Homemade Antenna Choice, Jeff, et al, can you give me your best guess?

I avoid using USB dongles unless they have an external antenna connector. Neither of the two you've mentioned has an external antenna connector. I have no idea which of the two is "better" since I've never used either.

The problem with putting a reflector behind a USB dongles is that you'll get different gains between xmit and receive. In receive, the reflector picks up the aperature area of the dish and sends alost all of it to the USB dongle. However, in transmit, the USB dongle sprays RF in every direction, including directions where it misses the reflector. This RF is lost, going to places unknown. I ran through the calcs in a previous rant, but am too lazy to find it right now. Bug me if you want numbers. As I vaguely recall, there was about a

10dB difference between xmit and receive. See dish illumination issues under:

(see spillover).

Methinks you'll be better off with a wireless client bridge, which talks to the computer via ethernet, and usually has an external antenna. The nice feature of this arragement is that it will work with any OS that talks ethernet and that there are no flaky USB drivers to deal with.

I haven't checked, but I've had "issues" with USB dongles under Linux in the past. Make sure it's well supported before buying.

It's usually on the pointed end opposite the USB connector. Something like this:

I'm always busy, never have time, and am lazy.

I prefer my veggies raw instead of steamed.

Like the above salad bowl, the colander is closer to a sphere than a parabola. The result can be seen with the shiny bowl and a light as a ring of light. No energy in the middle, along the center axis, then a ring with all the power, After the ring, nothing. It's nowhere near a point, so therefore, lousy gain.

Much better. If you can get it close to a parabola, you'll have a fine receive antenna, and a lousy transmit antenna due to reasons previously mentioned. The struts will have little effect if they're reasonably thin. If the wood is dry, there's little attenuation.

I don't understand the question.

RF travels on the surface of conductors. At 2.4GHz, the skin depth is much smaller than the aluminized mylar plating. Making the aluminum layer any thicker just adds weight and does nothing for gain. Try the same plating thickness at lower RF frequencies and it won't work.

What are you trying to accomplish? Why are you using a USB dongle?

While I agree with Jeff that USB sucks, this is a decent USB wifi dongle. The antenna is external via reverse SMA. It is quite cheap through Data Alliance and these guys ship fast. I got this one and the N version, though the 036H is the better long distance unit.

While it's fun to build antennas, there are some cheap wifi antennas that are pretty good. Try the Laird PA24-16. The radiation pattern doesn't have a lot of lobes. It is about $35 if you look around.

I built the wifi biquad a few years ago for yucks. It worked OK. There are some tips on this list on how to build it since many of the websites do it incorrectly.

I like these panel antennas. They are easy to transport. All the elements are under a radome, so other than water infiltration problems, they are well suited for outdoors. Avoid units that come with a pigtail. It is better to have a connector on the back.

Given construction tolerances, antennas that depend upon aperture rather than elements re-radiating the signal will work better. That is, a yagi looks good on paper, but a lot depends on how tightly all the dimensions are controlled. These panel antennas just sum a lot of smaller antennas, so the bandwidth is broader and the design is more tolerant to manufacturing tolerances. Some of these panel antenna are air isolated, and some are just etched PCB. I would imaging the antenna with air spacing works better, though I don't know if there is a list of how who builds what.

Jeff Liebermann wrote in news: snipped-for-privacy@4ax.com:

Thanks much for your reply. I don't have a choice, money is tight and I got these two at an estate sale for $1 each.

Did not know that. Read one source that said a deeper curve than parabola will attenuate some of the x-mit losses?

Haha, you're not lazy far as I can tell

I mean the backing on the umbrella will be mostly continuous alum foil., whereas backing on the collander is mesh with holes. Also umbrella has metal pole in center that I will cut partially but there will still be a metal post at the center, so losses from that, probably minimal.

I guess I don't have to worry about removing the thin nylon of the umbrella as the waves will go right through it and reflect on the foil on the other side, right?

Just trying to get any reasonable surfing connection to a hotspot. I have another old computer running xp with an alfa radio and half parabola behind it (quite large) that is getting me about 100kb/s on what I think (but do not know) is a pretty distant AP, not sure exactly who owns it specifically, but they leave it wide open, I think it's a major COM company.

