" Cantenna " - yagi design for 802.11b wireless application



Please Note: This page is under construction. This page was last Updated by me on Wed Mar 19, 2003. Thanks, Andrew S. Clapp

Please help. I'm unemployed! I'd like to work as a Unix Sysadmin, DBA or Programmer and stay in or near my home in Portland, Oregon. Please have a look at my resume. And while I'm at it, if you guys at O'Reilly would like it, I'd love to write a book about antenna construction. I could use the royalties, and you could use the profits. Whaddya say? :-)

What _IS_ Here? What's _NOT_ Here? (Index)

IS Here: NOT Here:
  • Antenna Sales
  • Network Security Issues
  • Information about 802.11a or 5GHz Equipment
  • Legalities of Use

Thanks, Credits and Disclaimer.

Thanks to Mark and his help with the first round of tests. And to The Feef, for his inspirational words taken from their original context, "I am the mad beamer, who beams at midnight", and his help with the distance testing on the Island.

Since the haphazard inception of this page, I have received well over 1000 emails from various folks in all corners of the earth, asking questions, seeking information or advice and offering tidbits of useful knowledge that all contribute to the success of this page. At it's peak there were more than 10,000 hits per day. I was completely surprized at the substantial interest for Do-It-Yourself antennas that was present in the network community. Thanks to you all.

I did not devise the term "Cantenna", but to the best of my knowledge, I did build the first one using the popular brand potato crisp can. I've adopted the use of the word simply because it's appropriate. Thankyou to the many folks who simultaneously discovered this word.

These antennas were the design of two Japanese people, Hidetsugu Yagi and Shintaro Uda, and are sometimes refered to as Yagi-Uda antennas. They were originally used for radio, long before modern computers. The CQ Amateur Radio Hall of Fame, http://www.cq-amateur-radio.com/hrhof.html is pretty cool, and is a good research starting point. This article from Alternative Wireless has some good things to say. http://www.alternativewireless.com/cellular-antennas/yagi.html

Interference, Health and Security Concerns

I'd like to say a little bit about Electromagnetic Waves. I'm sure you are eager to build an antenna or two for your 802.11b wireless network. Before you do anything else, it is important to think carefully about all the possible consequences of what you are doing. Simply put, you are going to be sending electromagnetic waves through the air. This can cause much distress to folks who are not expecting it, and when they get upset, they will call the FCC. So, you may want be knowledgeable about what you are doing before you do it. A few terms you should be familiar with are Radio Frequency Interference, Electro Magnetic Interference and Bandwidth Saturation.

It has been mentioned in this article that it is not legal to attach a non FCC-approved antenna to a wireless device. I suggest you read the FCC rules and regulations before doing anything. Seattle Wireless has a good collection. The antenna design I illustrate below is extremely experimental. I have heard that it's use could cause interference in near-band frequencies that are commonly used in things such as portalbe wireless (not cellular) phones that people may have in their homes near you. There are all kinds of wireless devices out there that operate around the same band as 802.11b, and these are potentially disrupted by use of the equipment described here.

You may consider purchasing an SWR meter and you may excercise much care, consideration and caution for others if and when you decide you need an RF amplifier for use with your antenna. For most applications you will not need one. I have also heard that if your antenna is too efficient, that you may even damage your 802.11b device with too much current/feedback. If you do not know what you are doing, study until you are confident that you will not break people, places or things when you start experimenting. I am providing this information for the sake of information and I am not liable for any damages, injuries or other accidental or intentional harm caused by the use of it. Play nice. We are all in this together.

As if this was not yet enough to keep you from messing around with fast flying electrons, I have received many emails from folks who are very involved with HAM radio and other professions and hobbies that involve work with high frequency microwave radiation. They warn that 2.4 GHz just happens to also be the resonant frequency of plain old water. This is why a microwave oven works. The energy of an 802.11b device is the same kind of energy that cooks your food, but on a much smaller scale. This is important considering that we as humans are 98% made of water. I have been warned that exposure to even as little as a 1/4 watt amplified with a 14db antenna, such as described here, could lead to severe vision problems and possibly other health issues.

I like the idea of using wireless to do cool stuff. It's really nifty. It's not very secure though. Nothing really is "secure". But wireless is even less so, because it is wireless. It does not require some guy breaking in to your home/business and tapping a physical connection for them to see the data you are sending. It's flowing freely through the air, like the sound from you stereo speakers. If someone was standing outside your house, and you were rockin' out with your stereo up loud, they could hear it as well.

