I would like to build a cantenna that uses a usb adapter as the active element so that I can avoid the problem of losses in a long run of coax cable and just use cat5 instead. But I haven't been able to find much info on best dimensions and usb adapter placement within the can. What is the best diameter can and how long should it be? Should the usb adapter poke through the back of the can in the middle or into the side of the can like in a regular cantenna? If through the side, how far from the back of the can?
I plan to use an inexpensive Zonet ZEW2501 usb adapter.
Has anyone built something similar and how did it perform?
Thank you both for the info and links. What I have in mind is the same as Clarence Dold's coffee can. I want to use a USB dongle. But he doesn't give any dimensions or formulas for determining them on his page. I suppose I could just experiment some but I'd prefer to have a calculated starting point. What kind of antennas do the dongles use? Are there different types? Is there a way to solder a pigtail onto a dongle?
Are you talking about a game adapter at the end of cat 5, or a USB dongle at the end of USB cable? cat 5 can go a lot farther. The USB should go 5 meters without an active repeater.
A friend was using a 6" diameter coffee can, which I found to have lower gain than the 4" diameter two coffee can setup, but the larger can was easier to point at an access point about two blocks away.
Bob Alston's coffee can
Dold's coffee can
used the turnpoint calculator to decide where to poke the hole.
There are also many haphazard designs on the New Zealand page, using USB dongles.
Using a "standard" USB adapter, instead of the dongles, is on David Taylor's site.
The only calculation I used was the turnpoint page that I referenced.
My coffee cans are 4" inside diameter. Plug that in to the turnpoint calculator, and it will tell you that a wire radiator should be 1.7 inches from the closed end of the can (measure inside). I made the assumption that a USB dongle antenna should be at that same point.
The other dimension of interest is the length of the antenna element inside the can. If I were putting the USB dongle in front of a simple reflector, I would want it 31mm from the reflector. This suggests that the antenna element of the dongle should be 31mm away from the wall of the can. The antenna is at the tip of the dongle. Simply inserting the dongle inside the can is conveniently near that dimension.
Here are some of the cans that I tried. I liked two 12 oz. Yuban cans the best, but my friend likes the 3lb Yuban can, because it's easier to point. According to the calculator, the 3lb can would be more efficient at frequencies below the 2400 MHz of WiFi, but it works well for him. The 12oz is even too large, but I couldn't get my hand inside a smaller can to put the dongle in place. If this were a permanent setup, maybe you could make it fit into a smaller can, but I was only playing, and I carry the dongle with me, and don't normally use the can.
Well done! It's encouraging to read that you got a connection a mile away with it!
One thing that bothers me about soldering a pigtail onto a circuit board is that it may cause an impedance mismatch. What kind of circuit is typically used to drive an antenna? A power transistor with low output impedance? Some pictures I've seen show a dongle with an antenna made of pc board traces. Where would I cut into the traces to add a pigtail?
You can go further but that would require more antenna gain.
True. I would have used a smaller diameter semi-rigid coax cable for the sole purpose of reducing the amount of exposed center conductor. That appears as inductance and will create a substantial mismatch loss if not properly compensated. If you look carefully at the motherboards on the WRT54G series of router, you'll see a matching section next to the TNC antenna connectors. That's specifically to compensate for the exposed center conductor in the cheezy TNC connectors or the coax cable to circuit board connection. |
the small parts on the leads going from the diversity switch IC to the antenna connectors or coax. Those are the compensation network.
Actually the antennas are usually connected to PIN diode diversity switch. That gets driven by a power amplifier IC. This is fairly typical.
Nope. Usually an integrated power amplifier.
That's the antenna.
I don't have a clue. Give me a model number and photo and I'll make a guess.
Unsolder 3 solder connections and an SMA connector. You can solder the antenna back if you don't like it. No big deal. Unfortunately, it's on a neighbors roof and I can't find the photos I took. Maybe I'll do another.
That reminds me.... If you don't know how to solder, find someone that will do it for you. It's way too easy to destroy these small devices.
I think I've found a problem with this article on biquad antennas. |
general construction is great but there's a problem with the way the biquad wire ground connections were soldered to the outer copper sleeve. The way a biquad works is that it's two full wave loops in parallel. The only dimension that's really critical is the length of the loops. They can be any shape including a circle, but the total length is what determines resonance. In the photos, he soldered the ground wires to the sleeve leaving about a 2mm length of tubing between the ground points. Aesthetically, that looks very nice, but the extra 2mm of loop length will cause a slight tendency to detune the antenna lower in frequency. I'm not sure exactly how critical this is. I'll play with an NEC2 model first to see what happens. My guess(tm) is that the loop has to be shortened about 1-2mm to compensate or the ground ends need to be soldered together.
I also take offense to the use of oversize coax and the large amount of exposed center conductor at both ends of the oversized coax. Losses at 2.4GHz hare high, but insignificant for short pieces of coax. I would have used semi-rigid coax. The transition between the huge center coax support for the biquad and my suggested semi-rigid coax will be a big problem. There's no way to keep the center conductor short. Therefore, methinks it would have been best to use a single piece of .141 semi-rigid for *BOTH* the center support and the pigtail. The .141 coax is stiff enough if the copper wire used in the two loops is reduced in diameter. This will slightly affect the bandwidth of the antenna, but not catastrophically.
In other words, change everything (one of my bad habits).
Otherwise, it's a very nice construction job and article on building a biquad. However, I would like to have seem some test results and comparisons with known antennas.
