dold@92.usenet.us.com hath wroth:
>Mesh and solid are the same, as long as the mesh is small enough for the
>frequency that you are working with. The freeantennas page suggests a 1/4"
>mesh. The mesh should be closer than 1/10 of a wavelength, or almost 1/2"
>for 802.11b.
I've been wondering where the 1/10th wavelength rule of thumb came
from. So, I just wasted the last 30 minutes Googling for references
and another 30 minutes digging through my text books. It's all over
the place, but never really explained. As near as I can guess(tm) it
came from the radio telescope designers, that needed a magic tolerance
for the smoothness of the parabolic dish (i.e. deviation from perfect
parabola). Same with optical telescopes, where the reflector has to
be within 1/10th of the wavelength of light to be usable. Why 1/10th?
I dunno.
However, that's flatness, not grid spacing. My guess(tm) is that the
flatness specification magically morphed into a grid spacing
specification, with little effort.
Next, I stumbled over to the junk pile, evicted the spiders, and
measured the spacing on various welded wire dishes (mostly
PacWireless). 2.4cm center to center, although it did vary somewhat
from dish to dish and across the surface of the larger dishes.
2.4cm / 12.5cm = 0.19 = 20%
So much for the 10% rule-of-thumb.
So, I dug out the antenna muddling program (4NEC2) and proceeded to
fumble around with a simple parabola with various grid spacing. I
didn't see a visible change to the gain pattern or phase distribution
until the grid spacing was beyond about 0.6 wavelength. Even then, it
was minor until about 0.8 wavelengths.
Next, I asked myself if it was better to have a very tight mesh (small
grid spacing) but with high irregularity, such as found with trying to
get welded wire mesh to conform to a parabola, or if was better to
have wide spacing, but a better approximation of a parabola using much
fewer wires. As expected, the dense, but irregular tight mesh had
lots of strange lobes in strange places, but they were minor and did
not affect the main lobe in the slightest. In other words, they're
the same. (I'll guess I'll have to stop complaining about wrinkle
finish aluminum foil reflectors.)
Going to the sacred online book of microwave antennas at:
<http://www.w1ghz.org/antbook/contents.htm>
I find that focal length accuracy is the most critical dimension, with
feed illumination angle a close second. I won't delve on the details,
but I should point out that these are two items that are being
literally ignored by the typical parabolic reflector that uses an
omnidirectional antenna (or a USB dongle) as a feed.
So what's the problem? Well, there is none. The errors introduced by
focal length inaccuracy (i.e. sloppy dish construction) are perhaps
1-2dB at worst. That's fatal for a satellite dish or high reliability
terrestrial link, but perfectly acceptable for a typical Wi-Fi system.
Feed illumination is a more serious problem, as much of the RF
generated by the omni antenna (or USB dongle) goes drifting off in a
useless direction. My guess was about 5-6dB gain loss, but only in
the transmit direction. In receive, all of the energy that hits the
dish, also hits the feed antenna. So, it's asymmetrical, but can be
compensated for by cranking up the tx power.
Conclusion: Use 0.2 grid spacing (about 2.5cm) for now. Don't worry
too much about precision. Try to build a proper matched feed for the
dish, but if that fails, the omni or USB dongle will still sorta
function. Any manner of gain is better than the stock rubber ducky
antenna.
Let's do the math. Pretend we have a parabolic reflect wrapped around
an omni antenna feed as in the various FreeAntenna models. The
illumination angle of the dish is about 120 degrees at best. That's
1/3 of a full circle, so 2/3's of the tx power goes off to who knows
where. Gain loss is:
10 * log (0.3333) = -4.77dB
So, the tx gain of such an antenna will about 5 dB less than the
receive gain.
I've never bothered to measure this one:
<http://802.11junk.com/jeffl/antennas/Misc/slides/lantern-reflector.html>
and probably never will. It is closer to a hemispherical section,
than a parabolic section. It sorta works, but I did it more for a
joke than for anything practical or useful. Also, the aluminum is
very rigid and difficult to bend into a parabolic shape.
>Noted on the www.freeantennas.com home page, at the bottom,
>"Undoubtedly the best implementation of the Parabolic Template Design I've
>seen so far." http://www.nodomainname.co.uk/parabolic/parabolic.htm
>Is a fine mesh parabolic.
Nice. However, the large size JPG images are missing from the web
pile.
--
Jeff Liebermann
jeffl@cruzio.com
150 Felker St #D
http://www.LearnByDestroying.com
Santa Cruz CA 95060
http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
why is SMA cable so expensive? 
Linear Mode
