Range on a pair of Belkin 54G routers in wireless bridge mode?

Probably. (I have no experience specifically with Belkin equipment, but unless there is something very unusual about it there won't be any real difference as far as range goes.)

However, 400 feet from one building to another, works if and only if certain things happen to be true. You simply will not get good results going through walls, and some walls (metal) are absolutely killers. Hence you will do much better if you can place both units in a window, for example. Or even outside if you manage that. And the higher the elevation the better, because reflections from vehicles on the ground can cause signals to fade too.

Hence worst case is two units at ground level with a two or more walls between them, and best case is both units sitting in a 4th story window with an unobstructed view of each other.

With that said, I have one unit sitting in the window at ground level, which works fairly well to a repeater about 400 yards away. Moreover there are two houses blocking a direct view! Near as I can tell the signal is bouncing off a building on the opposite side of the street (none of them have metal siding or anything obvious, so I'm not really sure just what the signal is bounding off).

The same AP is just barely useable through one wall, on the ground floor of the building directly across the street! I put a repeater in a second story window, and it wasn't all that good either! But when I moved it from one side of the window to the other, a matter of about 20 inches, it became rock solid!

The moral of the story is that you'll have to try it and see what you get. Under the right circumstances it can certainly be done.

They all use the same set of channels. A cantenna optimized for channel 11 might not work as well on channel 1, but I'd guess that the difference would be slight. Certainly an antenna cut for say channel 8 will do equally well for everything from 6 to

10, at a minimum.

The typical construction of a cantenna is not adaquate for outdoor use, except perhaps in a very dry climate where there is also not much wind. But for mounting inside, they make a significant difference, both in signal strength at the desired location but also to reduce signal strength in undesired directions.

This might help. I ran Trevor Marshall's model of a can antenna through 4NEC2 to get some numbers:

gain and vswr are fairly flat across the band (2400-2483.5). However, there are some differences that I'm guessing (this is a guess, not a calculation) that will yield about 1.5dB difference in gain and reflected power loss difference between the center (chnl 6) and the edges (chnl 1 and 11).

However, also note that the design uses a conical 1/4 wave feed and not the usual 1/4 wave single wire to obtain additional bandwidth. I've been meaning rework it a single wire and see what happens. I'm fairly sure that the bandwidth will be much narrower and band edge losses will be much higher.

Agreed. The big advantage of the can antenna is that it is cheap and easy to build. Everything else about it is wrong or can be done better. I've compared the coffee can flavour[1] of cantenna with my rather sloppily constructed biquad panel implimentation. In general, the biquad is superior in all measurements and crude tests. It's also much smaller, I think easier to build (there's some debate here), and much easier to package.

don't have an easy way to measure absolute gain accurately, but I can easily measure relative gain difference between antennas. The single biquad has about 3dB more gain than my 1lb coffee can implimentation.

[1] Rumor has it that caffinated coffee cans have more "punch" than decaf.

One reason that the coffee can type of antenna is fairly popular is that it was the first of the do it thyself antennas and therefore has some kind of "image" that it is better than a commercial antenna. Never mind that almost any type of external antenna will be better than what's delivered with the stock AP's and client radios. Argh. phone call...

No. It's the feed element. In most coffee can antennas, it's just a

1/4 wave piece of wire. In his implimentation, it's a conical shape. Here's a photo from his site of a 5.6Ghz version: 's kinda hard to see on the 4NEC2 similation, but if you look at:

'll sorta see the conical feed in red.

Like I said, it's arguementative. I find that tweaking the coffee can for maximum gain is more difficult than the biquad, which usually works the first try. I post a photo of a biquad that I'm currently building that's held together with solder and hot melt glue.

Notice I didn't say anything about planting USB radios inside cans. The original comments were related to a wireless (transparent) bridge, which methinks a USB radio is generally incapeable.

Pill bottle? Well, someone showed up with an antenna made from a large aluminium funnel, a TNC connector, and some mangled coax. At first, I thought it was a joke, until he showed me some comparison tests, vswr measurements, and antenna patterns. I vaguely recall about 8dBi gain. The moral is that you can build an antenna out of just about anything. Pill bottle?>

That seems to be the consensus. Actually, I think it is a 1lb coffee can (100mm x 185mm) for the Trevor Marshall design. Reducing the numbers in the table, I see random rubbish indicative of significant errors, probably from reflections. Reducing the tables to differences between the can and biquad 1 and 2, I find that the two biquads are anywere between 4dB better, to 4dB worse than the cantenna. That's an

8dB variation in results which methinks is rather inconsistant. That's what I expect from a random test range. Reflections from the ground are always a problem with street level tests.

