In article , Jeff Liebermann writes:
Thanks and interesting. I discarded synthetic aperture imaging
because I assumed that either the sensor array or the object being
imaged had to be moving roughly perpendicular to each other. That
seems to be the case with SAR (synthetic aperture radar). I'll read
some more (later) as I have no experience with the technology.

You mean you were planning a 30,000 foot eyeball and no way to
aim it?

Yes, you are probably right - there would be issues with off-axis
imaging, especially if the individual antennas were widely spaced.

Unfortunately what I know beyond what I talked about is rather
sketchy, but I do know that synthetic apertures are used for
optical telescopes. Instead of a single, perfectly polished
mirror, you place multiple mirrors somewhat distant from one
another and use optical magic (smoke and mirrors?) to put it
all together for form an image.

Anyway, if you have a telescope mirror with holes in it, you
have tradeoffs.

I'm guessing that what happens is that there are
aliasing effects. If the spacing along, say, the
x axis, is s and wavelength is w, you will have
alaising - images of off-axis points that appear
to be on-axis, for example - and I would expect
those to be at angles

arcsin ( N w / s )

relative to the normal (read arcsin as
"the angle whose sine is")

If you want to see something that is off axis, you
might be able to leverage this if each antenna
is directional and blocks most energy from outside
a main lobe narrow enough that, for small N
at least, the antanna only picks up signals from
one of the aliased angles and blocks the adjacent
ones - kind of like an RF amp passband that allows
a desired frequency through and not its image
frequency.

And you might be able to tune the pattern so the
nulls in the pattern at least partially null
out aliases at the N-1 and N+1 angles, where
you would have to have some lobe width adjustment
if you wanted to use this technique for more
than just a single value of N.

If you don't want to use a dish, perhaps you
could use a 'Pringles can' antenna with a dipole
at the far end of a long cylinder - your "telescope
body".

You would feed measured magnitude and phase from
each antenna to your computer to have it produce
an image.

And if you were really good, and used a UHF
illumination source, you would interfere
the illumination source with the received
signals and via holographic techniques
produce true 3D.

Just speculation. But if it's doable I
would guess the military has already done it.

George

In article , I wrote:
If you want to see something that is off axis, you
might be able to leverage this if each antenna
is directional and blocks most energy from outside
a main lobe narrow enough that, for small N
at least, the antanna only picks up signals from
one of the aliased angles and blocks the adjacent
ones - kind of like an RF amp passband that allows
a desired frequency through and not its image
frequency.

Actually, the angular spacing increases with N,
so if the dish excludes alias images when aligned
with the overall "optical axis" then it excludes
them when aimed off axis as well.

And I am assuming s w, or the formula gives no
aliasing at all.

George

On 16 Jul 2015 01:18:17 -0400, (George
Cornelius) wrote:

In article , Jeff Liebermann writes:
Thanks and interesting. I discarded synthetic aperture imaging
because I assumed that either the sensor array or the object being
imaged had to be moving roughly perpendicular to each other. That
seems to be the case with SAR (synthetic aperture radar). I'll read
some more (later) as I have no experience with the technology.

You mean you were planning a 30,000 foot eyeball and no way to
aim it?

That was the Mark I model. Future models will involve some
miniaturization.

I can't comment on your speculation because (1) I don't know much
about synthetic aperture imaging and (2) it won't work anyway. I
tried to resolve the first problem by doing some light reading on the
topic:
https://en.wikipedia.org/wiki/Aperture_synthesis
What this demonstrated was that either the telescope of the imaged
object needs to be moving. In the case of the optical telescope, it's
the earth's rotation that does the moving. I don't think this is
compatible with an RF eyeball that fits on my workbench. The 2nd
problem is easily solved by what I consider to be a better method. But
first, I need to define an objective in electronic terms.
What I'm trying to accomplish is build an antenna array that
has extremely good resolution, without making it brobdingnagian.
If this can be done with just one antenna system, it could be moved
around in the form of a flying spot scanner to obtain an image,
similar to an optical "flying spot scanner".

The basic problem (for me) is how to get obtain good angular
resolution from an antenna with not so good angular resolution. I
solved this problem with an idea I stole from the WWII Lorenz blind
landing system, using 2 directional antennas or one switched antenna.
https://en.wikipedia.org/wiki/Lorenz_beam
However, I reversed the location of the transmitter and receiver. My
system consists of two identical wide "beams" similar to the beam
pattern produced by any directional antenna. The angular resolution
of the beams causes the amplitude of the signal to vary depending on
it's location along the beam pattern, just like any directional
antenna. By itself, this angular resolution is useless for imaging.
However, if I take two identical antennas, position them at a slight
angle from each other, and switch rapidly between rapidly, the line of
equal signal level half way between them is VERY narrow. In my
testing on VHF, the equal signal null produced was less than 1 degree
wide and could probably be improved with a better test setup. I have
some sketches and photos buried somewhere and will post them if I can
find them.

The circuitry is fairly trivial, consisting of a synchronous antenna
switch and a synchronized AM demodulator charging two capacitors (one
for each antenna), and a comparator. See the block diagram for my AM
homer system:
http://802.11junk.com/jeffl/AN-SRD-21/
http://802.11junk.com/jeffl/AN-SRD-21/Block%20Diagram.pdf

So, how do I produce an image with a null generating derangement? It's
somewhat like a photographic negative, but not quite. It's also great
for direction finding, but not so great for imaging. Simple inversion
of the negative will not produce a usable image. I have some tricks,
but all of them rely on the dynamic range of the AM demodulator, which
frankly sucks, especially in the presence of noise. Reflections also
caused major problems. That's where I stopped working on the idea.

Assuming I can extract an image, a rotating or scanning antenna system
would only produce a line in one axis, which is hardly an image. So,
I propose to store the horizontal line scan, rotate the directional
antennas 90 degrees, and scan again. Where the detected (stored)
voltages in both axes are equal, it produces an output dot. More can
be seen by adding frequency (color) to the output.

While at first glance, this might seem like something thrown together
using WWII technology, implemented with 1970's hardware, and lacking
the benefits of modern acronyms. Yeah, that's probably accurate.
Still, it's something that can be built using technology available to
the average ham. However, instead of using it to RF image a PCB or an
antenna, it might be better to start outdoors by imaging the
neighboring RF environment with a rotating antenna on the roof or
tower. In theory, one could "see" RF sources and reflections off
building and mountains. For indoors, I visualize a motorized X-Y
track mounted on the ceiling, with an antenna array similar to a yagi
pointing downward towards the device under test.

Good luck.

If you don't want to use a dish, perhaps you
could use a 'Pringles can' antenna with a dipole
at the far end of a long cylinder - your "telescope
body".

I have an aversion to using a waveguide beyond cutoff for anything
more than a parabolic dish feed. The main problem is the asymmetry of
the pattern caused by the feed being offset from the centerline of the
can. See horizontal pattern:
http://802.11junk.com/jeffl/antennas/coffee2400/index.html

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558