Designing a 915MHz Yagi Antenna on the Cheap
In a previous post, I discussed the possibility of linking my house’s network to my buddy Doug’s house, 6.5 miles (~10.5km) away. Using 2.4GHz technology with 200mW and 14dB yagi antenna’s didn’t do the trick. I wondered, though… could I utilize wireless networking in the 900MHz range? 900MHz can definitely go through trees and what not a lot better than 2.4GHz. I heard about a ship-to-shore link that used both 2.4GHz and 900MHz wireless networking technologies… once the 2.4GHz signal disappeared, the 900MHz network link would kick in and give the ship some extra distance. (Of course, though, that was over open water! I need to go through trees, houses, and bounce off rocks.) O’Reilly’s Wireless Hacks book states, “As the frequency of a signal increases, the apparent range it can cover at the same power and gain decreases. For example, a 100mW signal at 5.8 GHz appears to travel less than half the distance of a 100mW signal at 2.4 GHz, which appears to travel less than half that of a 100mW signal at 900 MHz.” (Wireless Hacks, Hack #7). I also read a post on Broadband Reports where someone was able to shoot a 900MHz link 9.7 miles through thick pine trees.
Having a private link to my buddy’s house that doesn’t eat up my precious 0.8Mbs DSL bandwidth would be nice. Wireless network equipment like the Lucent/NCR WaveLAN (or DEC RoamAbout) 915MHz cards would work. So would the self-contained (Cisco) Aironet Arlan 915MHz access points and bridges. (These wireless products utilze ~22MHz of spectrum for attaining 1-2Mb/s throughput, usually between 902MHz and 928MHz. 915MHz is considered the midpoint.) In order to utilize any 915MHz wireless networking equipment, I need some 915MHz antennas. 915MHz antennas aren’t exactly that cheap especially if I want one with around 10dB gain (around US$100 apiece.) I turned to the amateur radio community on the Internet, to see if I could build my own cheap yagi antenna. Total estimated cost: US$10 for two antennas.
I chose a yagi antenna design due to its simplicity and that it’s directional. I don’t want to build an omnidirectional antenna; I am trying to create a point to point link, and to do so, I need an antenna that can focus a beam in one direction, and cut out most of the noise coming from other directions. To understand how yagi antennas work, I read How to Build an Antenna from Sky Scan. To build a cheap antenna, though, I discovered a very informative article, Cheap Yagi Antennas for VHF/UHF by K. Britain and J. Maca. It helped me understand exactly how far I could go with plastic PVC tubing (or wood), wire, and coax cable. I discovered that with a little computer modeling and a trip to the hardware store, I could build a yagi antenna that does what I want it to do — have a gain of ~ 10dB, resonate at 915MHz, and be CHEAP.
Modeling the Yagi
After trying out a few antenna modeling programs, I attempted my first design on a DOS program from 1991, YagiCAD at the TEARA website. (I also downloaded one of the YagiCAD examples from that site; the author’s example for a 10 element 900MHz yagi on a 1.05m boom.) From what I’ve read and what I saw in the different calculations, 10 elements in a yagi is a “sweet spot.” Adding more elements past the 10th element only increases the gain marginally. I also decided on using a folded dipole for the driven element based on the “Cheap Yagi…” article. The size of the gap in the dipole’s bent element determines the feed’s impedence. The larger the gap, the higher the impedance. Using the TEARA web page’s 900MHz 10 element example, I used YagiCAD to scale the antenna design to 915MHz. I also compensated for 1/8″ (0.32cm) thick (also known as 8AWG or “8 gauge”) solid wire for the elements.
Testing the YagiCAD Design
From the dimensions YagiCAD produced, Doug and I built our yagi antennas using PVC pipe for the booms. The PVC pipes were very difficult to drill through with a hand drill (I’m sure a drill press would’ve been easier…) The elements ended up being a bit uneven. I’ve read that the elements can be tilted or outside the element plane with an error of 45 degrees… which is a lot! The Yagi design is a bit forgiving in this aspect. Apparently, a vertical polarization of the antenna would require the elements to be more in alignment with each other. It’s the spacing of the elements that is very important, and needs to be as correct as possible. After some initial tests with NCR WaveLAN cards, we were getting a better signal when the antennas were pointed in opposite directions! Meaning, there was better gain coming from behind the yagi. This wasn’t good. Back to the drawing board.
