At 04:01 PM 9/3/2005, Mike Harris wrote:
>G'day all,
>
>I'm wondering how the "effective diameter" of element to boom mounting
>plates is calculated.
Empirically.. that is, someone goes out and fiddles with a model of the
antenna using round elements until its behavior matches that of the actual
beast with the flat rectangular plate.
Likewise, one can convert a stepped diameter into an equivalent single
diameter element of a different length. This has been done both
experimentally and with a fair amount of analytical rigor, back in the days
before various finite element modeling techniques (e.g. NEC) became popular
with ever dropping computer costs.
For instance, there's a whole body of yagi design that uses mutual
impedance coupling analysis between the elements. For instance, you'd work
with a 4x4 admittance matrix on a 4 element antenna. This is just simple
enough to be tractable to calculate by pencil and paper or with a
calculator or low end computer. You'd use expressions from papers by King,
for instance, to get the various Zs, and go from there.
Since the basic equations that are used assume constant diameter elements,
you'd need some way to turn a real element with mounting plates and steps
into an equivalent one. And that equivalent would have to be "close enough"
that the rest of the design process would work.
For straightforward designs, this works quite well. As the designs get
more exotic (lots of elements, radical taper schedules, elements very close
together), the approximations don't always work as well.
These days, with reasonably fast computers, you can just model the whole
thing, warts and all. If you're ambitious, you can even model the bolts
and nuts.
I'd say that the transition from "algorithmic" methods to "finite element"
methods for amateur antennas probably really occurred around the early
90s. MiniNEC, NEC, etc., have all been around longer, but their
limitations, and all around clunkyness of use, as well as the tedium of
running a big model (if you spent the time to create that 1000 segment NEC
file) made them a pain to run. And it's not til you get to those many
hundred segment models on touchy designs that NEC starts to work a lot
better than the older analytical approaches. Even today, a 2000 segment
model takes several minutes to run on a 2GHz processor. Back in 1991, when
a 66 MHz 486 DX-2 was hot stuff, this would have taken hours, assuming you
even had enough RAM to hold the model in memory (a 2000 segment model needs
a 2000x2000 matrix of complex numbers. In double precision that's 128 MB...
Phar lap extenders notwithstanding, it was still a chore)
Even today, with all the supercomputers and fancy software (like HFSS)
available, a lot of designs start out with rule of thumb kinds of
dimensions. Then, you build your model in the FEM program, run the
patterns and impedance curves, and start adjusting (or let the optimizer
beat on it).
-- The upshot is that if you're worried about small changes like this, get
one of the many nice free NEC programs out there (I like 4nec2), model it
up, and see.
_______________________________________________
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
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|