I do not trust any connection I make with my 802.11b network, so I do not send anything I would not want others to know about over the air waves. This however, that would be silly and impractical. I encourage the use of WEP and SSH. It may have bugs and it may be weak encryption for some value of weak. It seems to do the job though. It does add an extra quantitiy to the time it would take some one to break in. Here is some interesting info from ISAAC on WEP. I have heard of, but not yet used, Cisco's implementation of 802.11x, which has all kinds of improvement on the security of your wireless connections.

History of the Pringles Yagi "Cantenna"

I looked at this cool Seattle Wireless page with wonderful pictures of a manufactured directional yagi antenna and then I did some math and built one as much like it as I could afford. When I was done, I had the basic model for the antenna, but 36" long. Both are similar in construction and materials, and identical in theory. I just scaled down the materials and made a smaller one, and the potato crisp can just happened to be about the right size and it worked out. Another good point to mention, is that for smaller pcmcia powered devices, the smaller antennas worked better. I believe it's because the ammount of energy required to drive the larger antenna efficiently exceeds the 1/4 watt produced by my Lucent Orinoco card. Anyone know if that is the case or not? I suppose it could be calculated using the mass of the antenna and Plank's constant, but I've not done that . . . yet.

Simple Alternatives

A simple way to make an antenna is to build a "di-pole". These are simply a piece of wire that with a length that's a even multiple of the wavelength. These can then be cut in the center and attached to a piece of coax which is run to your wireless card via the nifty-but-expensive "pigtails". I have used many di-pole antennas for HAM radio projects. A disadvantage is that they can take up large ammounts of space. On the other hand, they are very easy to make.

Buy one. Yup, if you want to have an antenna and you don't want to build it, you should buy one. I'm sorry, but I won't build one for you. :-) There are many places out there that make antennas for this kind of thing. This is a page for the adventurous person who wants to build an antenna. Buying an FCC approved device is also a very good idea.

Simple Antenna Theory

Waves
Imagine a boat making a wake in a river. As the wake laps up against the poles of a pier, it imparts a vibration of the same frequency to the flat boards in the pier itself. If you put your ear on the pier, you can hear them quite loudly because the pier amplifies the sound at the water level. The frequency is very low, because the length between waves is long. It can be up to a few seconds between waves. This length is called the period of the wave. With wireless, the waves are much shorter, the period is less than a second, much less. It is because of this that we name the wave by it's frequency. For 802.11b, it's called 2.4GHz, because the wave repeats 2,400,000,000 times every second.

Resonance
An antenna is an amplifier in a sense. It is the antenna that catches the signal from the air, much like a sail of a ship catches the wind. Any piece of metal will serve as an antenna, and the bigger the better if you use the sail/wind analogy. But regardless of size, some work much better than others. Why is this? Resonance. Some bits of metal are the same size in one dimension as the radio wave it's catching. This causes the wave to be felt stronger by the metal, resulting in the a stronger electrically induced current. It actually works well if you are some even multiples and fractions of the wave's size. Quater wave length segments are very common and useful.

Frequency Bands
We need to know the electrical characteristics of our signal in order to understand an antenna. These are the wave's frequency, wavelength and amplitude. The frequency is inversely related to the wavelength. 802.11b uses several frequencies that are close together. The range as a whole is called a band. Because they are all close to 2.4GHz, it is generally referred to as the 2.4GHz band. It is the begining of a band of frequencies on up to 2.4835GHz. A useful fact here is that one antenna can and does serve for more than one frequency, but generally the ones in the middle will work better than the ones on the ends. You should design your antenna around the middle of the band you are working with so that it will work more efficiently.

Light Speed Ahead
The frequency is the number of times your carrier signal cycles or repeats in one second. The wavelength is the distance your signal goes in one cycle. To calculate wavelength for 802.11b, we must know the speed of the signal. Electromagnetic waves, our radio waves, travel at the speed of light. That means the velocity is about 3.0 x 10^8 meters per second. It's a constant. This means that distance it goes per second will always be the same. One can imagine a sine wave flying through the air. This picture shows one cycle of a sine wave. No matter how long the cycle of the wave, it always goes the same speed, and therefore the same distance in one second as any other electro-magnetic wave. Now that we know the velocity , the frequency and the band or our signal, as well as a little bit about resonance, we are ready to do some calulations.

Here's a really nifty sine wave applet that I found on the web.

And a handy unit conversion table that I also found on the web.