Designing a “Better” Antenna
After some additional research, I found a better yagi modelling program, WA7RAI’s Quick Yagi. This program is far better than YagiCAD — it can auto design and optimize a yagi antenna. To see how Quick Yagi modeled YagiCAD’s data, I manually entered the YagiCAD data I utilized for the PVC antennas. Even though the YagiCAD plotted it’s data in a nice area pattern, Quick Yagi thought differently. No wonder I was getting a good signal from the back of the antenna! Not only that, but YagiCAD stated this antenna should have a gain of 15dBi, and Quick Yagi computes a gain of ~ 5dBi! (I used a DOS program, Screen Thief, to capture these plots.)
I believe another problem with the PVC antenna was the folded dipole. Since it was difficult to drill holes in the PVC, the gap of the folded dipole was tapered. The holes in the boom need to be drilled so the gap in the dipole can be even. As stated before, it’s this gap that determines the feed impedence, so, using a wood boom to drill even holes will help in this aspect.
Utilizing Quick Yagi’s “auto” features, I had the program design a 10 element antenna with a folded dipole. Since I will be directly soldering some 75 Ohm cable TV coax to the folded dipole, I need to make sure the antenna’s driven element (the folded dipole) is 75 Ohms. To do this, I set the gap of the folded dipole to 4cm (40mm) and adjusted the length of the fed element until the feed impedence reached ~ 75 Ohms. Quick Yagi produced this data (where I rounded off the numbers) and this plot. The new plot definitely looks a lot more promising than the YagiCAD data. Quick Yagi created a ~ 12dBi antenna with a Front to Back ratio of 29.21 dBi. (The higher the ratio, the better, as this is the ratio between the gain in front and behind the antenna.) Below is a table of the antenna’s elements from the Quick Yagi data:
The gap distance in the folded dipole is 2.0cm. The distance from the Reflector element to the last (D8) element is 53.2cm. The SPACING column states the distance from the previous element. The Cumulative Spacing column states the distance from the reflector to that particular element. The LENGTH column states the length of the element. The diameter of the elements is constant at 0.4115cm (4.115mm) or 6 gauge wire. (The previous antenna I built used 8 gauge wire; I changed to 6 gauge wire as it seems to be easier to find in hardware stores.)
| ELEMENT | SPACING (cm) | Cumulative Spacing (cm) | LENGTH (cm) |
| REFLECTOR | 0.0 | 0.0 | 16.6 |
| DRIVEN (Fed Element) |
4.1 | 4.1 | 16.2 |
| D1 | 3.3 | 7.4 | 14.8 |
| D2 | 3.8 | 11.2 | 14.2 |
| D3 | 4.5 | 15.7 | 13.7 |
| D4 | 5.3 | 21.0 | 13.2 |
| D5 | 6.2 | 27.2 | 12.5 |
| D6 | 7.3 | 34.5 | 12.1 |
| D7 | 8.6 | 43.1 | 11.5 |
| D8 | 10.1 | 53.2 | 11.2 |
Next… building this antenna, but with a wood boom for drilling better element holes.
October 26th, 2005 at 10:10 pm
Getting inspiration from your article, I have also built an antenna using the “Cheap Yagi . . .” 902/903 MHz design. I just filed shallow slots on the pvc pipe for the 1/8″ bronze welding rods elements. Then glued them using gutter adhesive/sealant and further tied them using plastic ties that you pull to tighten. I used it as an external antenna for my GSM 900 modem.
The variations that seems to make a difference are
1) tightness of the elements on the boom
2) keeping all the elements perfectly aligned
3) losses on the feed line from the antenna to my GSM modem (had to join two different coaxial cables)
Slight variations in length of elements and spacing of elements did not seem to matter so much in my experiments.
The results are good. When properly oriented (vertical polarisation for GSM 900) and sited, it gives much better signal strength than the small 3 inch whip antenna provided.
