You can extend the range of your WiFi-enabled laptop with this easy-build homebrew microwave waveguide aerial for 2.4 GHz.
A 'cantenna' is a homebrew waveguide aerial, short for tin-can antenna. In some countries they use the word antenna, others aerial; but it's the same thing, meaning the device on the end of the cable that receives and transmits signals (and canaerial doesn't sound so good). For lower frequencies, this might for instance be a short rod of around 500 mm. This is what you've got at your masthead for the boat's VHF aerial. It is a quarter-wave at around 150 Mhz. An HF aerial, being for lower frequencies still, is much longer: an insulated backstay for instance. Lower frequencies = a longer aerial; higher frequencies = shorter aerial. Microwave aerials are physically tiny - just 31 mm for the frequency we look at here so that complex types of aerials have to be used in order to get a decent signal.

Extending Your WiFi Range

We can usefully increase the range of the laptop's WiFi LAN modem by attaching a better aerial. This is only possible, though, if:
1. You have a WiFi LAN PC card.
2. The card has an extension aerial socket.

If not, there are several workarounds:

1. If you have a PC card but without an extension aerial socket, sell it. It was a bad buy in the first place. Get an Orinoco or Enterasys RoamAbout type of card instead: these have the required socket.
2. Or, find a friendly RF engineer who will hack the card and install an extension socket for very little, as he is a member of your family or owes you. A new card is 40 pounds or so; no commercial engineer would do it for less than that, plus there's the risk he will break it as no doubt it will be the first he's done; so on balance you're better off buying a new one.
3. If you have an onboard WLAN, then find a friendly RF engineer who'll modify it and install an external N connector on the laptop case for you. We do this for 75 so you can do the maths and decide whether it's viable. Add on the shipping of a laptop, two ways, and it usually isn't. Many will feel a better (and less risky) choice is to disable the onboard LAN and buy a PC card. If, however, you have an all-USB notebook with no PC card sockets; or even, dare we say it a Mac, it may be your best option.
4. The last option is to buy an active external USB AP setup. This means a self-contained Access Point itself (which would of course normally be connected to a DSL router). This needs to have an extension aerial option, and also a Disable AP And Work As Guest (i.e. not as the usual Host) option, in order to be a viable solution. In other words, not all USB APs will work for this purpose. Mac laptop owners with an onboard AirPort card have done this, to avoid hacking the card and fitting the N connector. One advantage is that you get more power output.

Microwave Antennas

Up at microwave fequencies such as the 2.4 GigaHertz band that WiFi LANs work at, the aerials are so small that a different technology is used. Here, a quarter-wave aerial like your VHF masthead one is 31 mm long - so there's not much of it. Since physical size matters a lot with aerials, as you can simply pick more signal out of the air with a bigger aerial, some other solution has to be found. The answer is a dish aerial, as you will use for instance for satellite reception; or a waveguide.

Here is a simple example of how size always matters, whatever the dimension and whatever the frequency: if you made your masthead VHF aerial from thick pipe instead of a thin stainless steel rod, it would be much more efficient.

At 2.4 gigs, then, we need to use a bit more muscle than you can get from a 31 mm quarter-wave rod. As mentioned, a dish antenna is the best, as it picks up the RF sigs from all around and reflects them onto the widget placed back off the dish at the focus point your little tiny quarter or half-wave aerial element. Dishes have a gain of around 20 db or so, a massive signal increase. Compare this with your quarter-wave at unity (zero), or 2 to 3db, depending on how you measure it.

However, a dish is a little over the top for our purposes; though if you can get hold of an old satellite dish, it will suit admirably. You will have to change the central element for one to 2.4 GHz spec. Most will feel, though, that a 600 mm or 1-metre dish at their boat's companionway hatch (or house window, or car top) is going too far. There are better solutions size-wise, though nothing can touch the dish's gain.

The next step down from a dish comes in the form of a double-biquad + backplane reflector at a tested 13 db gain (a 'pantenna' or 'woktenna' in fact), then a biquad + reflector at a gain of 11 db, then a horn waveguide at 9 or 10 db, then a cantenna at around 8 db, then Yagis and beam helicals of various formats, then a co-linear, then a high-gain magmount or a 1/2 wave groundplane, then a 1/4 wave magmount and then your built-in WiFi aerial in the laptop or PC card. Phew.

We're going to forget all about those other types and make a cantenna, as it costs nothing except for the cable and connectors, works well, and is easy to make without fouling up the dimensions. At microwave, tiny mistakes in dimensions (or in the calculations made by someone giving you advice) can easily ruin the results. As a very good example of this, in theory a Yagi, a shotgun disc-element Yagi, or a big-mandrel helical (used end-on as a beam for microwave) all examples of beam aerials will perform well at microwave. This is true; but very few people have actually got them to work, as the dimensions are critical and unrelated to those normally found in the advice often given. Aerials built to the dimensions commonly circulated don't work. The need for absolute precision, the velocity factors of the cables and of the wires used, and even of the mandrel core material, critically affect the result, meaning in practice that instead of achieving a high gain, they are outperformed by a good cantenna. A Yagi is the form you see for TV aerials, with the directors stacked along the array. At 2.4 GHz, it's the size of a pencil with matchsticks on each side, and half a millimetre this way or that affects the results.