This simple formula is from a high-school physics book.

Rate * Time = Distance

Or for our special case,

C (the speed of light) * Time = Wavelength

Notice how the time and the distance portion of our little equation are related. The time we are concerned about is the time it takes our radio wave to complete one cycle. This time is called the period of the wave. This will typically be measured as some small fraction of a second. Since the measurement of frequency is the number of cycles in 1 second, the frequency multiplied by the period will always be one. If we know either the frequency or the period, we know the other as well. We use frequency to name a wave because it's a number larger than one.

Period (length of time for 1 cycle) * Frequency (cycles per second) = 1
or
T * F = 1

The Frequency of our 802.11b wireless equipment is 2.4 x 10^9 Hz.
Wavelength is the distance traveled in one period, or the speed of
light times the time it takes the wave to complete one cycle.  And
the period times the frequency is equal to one.  So now we have two 
equations with two unknowns.  Sound familiar?  Good, we all did this 
in grade school right?  It's basic algerbra to solve this problem.

Equation 1: W [meters] = 3.0 x 10^8 [meters/sec] * T [sec]

and

Equation 2: T [sec/cycle] * 2.4 x 10^9 [cycles/sec] = 1

Equation 2 in terms of T is: T = 1 / (2.4 x 10^9)

Using the first equation in terms of W, which it already is,

W = 3.0 x 10^8 [meters/sec] * (T)

and substituting the second equation in terms of T, 
for T in the first equation we have,

W [meters] = 3.0x10^8 [meters/sec] * (1 / 2.4x10^9 [cycles/sec])

W = 3.0x10^8 * 2.4x10^-9 [meters/cycle]

= 3.0/2.4 * 10^-1 = .125 meter, or 1/8 meter, or 12.5cm

The same calculation for 2.4835GHz, the top of our band, yeilds 12.0797cm, which rounds nicely to 12.0 for now. So, we are dealing with waves that are between 12.0 and 12.5 cm in length. Cool, huh?

Now that we know our wavelength, what do we do with it? Well, remember that we should design our antenna for the middle of the band. This would be something with a wavelength of 12.25 cm. I've made them with all kinds of lenghts from 11.8 to 12.7, due to the tools and parts I have access to, and the ammount of time I'm willing to spend filing down spacers, and they all seem to work well. But for our calculations, we should try to be exact. The more effort you put into your antenna, the better it will work.

Finding Parts (Hardware, Pigtails, etc.)

If you use a device that has a spot to connect an antenna via coax cable you will need to get a pigtail that will let you get the signal out of it and into something useful, like your antenna. A pigtail is just a coax wire with some custom connectors on the ends of it. Many cards out there use custom ends that you may find at this place, Electro-Comm Distributing. I believe the part number for a Lucent Orinoco to N pigtail is RMC-6010-B (Thanks Garry A.).

Many of the antennas out there use what are know as N Connectors for the jack that you plug your pigtail into, so that's what I used. I found my N Connectors at a local electronics store called Radar Electronics, www.radarinc.com.

The rest of the antenna is really just PVC or ABS plastic tubes, metal washers, metal rods, a can of some sort. I acquired all this at a local hardware store. Between Fred Meyer, Home Depot and the best True Value store on the planet, I found more than adequate supplies.

Parts List

This model is the one I suggest you build.

18" directional/omni, 1.25" dia.

This model was the first one I made.

36" directional/omni, 2" diameter

Building a Cantenna

The Basic Goal
Here are some Cantenna measurements to help give an idea of what we are trying to build. The 18" version, uses a small 1-3/4oz pringles can for a reflector. The can itself is layered inside with a metallic film and a thin plastic-like coating. The bottom of the can is metal and should have electrical contact with the sides of the can. All the ones I tested did. If you can't find the little ones, you can use the big cans, for this I cut the can to about 4 1/8" tall the first time around. I still need to investigate what good length would be in theory, but the longer the reflector, the more directional and less omni your antenna will be. You don't need to cut the can at all. You may get different results. The can portion of the antenna is called the reflector because it reflects the signal back out of the can.

You will use a pigtail to connect the can to your wireless card. There is an N connector mounted in the side of the can centered at 1-3/8" to 1-1/2" up from the bottom, and I found that small zip ties secured it nicely in place. Also, soldered to the center of the N Connector is a small piece of metal rod or tubing that I call the injector.

