Although I have
been aware of the guerilla.net
low power collinear omni for some time, I have hesitated to build it due to
its complexity, and the fact that I didn't fully understand the design. To
complicate matters, the website has been down for several weeks, making the
design inaccessible. I was finally able to piece it together by using a copy
from google's cache, and archived JPEG's from this site: http://www.tux.org/~bball/antenna/.
My google searches also yielded an
attempt to explain the theory behind the antenna.
Although the
guerilla.net folks claim that the antenna is their own original design, one day
while surfing the FCC site for information about Lucent's ORiNOCO cards, I
happened upon an interesting
document. On page 13 of this document was a photo of antenna labeled Maxrad which looked almost identical to
the gnet collinear omni. sparafina of the NetStumbler Forums did some initial
analysis of this antenna, and found its dimensions to be very similar to the
gnet design. Later, I found a test
report which refers to the antenna as external antenna AUO24-OD-10,
and lists its gain as 10dBi. Once I sat down to build the antenna from gnet's
design, I found their description of the coil dimensions to be confusing - maybe
I am just too dense. My initial attempt was a complete failure with very little
gain, probably due to my confusion over the exact dimensions. Therefore, I
decided to try again, but this time building it based on measurements taken from
the FCC photo. Below is an internal photo from the FCC archive which I've
modified to ease the estimation of the measurements:
click on photo for larger
version
The photo is rather blurry and pixellated, but I've
sharpened it a bit to increase clarity. If you'd like to check my measurements,
an easy way to do it is use Paint Shop Pro on
the larger
version of the photo. Simply use the selection tool to cut out a piece of
the ruler. Then you can use the selection as a ruler by dragging it around the
photo. For vertical measurements, rotate the selection by 90 degrees. Very handy
indeed. Not only is the photo blurry, but the quality control on the antenna
appears to be deficient; from segment to segment the measurements are not
identical. Here are the measurements I estimated from the photo:
Coil wire diameter: 1.5mm
Coil OD: 6mm
Coil pitch: 3.5mm center to center
Tube diameter: 3mm
Tube length: 56.5mm
The main difference between the gnet collinear
and the Maxrad is just the wire diameter. The gnet design uses 3/64" wire (which
is alot easier to bend) and 3/32" tubing. Also, the gnet design specifies
to leave a 2cm tail on the coils, which I found to be a little too short.
Otherwise, the two designs are pretty much identical. Since guerilla.net seems
to be down, I hope they don't mind that I've reconstructed
their page as best I could for your convenience.
Since I already had
the tubing and wire from the gnet parts list, I used 3/64" wire and 3/32" tubing
instead. I think the 2:1 ratio of tube to coil diameter is more critical than
the actual diameters anyway. Since the Maxrad decoupler is hidden in the
plastic, I used the gnet dimensions for the decoupler, but substituted a
N-female jack for the SMA jack.
For those who don't feel comfortable
cutting and bending the parts, aerialix sells the antenna in kit or
assembled form. Their prices are very reasonable.
Parts List
Below is the complete parts list with prices I paid:
(2) 12" long 3/64" diameter brass rod 2x$.59
(2) 12" long 3/32" diameter brass tube 2x$1.09
(1) 12" long 11/32" diameter brass tube $1.99
(1) 2ft long 1/2" diameter PVC pipe $0.10
(1) 1/2" diameter PVC end cap $0.29
(1) N-female panel jack $4.50
Grand Total: $11.50 including tax.
Everything except the N-jack was bought at Osh
Hardware. It's a good idea to buy 1 extra piece each of tubing and wire in
case you mess up. I used Schedule 50 pipe because it's the thinnest (I'm
paranoid about losses through the radome).
Construction
measurements
Below is my assembly procedure:
NOTE:
Try to be as accurate as possible in cutting the parts and spacing them during
assembly. .5mm accuracy is difficult but at least try to get within 1mm of
what's specified. Otherwise, you may be sorely disappointed with the performance
of your finished product. I used a pair of vernier calipers extensively during
the construction of this antenna.
Hold a piece of 3/64" brass wire against 9/64" drill bit and wind 4 turns,
each turn separated by about 3.5mm. Leave about 10mm of straight wire before
the turns and 24mm after the turns. Adjust the spacing of the turns with
needle nose pliers while the coil is still on the drill bit. Bend the ends
neatly with the needle nose pliers. Repeat until you have 4 coils.
Cut four 57mm long pieces of 3/32" tubing and one 91.5mm long piece. I
used the cutter on a pair of needle-nosed pliers to cut them; this flattened
them, which is not bad because the hole in the tubing is actually a loose fit.
Next, I used the needle-nosed pliers to squeeze the tubing until the hole
opened up large enough to pass the wire into it. Finally, I filed the tubing
smooth, which took off about .5mm off its length.
Cut a 30mm piece of 11/32" brass tubing. I used a hacksaw and mitre box.
Solder the 91.5mm long tube to the center pin of the N-jack. This is the
feedline
Slip the 30mm tube over the 91.5mm tube, and solder to the N-jack. This
tube is the decoupler. The feedline must be supported in order to keep it from
shorting on the decoupler; I slid a piece of rubber hose over the feedline.
Solder a coil to the feedline, leaving about 3mm of wire between the
feedline and the start of the coil.
Solder a piece of 3/32" tubing to the other end of the coil, leaving 22mm
of straight wire between the last turn and the tubing.
Repeat until you get to the top of the antenna.
Cut the PVC pipe to the proper length to enclose the antenna and glue on
the end cap.
Cut pieces of foam to support the antenna inside the PVC pipe - I stuffed
the foam into the coils - and carefully slide the antenna into the pipe.
Attach the N-jack to the other end of the pipe. You can screw it onto a
flat end cap if you can find that style. I chose to just tack the N-jack
directly to the pipe with hot glue. Then, I used epoxy over the hot glue for
strength and heat resistance.
decoupler
nude antenna
completed antenna in radome
Performance
I tested the collinear down the street from my AP using
MiniStumbler. Below are the results:
Antenna
SNR
ORiNOCO built-in:
10 dB
collinear:
20 dB
For some reason, the signal strength was
fluctuating wildly, so this is the best estimate I got. Overall, I'm pretty
happy with it, which is good because this antenna cost more and took by far more
time to build than anything else I've made to date.