Seriously, there is hope. However, I have problems resisting the temptation to interject some humor in an otherwise dull and overly technical topic.

Yep. Remind me to send you my hacked version of Ivor's antenna design spreadsheet.

has some problems that I'm trying to fix. However, it's close enough to demonstrate what happens with different lengths of can antennas. Methinks you'll be rather amazed at the effects of the can length. As I recall, the problem with the Pringles can was that it was close to the exact wrong length for the worst possible performance. There's also an issue of skin effect conductivity. The aluminized paper is probably not the greatest. Neither is whatever is used to plate the insides of a tin can. I build my stuff from copper sheet roof flashing. When it works, I silver plate it.

Drivel: Gotta run to the local hospital for part 2 of a radioactive treadmill test. It was LOTS of fun demonstrating how radioactive I was (100mR/hr) after yesterdays test to my friends and customers. Those with a clue were tempted to run for their lives after they heard my geiger counter clicking. 24 hrs later, I'm still doing about

8mR/hr. Normal background is about 0.1mr/hr. Wheeeee...

I don't see any cantenna references on Trevor's site.

Are you talking about the cone in front of the can, or something different about the feed wire itself?

"Can with Conical Horn".

pretty easy to build. I don't see how a BiQuad could be easier than that. My USB-dongle coffee can wouldn't even require any soldering in the one-can configuration. I'd have to look at the numbers to see how much gain I got from the second can. The USB-dongle in a pill bottle takes the prize for ease of concealment in an outdoor setup. I can't locate the reference page, but they just put the USB dongle into a pill bottle, and arranged it over a light on a metal building doorway. It looked like a motion sensor, and the metal building provided some reflector, giving them a shot to another building that wasn't too far away, and was unreachable with any antenna inside the metal shop.

doesn't agree.

Of course, that's not a 1lb coffee can. When I fussed around with the calculators, and some trials, I found that two one pound coffee cans soldered together made the best antenna of the cans I tried.

I lose any hope of a serious conversation with you, sometimes.

No, I think that would be the difficult-to-build Pringles can. It generally comes out poorly when compared to the cantenna.

I still don't see any on Trevor's site ...

Wow! Okay, that might be difficult to construct, but I still don't see anything on his site about cantenna. That cone is used in a flat slotted waveguide

Ah, tweaking. Given how hard it is for the home hobbyist to get a reasonable test range, tweaking is just out of the question in a coffee can. That's why you use a calculator and poke your hole, and take what you get... Unless you drink enough coffee to have a steady supply of cans. And then, where is the tweaking? Assuming that someone got it right, that's it. Poke a hole in the place that you're told to poke it.

Using the calculator at

decided that a pair of Stagg's Chilli cans were the best size for a wire cantenna, but it was too small for me to fit my USB dongle. The Staggs is 85mm x 220mm. I use that with my Orinoco. For the USB, the 4" coffee can was a usable size, similar to Trevor's.

I played with that, and didn't find any difference from turnpoint, for the sizes I had on hand. Some commentary on one of the sites made me think that the differences were from measuring on the outside of the can, which didn't take into account the raised bottom. I measured the inside.

How long do you think it should be? Doubling was good for the chili and the coffee. The chili really needs to be a little longer, and the coffee could stand to be shorter, if the goal is one wavelength. Adjusting that would be rather difficult with cans. You'd have to prototype, and then build the right one. A rolled copper sheet suddenly gets a lot more attractive.

This weekend, I'm going to poke a hole in the bottom of a 3lb coffee can for my USB, mounted 31mm tip from bottom, kind of like the sardine can at

but bigger and more shielded.

He didn't post an exaustive analysis. That's what I wanted to do (and never finished). See:

proclaims: "Another model that you might simulate is the ubiquitous Coffee Can feed. Lower gain than the BiQuad, it also tends to radiate significant sidelobes. I produced an optimum design for a Coffee Can Antenna, with about 9dB of gain. You can copy the NEC model from this link."

Yeah, that's what I said. The photo is for a 5.6GHz model, not

2.4Ghz. It's not like the cone is some strange oddity. See:

where to buy a 2.4GHz version $15. I suspect this is overkill for a coffee can antenna, but it does improve the bandwidth.

