Homebrew Wifi Antennas
Posted 28 October 2012 - 07:37 PM
The only issue I have with these projects, is trying to classify them. I end up with all this crap under tech talk because it really doesn't fit well into other places. Anyhow, here goes.
This project is a bit more "simplistic" than my other projects. Namely, in that it isn't nearly as grandiose. What I lose in flare, I hope to make up for in practicality. Today, I want to discuss wifi antenna, how to make your own, where to find decent information and discuss what works and what does not. The bright side here, is that there is a LOT of flexibility in this project. The downside, is that radio work (of any kind) can get expensive quickly. So I will take you down the road needed to do this as cheaply as possible. Understanding that once more, I am a cheapskate at heart. As of right now, the project is incomplete, but very much in progress - and necessary. I will be discussing not only how to make it work, but how practical it REALLY is. I intend to find not just signal numbers from my equipment, but also get throughput testing done, as well as pinpoint additional delays. The goal here, will be to see if this can be practical for people in need of real range.
So, first up, the setup. I have 2 WRT 54G routers. They are old. They are finicky. And now, they run Linux. Sadly, I am stuck using the micro build on one of them, as it is an oddball version with a 2MBeeprom. In any event I will attempt to use the V5 router for now, though it will be on its way to Ohio as soon as the first round is done. The goal here, is to get this router to communicate with another router about 500meters away. Excessive speeds are not needed, in fact a rate of 11Mb/sec would be more than enough for this discussion, though I will of course, be trying to much more.
A quick bit of poking around on the internet turns up a few obvious places to start, namely the Pringles can-tenna. What you will find out in the end, is that the cantenna is A| too short for practical use, and B| actually a very complex Yagi style antenna. These are excellent directional antenna, but due to the limited space of the can, relatively ill equipped for the target goal. So for now, we will be discussing a wave guide style antenna.
So during my research I found a website that I would love to credit with 99.9% of the information that I will discuss about the antenna itself: Turnpoint.net
That site was used to develop the idea, and give me hard numbers to use for the project. They do, however, leave out an important detail: the cable. That will come up shortly.
Understand, that for now, I will NOT be using Type N for my RF connections. Firstly, because I don't really care if it is a standard connection, and secondly because I want to be able to run a few tests later on using a scope or two. For these reasons I will be using a BNC connection on the can. For obvious reasons, I am forced to use an RPTNC connection on the router end. Double check your gear before you buy ANY connectors/cables.
Important note: The wifi routers sold use a non-standard connection for a reason! This is to comply with FCC regulations regarding part 15 rules. Namely that you cannot transmit more than 1 watt peak to peak power. A modified antenna WILL change this number! For more details regarding the power limitations, see here. The FCC does allow limited unlicensed "testing". However, I will be broadcasting using my amateur radio station license during these test, and you will see my equipment identified as such. As a result of my licensed testing, I will not be testing for encryption verification. so remember that all these numbers will exclude WAP2 protections. Sorry, but that is how it goes sometimes.
OK, on to the nitty gritty. First off, the can itself is not all that important. Just find a can that is about 3.3" in diameter. I am using a pair of HUNTS Pasta Sauce cans. They are about 3.25" in diameter, and roughly 6" long. These seem to be about the cheapest cans I could find (about $1 each!). Open only one end. That is important, and we will discuss why later (it is safe to say, I already screwed this up! ). Then drill a hole in the can 2.49" from the back of the can (the closed end). This hole needs to be large enough to fit in the RF JACK. Afterwords, solder in a piece of SOLID CORE 12gauge wire to the center pole of your jack. That wire needs to be 1.21" in length. Pictures will come shortly detailing the measurement points. That is it - the antenna is done.