There are too many theorists giving advice who have never worked professionally in RF applications. A fine example of this is the often-seen advice given on how to make antennas such as halfwaves and quarterwaves, by calculating the length as 150 / Freq. Mhz.

Er wrong the Constant is unfortunately not 150...

Stage 1: Get Can

First we need a nice can. The dimensions are important, and the overall diameter is the critical factor. The ideal size of can is 92 mm diameter, but since we are unlikely to find one of that exact size, we must use the one we've got, or one easily found. A good bet is a baby milk powder can of 430 gm size, such as SMA Milk etc. This has a diameter of 98 mm (3.86 inches), a length of 125 mm (5 inches), and is fine for our purposes. It also has a nice plastic endcap that will come in handy if we use the cantenna outside in the weather provided the plastic is microwave-transparent, that is; we'll test this later. This is a big can as cans go much bigger than a baked bean tin for instance, which would be way too small. The milk powder can is a very similar size to a Nalley Beef Stew can available in the USA. This model tends to win out in WiFi homebrew antenna contests.

A larger can is better than a smaller can, for two reasons:
  1. As can O/D reduces, the rear standoff distance increases rapidly; therefore a placement error has greater consequence.
  2. The optimum can length also increases to an impossible degree (about 11.3 inches for a 3-inch can).

The Great Pringles Cantenna Myth

You can make a wave guide from any old round or square metal container with an open end. Pringles can tubes are sometimes used (they're not much good); the tetra-brik type 1-litre foil-lined milk and juice cartons work better. The Pringles tube is too narrow at this frequency, and also probably needs more metal in it. As well as being too narrow, it would need to be impossibly long at this frequency, as we have seen (and as you can calculate with our 'canculator').

You'll need, in addition to the can, a ruler and drill, with some drill bits of around 3 mm and 10 mm. A 13 mm or 1/2-inch drill bit is probably going to be too big, but this depends on your connectors.

Cables and Connectors

You need a pigtail to fit your WiFi PC card's extension aerial socket. This is a patch lead, usually of from 500 to 1500 mm long, having a tiny proprietary plug connector at the card end, and an N connector at the aerial end. For the popular Enterasys card, the connector is an MC Lucent type. With a long pigtail, you may not need an additional cable extension. With a short pigtail, you'll need another length of cable to get the antenna up or out where you want it. This cable needs to have an N connector plug on each end.

The cable type is critical. As we have seen before, the bigger or thicker, the better in cables, as in anything to do with aerials (and I know what you're thinking). The pigtail may well be in tiny thin 3 mm cable, and this cannot usually be avoided unless you are in the game and able to produce something yourself. Thin cable like this is not designed for such an application: open up your handheld GPS and you'll see it used to connect the onboard antenna with the electronics, a distance of around two or three inches. That's what it's for. Instead, any cable in the line here should be of a minimum 6 mm thickness; but usually isn't. For a run of more than four metres at microwave, you need 10 mm thick cable. Here are some cable specs, using as a base the excellent US-made Times Microwave cable. All RF coaxial cable is 50 ohm impedance.

Times Microwave LMR 100 cable: 3 mm diameter, 3 db loss per 1.2 metres. That's half the signal lost every four feet. A cheaper equivalent is RG174, which won't be as good.

Times Microwave LMR195 cable: 6 mm dia the standard solution. Use it. Equivalents: RG213, RG58, UR76.

Times Microwave LMR400: 10 mm dia, best for longer runs, thickness of your finger. 3 db loss per 12 metres at microwave; and that's about the best you'll get. Equivalents: RG8, UR67.


You will need an N-connector chassis socket for the can aerial element base, and to connect the feed cable. This has the screw connection on one end for the cable plug, a protruding centre pole for soldering to, and four 3 mm bolt holes.


The cantenna has two critical dimensions for the active element in the aerial. The can itself is a waveguide: the RF sets up a standing wave in the can and the active element is positioned at a point of maximum phase amplitude. The element, as we have seen, is 31 mm long, which is about a 1/4-wave at 2.4 GHz. It is placed at a distance of 44 mm out from the back face of the can. This dimension depends on the can diameter (not its length). It is calculated with the can diameter, and is 1/4 of the wavelength of the standing wave in the can not in free space (which is 31 mm). Different waveguide diameters (i.e. cans) need a different backspace measurement if you aren't using a 430 gm milk powder can, then:
   go to
our Can Calculator Page.

Purely as a matter of interest, a free-space quarter wave = 31 mm, 1/2 wave = 62 mm, and a full wave = 124 mm. These dimensions vary a little according to whether you are at band centre or either end of the band; there are 11 channels on the WiFi band. As we discussed, the free-space wavelength does not apply to a waveguide, except insofar as the active element is concerned (the quarter-wave element).

Get Drilling

Take your milk powder can and drill a 3 mm hole at 44 mm out from the back end. Then expand that small hole to 10 mm, using the larger drill bit. Your N-connector chassis socket should fit nicely here expand the hole slightly if it's tight. Drill the four 3 mm mounting bolt holes to suit. You can use PC case machine screws here, for instance; but you'll need the nuts & washers to fit. Place the screw heads inside the can, and the nuts outside, to minimise internal signal disruption.