I just tried a flood lite reflector about the same size and shape as the collander and it's not strong enough only getting

1-2 bars, 86 reading on Inssider. BTW just started using Inssider, cuz heard Netstumber does not work under Win7. I don't like the readout on Inssider-too hard to read from a distance for positioning of the antenna, unlike Netstumbler which gives a very clear graphical readout from a distance (maybe u can switch to graph view on Inssidder?)

The Alfa AWUS036H I suggested is Linux compatible. Kismet as well. Granted $25 can't compete with $1, but the unit works well.

I really think you can't skimp on the usb dongle. I suppose you can save some money rolling your own antenna if you have more time than money, or you just want the experience.

This is probably the best page for biquad construction. Per Jeff's suggestion, I used round loops rather than squares. The deal is you need the proper circumference. The shape isn't as important, and it is easier to get the circumference correct with a round loop. Marty does the construction properly by using a coaxial feed. Jeff ran NEC2 on the other scheme and it had a funny radiating pattern.

The reflector doesn't have to be copper, but soldering to anything else is a lot of work. In fact, soldering to heavy copper clad isn't all that easy unless you have a high wattage soldering iron or you have a small propane torch. Building your own antenna is cheap IF you already own the tools or have access to them. Otherwise you are buying hacksaws, copper pipe, etc.

You can use the biquad to feed a dish for more gain. You can often get a dish for free on Craigslist. The Dish or Direct TV dishes probably are too small to be useful since the dish doesn't see the all the RF from the biquad. You can get larger 24x36 elliptical dishes for free if you stake out Craigslist.

A bit harder to find are the old MDS antennas that were sold for "wireless cable". They use a parabolic reflector.

All that said, the Laird PA241-16 antenna I suggested is far less work, and the difference between off the shelf and home built might be $10 when the dust settles.

miso wrote in news:j8q7t4$lg$ snipped-for-privacy@speranza.aioe.org:

Read too many complaints about that outfit you mention, located on the Mexico border. If you buy wifi equipment stick with suppliers that guarantee and test their products:

Good price. Hopefully, they'll work.

Yes, but within limits. Visualize your USB dongle at the focus of a parabolic antenna. If the f/D (focal length divided by Diameter) ratio is large, you have a shallow dish, with the USB dongle well above the reflector surface. the result is that the dish only collects a small part of the RF belched by the USB dongle. However, if the f/D is small, you have a deep dish, where the dongle is located deep into the dish. More of the RF it belches is collected by the reflector. However, there's a limit. Let's pretend that the dongle is in line with the edge of the dish. Half of the RF belched by the dongle hits the dish, and the other half goes off in useless directions. So, even with such a dish, you have a 3dB difference between tx and rx. The steep sides of the deep dish also don't contribute much to the gain of the antenna. So, the gain actually drops as the f/D decreases. There's an optimum value of f/D. See the graph at:

Since you're cost limited, I suggest you try one of these:

Cardboard, covered with aluminum foil, works well enough.

Also, note that it's fairly easy to build an antenna with gain up to about 8-10dBi. Any more gain and the construction tolerances become more than what you can do with cardboard, baleing wire, and duct tape.

The holes in the colander are too small to have any effect. The non-parabolic pattern will cause some gain loss. The marginal parabolic shape of the umbrella, as aggrivated by the non-parabolic shape of the foil between the umbrella ribs will cause some gain loss. I'm not interested in modeling either design.

I think the nylon is RF transparent. However, if it's been painted or silk screened with something containing carbon or metallic powder, you'll have a problem. Find a small piece and put it in the microwave oven. If it gets hot, you loose. If it says cool, you win.

Ummm... I'm not really interested in helping you do that, especially since it seems you haven't obtained permission.

Nestumbler is ancient and has not been updated for about 5 years. Inssider is a good replacement. There are others. Bug me if you need a list.

Yeah, you complained the last time I mentioned Data Alliance. Hey, two orders, no problems. I order at night, it is shipped in the morning, and I get the item in 2 day. I should be so lucky with most vendors.

I suspect you are fronting for the competition.