Then comes the cable. This is where it gets interesting - at least for me. Looking above, we can make a pair of antenna for about $10 total. That right there can easily replace $100 in commercial products! But what about the cabling? Do you really need to spend $50 on a 30ft cable to run from your router to the antenna? I say no. I am working on the feed line right now, but understand that different cable types have very different properties. Generally speaking, RG8X would be the preferred cable - and keep it short! RG58 can be used as well (and will be what I test with initially). DO NOT use RG59 OR RG6. Those are impedance matched to 75 ohms, not 50. As such, you will likely damage your equipment! The last option I want to mention is LMR400. Without 'n' type connections on both ends, I wouldn't mess with this. Also, this would be used more for LONG runs, meaning more than a few feet - and for permanent installations. It has FAR LESS LOSS than either cable type I will be using - but is more difficult to work with. Remember to be very careful in buying your connectors, to ensure they are the right polarity AND for the right type of cable. In the end, my basic set of cables will end up costing about $20 for the pair at 12ft long. This should result in a cable with about 3.2db loss.
Pictures, test results, and additional details will come in the next day or two. Requests will also be taken.
Posted 29 October 2012 - 11:06 PM
Posted 30 October 2012 - 05:56 PM
Posted 30 October 2012 - 06:30 PM
I have very limited spaces to work in. The final product will be sent to Ohio, where there is plenty of room for testing, and VERY little interference. I am stuck in Texas with very little room to test in, and a LOT of noise....
Posted 30 October 2012 - 10:03 PM
I honestly don't know what to make of all this. Don't get me wrong, I know what was said, I just don't understand how you tested for these things. Maybe after I take some time to test, I will better understand what you mean.
Posted 31 October 2012 - 10:21 AM
Posted 06 November 2012 - 03:09 PM
A failure is obvious, but why it failed is not so obvious.
First, I have to note that I lost one of the original rubber ducky antennas that came with the router. This makes direct comparison quite difficult. But the real questions come down do - how does my netgear manage such incredible range?
So here goes. I pulled out a Dell D600 for this test as I can use my trusty Cisco Aironet A/B/G card. No fancy 'n' stuff here. This card has one of the most sensitive radios I have ever had the pleasure of using. Yet for some reason, I am hitting limits that should not exist. The equipment never dipped into 'basic rate' modes (a or <img src='http://forums.pcworld.com/public/style_emoticons/default/cool.gif' class='bbc_emoticon' alt='' />, yet never seemed to break 1.5MB/sec. All the signal numbers were taken from the client side with an active large file download.
So here are the results, at about 40 feet away (I do live in an apartment after all), signal strength was as expected. -57dbm with a noise level of -95dbm. This gives us a usable 38db signal to noise ratio. Again, at this stage, I was seeing a sustained 1.5MB/sec download.
From that same location, I turned the antennas first to their side, then facing away. I was hoping to see a significant signal drop indicating that the wave guide was, in fact, working. From the side: -65dbm signal, -96dbm noise. 31db signal to noise ratio. Ok, this means we have are at about 1/8th the original signal power. Which means that the cans are doing an excellent job focusing my signal, and reducing side emissions. During an active download, the signalling rates dropped as low as 18Mb/sec signalling, with an average throughput of 1.2MB/sec. From the back of the antennas, I recorded similar results as the side. No significant changes.
At this stage, the question became - what kind of distance can I expect? That answer is where I am confused. I don't have an exact distance right now, and will try to fix that when I can. But it is approximately 500 feet away that I ran my test from my truck. With the cantenna, I could not receive a usable signal. I was getting ready to blame the metal in my truck, when I noticed another signal from my apartment... the Netgear. Understand, the netgear has two little rubber duck antenna, and a patch antenna. Yet somehow, 500 feet away, I had a signal. Not a strong one mind you. It was a whopping -81dbm with a noise floor at -087dbm. But with 6db of signal room in there, that Cisco card latched right on. It was only good for a 2Mbit link, and a download speed of 15KB/sec, but the cold, harsh truth is that machine connected to gear that shouldn't have hit that far out, while not being able to even see the Cantenna.
So... where to go from here? I am not sure yet. I will have to do some research, and see if I can create an actual high gain antenna, not just a highly directional beast.
This post has been edited by waldojim: 06 November 2012 - 03:11 PM
Posted 06 November 2012 - 04:35 PM
Posted 06 November 2012 - 04:39 PM
The question I have here though, is why does the antenna act the way it is supposed to, though without the range? Or are you saying that I am coming across interference too close to the waveguide?
This post has been edited by waldojim: 06 November 2012 - 04:40 PM