I won't go into details about dimensions since they are well documented at the above link. I used the lid from a fisherman's friend tin (nasty menthol mints) for the patch and a piece of scrap galvanized steel for the ground plane. I cut all the parts with a pair of tin snips (actually it took me longer to hunt down a screw in my garage to fix my broken snips than it took to make everything!). I'm not sure how good the dielectric constant of cheap plastic pens is, but it's the only thing I had handy last night; that's what I used as a spacer for patch and ground plane. Hot melt glue holds everything together except for the N jack, which is soldered on. The whole assembly is contained inside a compact (4x4") plastic food container. It's a pretty messy job I did, partly because I didn't have much time and built the thing in a rush, and partly because I wanted to demonstrate what a rip-off commercial antennas are; that even with crude and imprecise contruction techniques and unglamorous materials, you can get decent performance.
Here is a handy tool for making odd-sized round holes. First, you drill a hole with the best matched bit you have, and then ream it out to the exact size you want with this:
I bought mine at Rat Shack when I was a kid. Hopefully, they still have them; if not, any decent electronic supply store should carry them.
In retrospect, hot melt glue is probably not a good idea; a hot car baking in the sun easily gets it gooey enough for everything to just fall apart. Next time I'll use epoxy putty or PC7.
Lacking any decent test equipment, I've only done some qualitative tests with this antenna, and it seems to have decent gain. First I tried to use ministumbler as a signal strength meter, but the numbers were fluctuating too much to get stable results. The signal strength and SNR w/ the patch antenna were anywhere from 1-3dB better than w/ my Hunts beans waveguide "cantenna" after adjusting pointing each one optimally.
Next, I did a ministumbler AP detection test. First walked far enough from my house that none of the antennas could pick up my AP. Then I started walking towards my AP with each antenna until ministumbler detected it. The patch antenna wins by a narrow margin in terms of range. There is a tree with thick foliage blocking the view of my AP from the street, so the range is greatly reduced, but anyway ministumbler found my AP 6 houses away the patch, about 15 feet closer with the cantenna, and with my Orinoco Gold's built-in patch antenna about 3.5 houses away. What's nice about the patch antenna over the "cantenna" is its broad beamwidth. The cantenna has to be pointed very precisely at the AP to get anything at that range, but the patch can be tilted several degrees and still get a signal.
All in all I was pretty impressed at the performance I got. This antenna is so tiny and portable and discreet compared to the cantenna. I'm going to use it as my main stumbling antenna from now on. Here are photos of the setup in my car.
I think I could probably eke out better performance with more careful construction techniques; when I have more time I'll probably redo all the metal parts with more accurate measurements/cutting and support the patch on all 4 corners to get more accurate spacing.
Does anyone know where I can get more info on designing patch antennas? I'm interested in finding out the effects of the dimensions, shape, distance between patch and ground plane, and accuracy in the dimensions. Please e-mail me at email@example.com if you have any good info.
Thanks again to outcast_one and otaku42 providing the design.