I beg to differ. You just have to know some tricks. For example, it's always necessary to use a reference antenna with known and controlled characteristics. I have such a beast that I've spent some time getting to play. I can easily measure the *relative* gain of the antenna as compared to the reference antenna. If I make the assumption that my test antenna is accurate, then I can easily calculate the absolute gain. I also make an effort to compare the NEC computah models with my measurements. I consider a 3dB total variation to be good enough. My test range is across a valley plastered with RF aborbing trees. My house is conveniently located on the side of a hill so that reflections are minimal. If I'm concerned about reflections, I just place a "tube" or "box" of RF absorbent foam around the beam path. It will also work with omnis, but I usually want to also characterize the mounting structure, so I'm usually forced to live with the ground and mounting hardware reflections. I would NEVER do the test on street level with a reflecting surface in the beamwidth or fresnel zone.

I never follow instructions. I would poke the hole in the bottom of the can and adjust the distance between the back of the can and a dipole or bi-conical feed element. For the more common side mounted derrangement, I would use a much smaller connector (SMA) with a thread mount instead of the usual N connector. The hole would be a slot in the side of the can. The SMA connector would be sandwitched between two copper or brass washers. That will allow it to be moved back and forth without leaving any (ungrounded) gapeing holes.

Quiz time: The driven element is 1/4 wave long. (Actually, it's about

0.95 * 1/4 wave due to end dispersion). So, from where do you measure the 1/4 wave? From tip to where the wire enters the shielded part of the connector, or from the tip to the metal can base ground? Yes, it's that critical if you want low VSWR.

Incidentally, a short discussion about modeling coffee can antennas in the NEC-LIST mailing list (on robomod.net) nailed me to the proverbial wall for posting an NEC2 model without first doing a sanity check on the model. It has some problems resulting in a potentially lousy gain correlation with reality. So, now I gotta fix the model as well as the spreadsheet generator. Grumble...

Ok, you answered my question correctly. You measure the 1/4 wave feed element from tip to where it shielding on the connector starts. That's fairly easy to measure with a small diameter SMA connector. It's a mess of trigonometry for a big fat N connector, where the effective shielding starts somewhat into the TFE part of the connector resulting in an unexpectantly long feed element. It's *NOT* measured to the base of the connector or the inside of the can which would result in too short a feed element. It's not magic, but it's also not trival to calculate. Therefore, I tweak and measure, also known as optimize.

It appears to be cyclic. Some lengths are good but a half wave longer causes the pattern to go really weird. I'm tryin got figure out if it's for real or merely a figament of my calculation. 4NEC2 version

5.5 just appeared, which should fix the problems modelling patches (that I found). Stay tuned.

Too expensive for initial prototypes. Use 0.090" thick aluminium roofing flashing. If you're careful, you can build a telscopeing tube for the ultimate in tweaking. Also, get some aluminium foil duct tape to hold the seams together. For something that needs to be soldered, use some copper tape. Only the inside is electrically important. You can get very sloppy on the outside. I usually plant some copper tape around connectors and joints for a better connection ground. The dissimilar metals won't be a problem as long as the antenna stays dry.

If you feel ambitious, you might wanna dry building a horn antenna, instead of a can, which has a noticeably higher gain per unit volume than the can.

Cardboard? Well, I guess that will work.

The copper flashing I use is expensive and is mostly for building horn, slot, patch, panel, or grounded backplane (i.e. biquad) antennas. I buy it by the roll from the roofing supply. I think the last roll cost me about $150.

Sigh. Best-o-luck. I don't think the sides of the sardine can are high enough or the ground plane big enough to offer much help with the antenna pattern. You might wanna compare the gain with something much simpler, like a chunk of box cardboard, wrapped in aluminium foil, with a hole poked in the middle. One of these daze, I'll rip my DWL-122 apart, replace the PIFA antenna with an SMA connector or install a real dipole antenna, and so some numbers. (Yet another project).

If you're serious about the can antenna, you might consider shoving an ABS (white plastic) 3/4" pipe across the diameter of the can, and cramming the dongle into the pipe. I think it will fit, but don't have a USB dongle type radio handy to test.

Aren't antennas fun? This should be interesting....

I meant physically. Moving the mount around in the coffee can is a bit problematic, compared to the slide feed on the biquad.

Tsk-tsk. A while back you suggested that not fully soldering around the base of an N-connector would have serious consequences. Now you're going to have a slot to play with, presumably with no solder at all?

That doesn't look like a difficult measurement, that's a huge measurement. You have the tip, you have the end of the solder bucket, you have the lump of nylon surrounding the solder bucket, the lump of metal holding the nylon, the side of the can, and the end of the threaded portion of the N-connector. Which are you saying is the right place? I would say the top of the metal, but the nylon would allow some radiation below there, so somewhat lower. Turnpoint says "the total length of both the brass tube and wire sticking out past the connector is 1.21 inches", so that's what I did.

But when you get done, you tell the world... Isn't that what 1.21" is?

Waiting patiently.

