find it didn't work.  What you had to do was apply some knowledge of the
real world and what the limitations of the program were so that you
could constrain the problem to come up with a buildable antenna... if
you did that it worked rather well... not 100% by any means, but much
better than trial and error.

I guess what I am trying to say is that the software models are great,
but you still have to know something about what you are modeling or you
won't know that the results are even reasonable.

David Robbins K1TTT
AR-Cluster node: 145.69MHz or telnet://

Where do you get ICE bandpass filters & beverage matching boxes?  The
same place that pays for the hosting of this list:  The eHam Store.
Order online at

FAQ on WWW:     
Administrative requests:

<Prev in Thread] Current Thread [Next in Thread>