[TowerTalk] Funniest thing I've seen in weeks

[Jim Lux]

At 04:36 PM 7/1/2004 -0400, Tom Rauch wrote:

> > the elements.  For instance, a log periodic (which has all
>elements driven)
> > doesn't really care which end it's driven at (aside from
>some matching issues).
>
>The log periodic uses crossfire phasing, and that means it
>must always fire towards the feedpoint.
>
>There is no way around that, so the statement a "log doesn't
>care which end is fed" is absolutely not correct in an
>aperiodic design.

Hmmm. I seem to recall seeing a log periodic fed backwards, but I'll have 
to take a look for the design and see if they were fooling with oddball 
transmission lines or loading.




>As for the Yagi, optimum pattern and gain require elements
>be excited in a binomial current distribution. While you
>could feed the front and use two rearward parasitics in a
>three element antenna (the middle in NO way would be tuned
>accurately as a reflector, or the furthest to the rear would
>have almost no current), by far the optimum feedpoint is the
>center which will produce a nearly perfect 1:2:1 current
>distribution and a null of at least 40dB with proper tuning
>and spacing.

The binomial case is just one aperture weighting that happens to be simple 
(both in analysis and in implementation). It's not necessarily optimum, 
depending on what you want the main and sidelobes to look like. Any sort of 
tapered illumination will tend to broaden the main lobe and reduce the 
sidelobes.  If one wanted a real narrow main lobe, and didn't care so much 
about side/back lobes, you could go with a reverse taper that has higher 
currents at the ends.  Especially if you have lots of elements (as in a 
long boom design), one could in theory, construct any sort of 
illumination/current taper you wanted (such as a Chebysheff or Taylor), 
although, I think you might have a problem with realizibility or bandwidth 
(you don't have total freedom with mutual impedances... an interesting 
synthesis problem... good topic for a thesis project).  And of course, it's 
probably not worthwhile.  Nobody is obsessing about 30 dB sidelobes for an 
amateur antenna.

The depth of null is almost entirely going to be determined by design and 
construction tolerances.  40 dB nulls imply controlling the current to 
around 1%, which would be quite challenging, especially over frequency. 
I'll bet the wind induced motion in the elements would change the coupling 
enough to cause problems.  As it happens, I have a paper I'm working on 
that addresses just this problem (i.e. how much does the mutual coupling, 
and hence the pattern, vary as a function of small displacements in the 
elements): For parallel dipoles  spaced 0.2 wavelength apart(one driven, 
one shorted), a 1% change in spacing changes the phase of the radiated 
field by about half a degree. For small angular changes in dipoles half a 
wavelength apart, 1 degree results in a phase change of about half a 
degree. That half a degree phase change corresponds to about a 40-45 dB 
null.  Normal element flexing in an antenna is probably on the order of 
several percent distance wise, and probably 4 or 5 degrees.  I don't recall 
the exact numbers for such large displacements (it's not important in my 
application) but it's probably roughly linear, so you're looking at phase 
shifts of 5-10 degrees.

Of course, this is what makes the SteppIR so cool, it makes getting 30 - 40 
dB nulls actually possible over a wide bandwidth (just not all frequencies 
at the same time). (if only the wind didn't blow<grin>)





>The idea of adding a reflector behind a reflector is a waste
>of time, because there is essentially no field there to work
>with. This is why Yagi's have developed into multiple
>directors with single reflectors. Even trigonal reflectors
>(staggered above and below the boom) have largely proven to
>be a waste of material.

Why would you say there's no field behind the reflector?  True, in the far 
field where the fields from all the elements have cancelled, but certainly 
not in the near field, and particularly not in a superdirective array.  Try 
running a NEC model on a Yagi and looking at the H field (which is where 
the energy is stored).

For instance, on a 3 element 20m beam, with 5.37, 5.31, and 4.70 meter 
elements, spaced at 2.72 and 3.89m, the H field has three distinct peaks at 
the element positions, but the field extends about 1.5 meters off either 
end before it decays to less than 10% of the peak at the driven 
element.  This is particularly noticeable if you do a cut at, say, 0.25 m 
above the plane of the elements.  Indeed, the fact that it decays that much 
implies that this particular antenna might not benefit much from an added 
element on either end, especially if it's spaced out comparable to the 
other two.



>Of course if you DON'T properly tune an element for null
>then adding a second reflector can help.

I don't think that adding a reflector (or director, for that matter) to an 
existing optimized design will have all that big an effect (at least not a 
good one!).  You'd have to optimize the whole shooting match, and you'd 
probably want the elements pretty closely spaced.  You'd probably get into 
a situation where the coupled currents are so high that loss becomes an 
issue, and, in any case, it would probably be horribly narrow band.



>I think the confusion in all of this comes from people
>thinking reflector reflect and directors direct. That just
>isn't what they do. What all of the elements do is cancel
>radiation towards the undesired direction. The minute a
>director starts "directing" as a primary gain mechanism it
>barely affect pattern. Patterns are formed by squeezing
>radiation by forming nulls, like squeezing a balloon.
>Stretching the pattern forward is a very inefficient way to
>obtain gain.
>
I would agree, although, superdirective arrays work by stretching to get 
gain, but have limits on how far you can take it.



>Traveling wave antennas, like the Fishbone, Beverage,
>Inverted V, and V Beam, use in-phase addition more than
>nulling to achieve directivity. That's why they require huge
>amounts of linear area for minimal gain.  I can't imaging
>why anyone would want to mess up a Yagi by making it work as
>a traveling wave antenna, unless they pound for pound had
>more element and boom material than brain material. Hi hi.

If they needed very broad band characteristics.  Of course, you would 
probably be better off with a log periodic or something else.