The whole assembly of spacers, washers and metal rod is called the Collector. The tubing spacers slide over the metal rod to seperate the washers. In creating the spacers for my antenna, I intentionally errored on the larger rather than smaller side of the calculated middle-band, quater-wave size. I did this because I used a hacksaw to cut them quickly, and that makes less exact spacer tubes, with sometimes even slanted cuts, and that's no good. I used a file to flatten out the ends and bring the aluminum/copper spacers down to a more precise size/shape.

Ideally, when you know what frequency your particular device is, you can build accordingly. All models with wavelengths within 12.0-12.5cm should work out well for 802.11b. This is not rocket-science. So don't stress if you are experimenting, just have fun.

I do not know much about the design of the reflector portion, but they seem to give about 12-14 Dbi. Attached to the center pin of the N connector is a metal rod, soldered into place that points straight into the center of the can, just shy of, but not touching the threaded rod with the washers. The threded rod should extend down past the pvc in wich it is housed at least 1/2" so that the space between it and the N connector's extension is very small. The PVC rests against the extended center pin of the N connector, but not with any weight, as it is held in place by cardboard or styrofoam between it and the can.

For the math, assume you are working with the frequency 2.4GHz. Our quater-wave length is .125 / 4 = .03125 meter. Remember this is the bottom of the frequency band. It's the "longest" end, for a margin of error that is corrected in the filing of the spacers, and not the middle of the band which would make the most efficient and well tuned antennas.

I chose to use materials that were most economic and readily available for my construction. If you want to spend money, you can go buy a parabolic grid antenna. Please send me one too. :-) If you want a fairly good antenna for a fairly reasonable price, go to Home Depot or your local hardware store and buy something like the following materials. I encourage you to try different things. Mine worked, but there are surely better ways to do this.

The Collector
Cut the tubing (using a hacksaw) into spacers that seperate the washers when placed alternately on the rod. I spaced them quater-wave fashion and I get somewhere between 10 and 15 db gain. You could use other lengths or combinations of different ones. I've seen many different designs that work.

Spacing between washers

It is true that the longer your antenna, the more gain you will get. More gain == good(tm). I've read somewhere (and it may or may not be the case) that the gain increase levels off rapidly after 6 or 7 wavelengths and it's not important to design beyond that. The 36" rod is (39.3700 inches per meter) about .9144 meter or a little over 7 wavelengths long. Overkill is a way of life.
Whole Antenna

The Reflector
For a reflector, you will need to experiment. I've gotten away with such things as a planters cashew nut can, a chicory coffee can, a pirouline can, a pringles can, and I'm currently looking at a old lighting fixture can because it looks neat. I've heard somewhere that some sort of important ratio for the reflector portion is .2 wavelength, but I don't recall exactly what that was. I used the pringles can for the smaller antenna.

Cashew Can


Pirouline Can

The Injector
This is the small metal extension to the N connector that is mounted in the side of the reflector. The extension does not touch the threaded rod. In fact, my tests show better results with a gap between the extension and the rod. Several Layers of Cardboard and Duct tape hold the abs pipe in place.

Reflector

Tuning
Here are some things to try to get better gain from your antenna.

Passive Repeater Construction

Using the same parts as two 18" directionals, minus the reflector can, you can build a thing that at least in my first tests, appears to bend the signal around objects it does not go through. This is not the best solution, but may work in a pinch, if your situation is demanding that you get arond something that is in your way, and you don't have very far to go. I have heard of others making this kind of thing work, but they were using two professional dish antennas and LMR400 coax between them as a way to change the direction of the signal. Regardless of the implementation, you will loose more than half of your signal strength. The power goes into exciting all the metal in the middle of your path rather than getting the signal where you want it. BTW, I'm told it's not a wave guide, it's a passive repeater.

To build one, take two of the directional portions of the 18" antenna, which are the parts made from the washers, metal rod and pvc with cap, and fine a 90 degree, 125 degree, or T pvc/abs connector that attaches the ends of the tubes of the directional elements. The elements do not actually touch in my first test, but they are seperated by about .125"-.25" with nothing but air.

As you know, a capacitor is just two pieces of metal with an insulating medium between, like air, so this contraption has a theoretical RLC resonant frequency, Q. If you could tune Q to something nead the center band, it might work better. One thing I've not tried yet is making the connection in the middle with a capacitor calculated to resonate at the band median to provide the impetus for the carrier to be sustainably resonated with much less power consumption from the end points. It's tricky and very speculative. Maybe a full LC in the middle would work. More on this, anyone?