I'm thinking about shopping the stores for slightly different coffee cans, to make a tight slide. Otherwise, it's cut and try. I thought about stove vent tubing, since that does slide together, but I'd like something I can solder when I get done.

Oh, certainly. I think the sardine can is just about silly. I did use a

9" pie plate with some good effect, but I was mounted flat 31mm in front of the plate. I didn't think about poking through from the back.

I'm going for a 3lb coffee can. That wasn't as good as the 1lb can for the side mounting, but it's what I have on hand ;-)

That would be as a slide mount, coming in from the side, instead of just poking the USB connector through? The can I used positions that antenna about in the center, so I thought that might be good. I did notice that this style of mounting allows the dongle to wobble off vertical, so the PVC (ABS is black) pipe would help with that, as well as allowing some adjustment. Tugging on the cable moves the dongle enough to make a big difference in the gain.

Grounding is good. More grounding is better. Tighten the nut on the connector and squash it between two brass washers. Good enough ground.

The feed element starts from where it extends beyond the shield on the connector. However, with a wide connector like an "N", it's kinda a guess as to where to start measureing. Looking at the connector from different angles, the difference between measuring perpendicular to perhaps a 45degree angle is about 2mm. Doesn't seem like much until you consider that at 2.4GHz, the difference for a 1/4 wave element between each of the band is: 3.0*10^8 m/sec / 2.400 * 10^9 cycles/sec /4 = 31.25mm 3.0*10^8 m/sec / 2.4835 * 10^9 cycles/sec /4 = 30.20mm So, a difference of 1.05mm will move you from one end of the band to the other. Can you say critical? This is why tweaking is important and a way of life with microwaves.

There are far too many effects to guarantee that 1.21" will be the exact correct length. For example: Differences in N connector construction Variations on can diameter affecting connector projection size. Aspect ratio (dia/length) of 1/4 wave element. end dispersion effects. The variations will be slight, but when 1.05mm will move you from one end of the band to the other, methinks it's important to consider them.

Well, I could have done it this weekend, but I prefered to do other things. I'll get to it eventually. I promise. Really.

I don't think you'll need to do much soldering. Copper roofing flashing would be a good start. Please realize that it doesn't have to be round. A rectangular waveguide section will work just fine. As I previously suggested, make a rectangular horn antenna instead out of flat copper sheet. Soldering is easy (with a really big iron). If you wanna avoid soldering, use flanges and fill the gaps with copper tape. However, I suggest you build your prototype out of aluminium

0.009" roofing flashing and use hot melt glue to hold things together. Aluminium duct tape for filling in the seams. This isn't fabulous, but will get you close to the final dimensions so that the final copper version doesn't degenerate to cut-n-try. Also, be prepared to silver plate the final design.

Yes. I think it's an easier way of mounting the dongle. Shove it into the pipe, pack it in place with some styrofoam peanuts, and no glue required for mounting.

Sorry. Y'er right. The black is ABS while the white and grey are PVC. I've had minor problems with the grey PVC pipe detuning the antennas, but no problems with the white PVC. Fiberglass would be the best, but unless you wanna pay boating supply prices, PVC is cheaper. To be sure, shove a piece of plastic pipe in the microwave oven and see if it gets hot. If it does, forget that idea. Don't bother with the black plastic ABS pipe as I know it melts in the microwave oven.

Hot melt glue it in place. I'm a big user of "adhesive technology" and "surface mounting" aids, formerly known as bondo and duct tape. Whatever works. The previous generation was bubble gum and hay baleing wire.

Guys, after receiving my pair of Belkin 54g routers, much to my chagrin I found that they had no removable antenna mounts! Having no experience with Belkin and tons with Linksys routers, I assumed they all had antennas that would screw off.

So I'm going to have to do any potential replacement of a stock antenna on the wifi NIC.

Oh I've been there, done that with a cantenna on a Linksys router.

My original question was focusing primarily on the range in BRIDGE MODE. That range may be identical to the unit's nominal range talking to an average wifi NIC.

Long.story.short: I'm going to go diagonally from one end of a 200' (deep and wide) building to the other with five Belkin 54g's, four acting as bridges off of the main master unit acting as DHCP server and router for the entire network.

At that extreme corner, I have a wood exterior wall. 200' away from that is a concrete block building with an A-frame roofline I can put a cantenna up in to get a good L.O.S. back to the main building. Now I'm trying to decide which PCI (or possibly USB) wifi NIC to order.

Obviously I'm looking for one with great specs (high end on transmit power, etc.) and a removable antenna to allow me to go up 8-10' to get above the concrete blocks for the best L.O.S. with the least bulk in between the two devices.