A 4-bolt N connector chassis socket is preferable to the screw-on version (although this is easier to fit), because the latter sticks up too far into the can.

Solder a straight and rigid bit of solid copper wire onto the centre pole. A piece from some twin & earth mains electric cable is good here; the thicker the better. Make it well over-length, and when cool, cut it down to 31 mm total positive element protrusion. This is emphasised since the whole centre pole counts as the active element.

Don't allow the N-connector to protrude too far into the can, as it will mess up the signal; that's why you try to keep the hole as small as feasible, starting at 10 mm and reaming out to the minimum necessary.

End Cap

Luckily our milk powder tin comes with a handy end cap, which must be used if the aerial gets any weather. First, you must test to see if the plastic is microwave-transparent: if it isn't, then you will lose signal, and will have to find an alternative.

Method 1: with a microwave oven fill a mug half-full of water, and place it in the oven along with the plastic end cap. Just lay the end cap down on the rotating table, don't let it touch the mug. Cook on full power for one minute. I forgot to say, put a teabag in the water first... Take out your mug and plastic cap. The water should be hot, but the end cap cold. If the cap is warm then it cannot be used, since it is absorbing microwaves.

Method 2: with no microwave oven, but an open-scanner boat radar available tape it to the front of the radar antenna, and run the radar. Run it for a couple of minutes; the effect is about the same as with the oven. A radome-type scanner won't work, as it isn't putting out enough juice in the short time the beam hits the end cap, once per rotation.

Improving The Cantenna

You may increase the gain by fitting a horn extension to the front and therefore increasing the size of the aerial. This takes the form of a cone of metal, spreading out at an angle of 30 degrees from the can wall. Most people who fit a horn extension tend to stick with simply cutting another even bigger can up, and rivetting the cone produced onto the main can's wall. It isn't then too large to handle, being perhaps a foot or so long. This type is harder to weatherproof.

If you want to be precise, then the ideal total extended length will be two (waveguide) wavelengths long, at 356 mm (14 inches). This isn't too big to handle, and should give an extra 3 db gain not to be sniffed at, as this means about double the signal strength; though in practice the effect is not quite as dramatic as that.


The cantenna is of course directional, and also polarisation-conscious. It is vertically-polarised if the connection is at the bottom; horizontally-polarised if this is rotated up to the side. Point the can's open end at the AP (or at another laptop for a peer-to-peer connection). When you have locked on to a signal, then rotate the can up to 90 degrees each side, to see if you can improve the signal strength.

The results you will achieve depend on a lot of different factors:

1. Your WiFi laptop card's quality, and transmit output power. You will be able to receive WLAN base stations, but they won't hear you unless you are putting out a good signal. The AP may be kicking out from 1 to 4 watts, but you have only around 60 mw (0.06 of a watt).
2. The quality of your cable and connectors. A lot of signal power never reaches the antenna. With only 60 mw output power, this is critical.
3. The line of sight to the AP (Access Point) aerial. Obstructions will block the signal even a tree in summer can block the signal, whereas in winter, with no leaves, the signal gets through.
4. An obstruction very close to the line of sight, but not actually within it, can also have a negative effect; this is one aspect of Fresnel effects.
5. The distance involved: 400 metres is easily achievable; half a mile getting tricky; 1 mile is for A1 gear under optimum conditions.

It all depends; engineers have achieved a point-to-point range of 125 miles, using a couple of watts of power at each end, with dish aerials; enthusiasts using the best gear, with a standard AP but on an external mast, and with homebrew antennas, have achieved 15 miles laptop-to-AP; and you may get 1 mile under optimum conditions. Let's face it, 1/4-mile across the marina to your pal in the other boat, or to an AP ashore, is more realistic. Even this distance is not supposed to happen: the system is designed for within-office use, and no more than that.


If you can just about make a connection, but need a little more power, then you have one more option. Your WLAN card is splitting its output between the onboard card antenna, and the extension antenna not all the juice is going up the cable to the cantenna.

You can hack the card and cut the connections to the tiny internal antenna, for just a little bit more output. If you manage this successfully (and it's too detailed to go into here), then note that you must always have some form of external antenna plugged in at all times. If you don't do this, the card's RF PA is working into zero load and may blow (translation: the wifi card has a radio frquency power amplifier as its last stage, which is normally not designed to handle driving into a no-load situation). This is easily achieved though, by making a little halfwave dipole on the end of an MC Lucent plug, or whichever one your LAN card uses; the dipole will be 63 mm (2.5 inches) long, with two elements of 31 mm each, and a centre gap of 1 mm (and is just a little longer than the theoretical half wavelength at band center 2.432 GHz, of 61.5 mm). Theoretical measurements are not used in practice.

For more can dimension solutions, try our 'canculator' see link below.

Have fun!

cantenna pic
<< click thumbnail to open larger image in new window
(or FF: right-click 'Open in New Tab'
Build a 'Cantenna' Aerial