OK, so they are only A when top of the line is A+.

Do you have some problem with Mexicans?

I have opened a couple of dongles, cut the antenna foil and installed an N connector. You need to plan mechanical support for the N connector, but they both worked better than anything I had. Now I have an Alfa AWUS036H and it works better than the inexpensive dongles I modified. At one time I connected a modified dongle directly to the MFJ Yagi and ran a USB cord into my boat. That worked well for a couple of years. Mikek

The AWUS036H is magic.

miso wrote in news:j8qp7b$o1$ snipped-for-privacy@speranza.aioe.org:

Nope I have NO monetary interest in the Wifi field, been all over Mexico, I like them. What I don't like are unprofessional "suppliers" who put up a webpage, thing they are a big outfit when they are small potatoes and treat customers badly and DO NOT WARRANTY their extravagant claims.

By the way if you READ the link I supplied you will see the OP said they were rated low until they made a payment to the BBB and overnight their rating went up to A. Like most gov. agencies the BBB is corrupt.

No it's not :)

The Realtek 8187L chipset that it uses is what's magic.

Any other cheap USB dongle with the same chipset and a removable antenna will perform just as well.

For "G" applications this chipset is hard to beat. The drivers work with the latest Linux flavours a lot better than they used to. Will now hold fulls peed.

Jeff Liebermann wrote in news: snipped-for-privacy@4ax.com:

They do, but you're right, not strong enough, maybe I have to buy another Alfa.

Yeah looked at all the homemade sites. That's how I got the umbrella idea.

Yeah thanks, I will test it in microwave.

How am I going to ask if I cannot determine who owns it? It's reverses to a big Telecom. I don't have the equipment to follow the signal to their door. Besides their intention clearly is to leave it open. For all I know they are sniffing my data; even SSL is no longer safe.

Insidder works on 7 but the graph output is very hard to read from just a few feet away. With NS you got a solid graph that you could adjust your antenna from across the room. I tried NS on 7, it installed and opened ok but cannot see anything, even when set to xp compatible.

I seriously doubt you can buy a good rating from the BBB.

The BBB is private enterprise. Of course, that makes them more likely to be corrupt.

If you want to go for cheap, I'd build the biquad then just solder wires to your $1 wifi dongle. You will have to inspect the PCB and cut out the internal antenna. If you have the FCC number, it is possible to find a photograph of the inside of the dongle.

You do realize that Netstumbler can be detected. It is not a passive sniffer.

I haven't run NS in years, but I don't think it finds stealth WAPs. Thus you can have interference without knowing it. There are Cygwin implementations of kismet, though I never tried them myself. I'd sooner try to set up a thumb drive linux.

It's easy enough to stick the USB dongle into the side of a coffee can. I theorized that two of the 2lb cans, soldered to make a longer can would be the good solution, but the "customer" said the single coffee can was better, because it didn't have such a narrow beam. He was just setting them on the back of a couch in the window, using pillows to clamp it into the right orientation to reach a router about a block away.

Other ideas:

My coffeecantenna:

I used the calculator at decide where to stick the USB dongle through the side.

It needs to make a tight fit between the USB shield and the metal can. later, I soldered a 10-24 nut to the can so I could mount it on a camera tripod, but my friend got cable in the meantime, so that ended that quest.

So does it work very well with a dongle ?

Trying to use a plain USB dongle with a reflector has been less than satisfactory from my experience. The radiation pattern is very irregular. Just turning a dongle around in a reflector dish sees the gain go up and down.

I've got a spare Belkin adapter in a drawer and might give it a try.

Might get a better result by using a dongle with a dipole antenna and just drill a hole into side of can, insert dipole and strap dongle to exterior of can with a rubber band. Will give that a go.

I suppose that depends on your definition.

The two 2 lb cans gave better signal improvement, but the single 3 lb can was more usable in the desired application.

I think, when I was doing careful testing with Netstumbler, that I did see a difference between front facing and rear facing in the can, but I forget. I didn't try side facing.

I also tried a few woks and pie plates with various spacing of the dongle from the reflector surface. I did some toying around with a used DirecTV dish, but that was too much effort compared to the success of the coffee can.