On 2/7/18 2:06 PM, Shawn Donley wrote:
I believe the article certainly is accurate with respect to trends. The NEC
tree models track well with the infinite lossy cylinder closed form solution,
which is a good double check.
But the real world, at least the one around my QTH, doesn't have a single tree
to consider but a forest of trees. One of my wire antennas is a full wave
loop in a vertical plane fed from a lower corner with ladder line. The top
and bottom runs are 100 feet and the vertical sections are 40 ft. The top is
about at 65 feet. Running EZNEC, I saw that I could get a lower takeoff angle
on 40M if I fed the middle of one of the vertical sides. But then the
polarization is primarily vertical, and those trees (hardwoods) are about 15 ft
away. I may try it anyway just to see if I can notice any difference.
This is where looking at "forest as dielectric slab" model might help. I
was just discussing this with Jan WB6VRN at work, where his observation
was that a large vertical loop in a forest worked terribly, even though
clear of immediate trees.
There's sort of two things to model in this situation. The first one is
"near field loss/detuning" - it would be easy to add a few more trees to
the QST model (NEC even makes it easy with the GM card). What I would
do is look at the loss (total power radiated in all directions before
and after the trees, and adjusting a matching network).
Then, I'd assume my antenna is a infinitesimal dipole and look at a slab
model with dielectric boundaries between soil, trees, and air - this is
more like what HFTA does, or even better, what ionosphere propagation
models do.
I've not looked at Tamir, et al's or Cavalcante, et al's papers
recently, but they might have a simple analytical model for this.
Otherwise, it's actually a straight forward ray tracing
(straightforward, not trivial, or I'd just write it out here). You might
be able to adapt a program for looking at UHF/VHF ducting, but most of
those seem only to deal with the change in refractive index, not loss.
For the Yagi-in-the-trees, I have that situation too. My Optibeam OB9-5 is at
70 ft but the trees are 80-100 ft high. I have 20+ feet clearance to the
nearest branch, but some of those branches approach a horizontal orientation
coming off the trunk. So like the vertical wire near the tree trunk, I suspect
there may be some tree losses.
Add those branches to the QST model - clever use of the GM card to
rotate, shift, and scale the trunk model is one way.
This is interesting stuff and not easy at all to model completely. Another
reference for measurements is at
http://nvlpubs.nist.gov/nistpubs/jres/68D/jresv68Dn8p903_A1b.pdf
But the measurements were in a jungle environment (most of this stuff was back
in the Vietnam days).
Yes, but the propagation model still works, if you plug in the
dielectric properties of temperate or boreal forest as needed.
Ultimately, in the far field, you model it as a mixed media dielectric
(i.e. ignore scattering and reradation) -> if you've got 10% by volume
of something with epsilon 80 and sigma 1 S/m, and the rest is air, it's
basically epsilon 8 and sigma 0.1 S/m
If someone's interested, I think I have some Matlab code to generate a
random forest based on some published distributions of tree diameters
and heights. I was going to use that to generate lots of lossy wires in
NEC.
There might also be some actual measurements made in the last 20 years -
there's a sort of perpetual low level interest in propagation through
forests and vegetation. yeah, the Foliage Penetration (FOPEN) radars
were part of it, but comm is also of interest.
Or rope some grad student into doing it for a Masters thesis
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|