Test Results

My test connection was made between an Airport with an internal Lucent silver card and IBM Thinkpad 600 laptop with a lucent gold card. I tested each end with and without an antenna that I built as described below. The laptop is currently running win2k, and I'm using lucent's software package for measurements, as well as Net Stumbler, for the gps distances. As soon as I get time to build some custom kernels, and possibly an additional drive for the laptop, I will be running NetBSD. Then we'll see more extremely nifty things.

Airport Antenna Laptop Antenna Distance Highest Rate Ping Times Signal Lucent Bars
None None 30 ft 11 Mbits <10 ms -40 dB 5 green
18" yagi 18" yagi 200 ft 11 Mbits <10 ms pings -43 dB 5 green
36" yagi 36" yagi 1000 ft 11 Mbits <10 ms pings -85 dB 5 green
18" yagi 36" yagi .7 Mile ? Mbits ? ms pings -89 dB 4 yellow
36" yagi 36" yagi .7 Mile ? Mbits ? ms pings -93 dB 4 yellow

I am currently suspecting that the higher precision I put into making the smaller antenna paid off. Or, another possibility is that because there is less metal in the smaller antenna, it is better for this low current application. :-) All my tests so far seem to place the gain on the 18" model at 12-15 dBi. This is really good for under $20.00, and I don't think you can beat that comercially!

Now for the repeater, with the first test, I used the same equipment as before, but seperated the laptop from the Airport by a non-2.4GHz-tranparent object (a house, a wet brick wall, etc...) and pointed them both at a right-angle assembled pair of elements.

Second tests were thwarted heavily by not being able to get the antennas on opposites sides of a dense enough object. However, using a piece of wire to Attach the closest ends of the antennas seems to be beneficial. I would like to test the concept of placing two full antennas with cans connected via LMR at the center point.

Airport 18" Antenna present? Airport (SNR/S/N) Distance A Right-angle "Repeater" present Distance B Laptop (SNR/S/N) Laptop 18" Antenna present?
Y 61/-35/-95 N/A No, Laptop pointed directly at Airport about 5 feet N/A 60/-35/-95 N
Y 52/-45/-95 N/A No, Laptop pointed directly at Airport about 20 feet (site for repeater) N/A 53/-43/-95 N
Y 51/-46/-96 N/A No, Laptop pointed directly at Airport about 20 feet (site for repeater) N/A 49/-49/-97 Y
Y (redo test) 20 No, Airport pointed empty repeater site about 20 feet, Laptop at 100 feet 100 (redo test) Y
Y 2/-95/-95 20 Yes, Airport pointed Repeater about 20 feet, Laptop at 100 feet 100 8/-88/-97 N
Y 24/-74/-97 20 Yes, Airport pointed Repeater about 20 feet, Laptop pointed at repeater at 100 feet 100 21/-74/-96 Y

Someone from the seattle wireless list wrote:
There's some discussion about passive repeaters on deja. I'm just repeating (um, no pun intended) what I read but it looks like the receiving antenna will re-radiate 50% of the received signal back to the source. There will also be a bit of thermal loss and loss from the coupling system. So the passive repeater will basically re-transmit somewhere under 50% of the original signal. That means we'll lose just over 3dB which probably won't hurt many links unless they're already strained.

After some math (plus some dumbfounded looks and obligatory cursing) it looks like we would lose about sqrt(2) / 2 of distance if we put the passive repeater in the mix (assuming 3 dB loss). So if you had a link that worked nominally at one mile you'd now need the endpoints 0.7 miles apart to get the same signal level. Will be interesting to see how the real world works.

Pictures


If anyone has anything to contribute or ideas to share or things to improve the design, please email me. :-) I'll keep this page updated with any new and useful data.

Thanks to Jeme for the use of his cool Olympus digital camera. And to Ricky for hosting the page until his server dies or his connection saturates (whee big pictures). Thanks also to Mark for inspiring the original beta test. Thanks to the gang at Seattle Wireless for the good antenna/airport info and pics. Thanks to Feef (the Mad Beamer, who beams at midnight) for the late night testing on the island. And thankyou great masses for all the feedback and comments. This page is getting bigger, and hopefully more accurate. Also thanks to Imperator Smod for the help with the most recent batch of pictures. I believe that only together could we have perfected the art of getting the blurryest possible image from those little disposable cameras.

Links

Well, many folks have good ideas. Here are some links to other sites:


clapp@aeonic.com -:- Home -:- Resume