Not so fast. I'm having 2nd thoughts about my guesswork on how to solder the biquad elements. I gotta do an NEC2 model before I can be sure. I might actually be wrong. (What a horrible thought).
Yep. It's magic. That's the excuse I use to justify my exhorbitant consulting fees. If it were easy or obvious, I couldn't be able to charge as much.
It's exposure to too much RF. When I got started in radio, I had a full head of hair, steady hand, positive attitude, and full bank account. After about 40 years of RF exposure, my hair is falling out, my hand is shaking, my attitude is totally cynical, and my bank account is depleted. Obviously, this could only be caused by RF exposure. Who do I sue?
Good start. See what it does before you destroy it. Take some pictures and post them somewhere. I'll suggest where to hack and cut.
5dB gain is not bad depending upon size.
I was in a local coffee shop about a month ago and saw someone with a ribbon cable extension from their PCMCIA card slot to the wireless card hanging on the back of the laptop. Whatever works, I guess.
Maxwell Smart (Agent 86) Don Adams is still very much alive and possibly amused. The wireless shoe phone is sure to be released shortly. We already have the "Cone of Silence".
Thanks for the excellent information. I'm a circuit slob going back a number of years but this RF stuff is like magic to a low speed guy. Speaking of soldering, I used to be able to solder anything to anything but somehow my fingers got thicker as I matured and surface mount componets now strike fear in my heart :)
I haven't gone very far into this project yet but my dongles (Zonet ZEW2501) arrived today and I ran a few quick and dirty table top tests using Netstumbler. I made a crude corner reflector by doubling up and folding in half a piece of aluminum foil and could see improvements of about 5 dB by placing it near the dongle and varying the position. A Trendnet TEW226PC pcmcia card that I already had seems to work slightly better but I can't put it on the end of a cable like I can the usb dongle.
I thought about that bit. I'm not sure that the redwood image isn't stolen directly from Marty. The picture shows the biquad with 30.5 mm sides, and a gap in the center, but the text on this page, Marty's and Trevor's indicates that the legs are equal length.
Trevor isn't clear on this point. He speaks of a 244mm piece of wire, but leaves a 1.5mm gap at the end. His shows the two legs right together at the copper tube. Marty and redwood show it spread a bit, as if it would go directly to the center pin if it continued.
You're suggesting that the ends soldered to the tube need to be just a little shorter, so that the distance from corner past the tube where the leg is connected into the (not connected) center conductor is 30.5cm?
I've been doing some more experimenting (nothing destroyed yet, but soon no doubt).
I mentioned earlier that I got about 5 dB improvement with a quick and dirty corner reflector. I rummaged around the kitchen and found a spaghetti strainer that's roughly a parabola. It seemed to work about as well as the corner reflector, but it was harder to hold in position since I don't yet have an extension usb cable for my dongle. That brings me to a whole set of questions about what is the best way to have the antenna at some distance from the computer that's cabled to the access point. But first a bit more about signal strength. Unsurprisingly, I found that my signals lose a lot of strength going through the brick walls of my apartment building. Maybe because they hold water. I'd like to be able to use my notebook both inside and outside my apartment. Is there a way to have both an inside and outside antenna connected to the access point? Kind of like a diversity antenna but with a separation of many feet rather than a few inches. If I buy one of the access points with two antennas, can I hook up an inside antenna to one connector and an outside antenna to the other? Is there anything that can be done for a router with only one antenna connector (parallel antennas - signals somehow isolated and combined where needed - half-baked)? What is the best cable to use? How far can I go? RF cable seems rather thick and difficult to work with. Can I get away with thin cable?
Another thing I will try soon is poking a hole in a can of beans (after dinner :) and slipping it onto the external antenna of my Netgear MR814 router. I figure I can just twist the can to get azimuth control and tilt the whole router up and down to vary altitude. I don't know how well the external antenna will feed into this crude waveguide but it seems easy to try.
Here are some thoughts/questions about cabling to the access point and wireless client:
I'd prefer not to use coax to the antennas since the coax is lossy and hard to snake around. I bought the Zonex ZEW2501 dongles thinking that I could put them in a can and use a long usb cable to put them anywhere I wanted. But now I see that there's a limit of 5 meters to usb cable length. Damn, I want to go about 10 meters. Etherenet cables can run 100 meters so they would be fine. I could perhaps mount the whole router inside a reflector and take care of one side of the link that way. But what could I do about the dongles? I bought two Netgear MR814 routers hoping I could use one as a client in place of the dongle, but the routers don't seem to offer a client mode. Any way to force it? Another (probably bad) idea is to use a junk computer running Internet Connection Sharing. I haven't thought that through completely but perhaps someone has done that or has better ideas.
My understanding is that this works, as long as the client can only see one antenna at a time. Jeff Liebermann expands on this in a posting in this group that you might find via a google search. My somewhat similar test is that my connection would toggle back and forth between the good connection and the poor connection when I had one stock antenna and one reflector.
The pattern is pretty wide. My reflector works better to the sides than the bare antenna did.
That likely won't work. The radiator portion of the antenna would have to be inside the can. Having it part in and part out would be unhelpful.
Better to build a reflector
EZ-12, printed on photo paper for thick stock, with aluminum foil glued to the sail, provides a substantial boost in signal.
5 meters of USB cable, but active repeaters can couple cables together. I bought a five meter cable with an active repeater lump at one end that was the same price as the 5 meter cable without the repeater.