The bandwidth is not too bad either. Could also use it for TV (600 MHz) and CDMA phone (800 MHz).
From hindsight, I will agree with you that using a square wooden boom would be an easier way to go.
September 11th, 2007 at 6:08 pm
I have my doubts about your formula for diminishing returns on higher frequency transmissions. While the lower frequencies may have their advantages I find it highly unlikely they have twice the range. If anything I’d bet higher frequencies are better at penetrating obstructions while lower ones travel further but are not as good at moving through objects. Perhaps this is the more realistic explanation of why you’d see ships using lower frequency for better range.
The fact is that technology companies are producing higher frequency devices for a reason. 2.4 ghz cordless phones for instance are designed for longer range and better reception over their lower frequency counterparts. Higher frequencies are also likely less prone to interference from existing electromagnetic waves. Also most cellular providers are using the higher frequencies for their future data networks. Sprint for instance, which is known to have the best data network here in the US, commonly uses 1900 instead of 850. They’ve also rolled out their data network to more places than verizon, so they have more coverage and better speed on a 1900 mhz network than an 850.
I think in unobstructed cases lower frequencies travel better. That’s why we’ve all heard the saying that bass travels further than treble. Therefore people with lower voices can effectively yell further, but I highly doubt this works out in the real world to a double the range situation as you suggest. It would be more or less unthinkable to use the higher range frequencies for anything if that was truly the case.
A 9 mile link shouldn’t be too hard. I’ve seen people do 20+ miles in the mountains. It’s mostly about the obstructions in your path not the total distance. Have you seen these antennas ? http://www.hyperlinktech.com/web/hg2424g.php
This antenna would be more geared toward what you want to do, though the beam could be too narrow for you exact application.
I vastly more likely your yagi setup was just flawed than it is that the lower frequencies are the reasons you have a connections. I think you’ll find in the long run making your own antenna is not cost effective. Antenna’s are very cheap and made for long term use. DIY antenna’s have high failure rates as they tend to not last through adverse weather conditions over the years. Yagi are nice but they I see more people using parabolic or panel antennas for these types of applications.
In the real world application you’ll never see a greatly reduced distance for lower frequency connections as you suggest. The tighter waveform of a higher frequency wave effectively makes it better for directional applications, while lower frequencies CAN be better for broadcasting style applications. This is why long distance wifi connections are usually 2.5 and not low frequency. Over ground based obstacles as I guessed lower frequency waves are theoretically not as good, but as anyone who has done any antenna work knows predicting the effects of terrain on the wave is difficult. While you say you have obstacles in your path you may have much more line of sight than you realize or much less. I think the very fact that yagi are direction make them a better design for high frequency applications. Omni antennas are more of what I’d think of for low frequency choices since lower frequencies actually spread out more over distance than higher ones.
https://answers.google.com/answers/threadview?id=772249
One think is for sure, don’t expect technology used for boats to tell you much about how to make an antenna or connection over land. Low frequency transmissions are ideal over water where their are no obstructions, but over land that is not the case. If only because high frequency waves are more capable of dealing with random obstructions than low frequency that still makes them the choice waves for most applications.
While your case is unique just look at the fact. The industrial world uses high frequency for wifi links. They must be doing that for a field tested reason. You could have formed you link at almost any frequency. I’m not sure, but don’t the higher frequencies also achieve higher bandwidth also ?
September 14th, 2007 at 1:41 pm
Thanks for your comment, Joe Doe!
I should’ve put down some citations of 900MHz vs. 2.4GHz and why I chose 900MHz. Unfortunately, I don’t have any scholarly articles on hand. These web sites, though, should provide sufficient information:
http://www.maxstream.net/wireless/wireless-900-mhz-2.4-ghz.php
In the next link, the author discusses how 2.4 GHz can go farther than 900MHz, but 900Mhz is less prone to obstructions than 2.4GHz.
http://www.dslreports.com/forum/remark,14095652
Since I have many obstructions between my house and my friend’s, I decided that 900MHz *may* be a better choice over 2.4GHz. It’s alright if the top speed would be 1MBs, if it all works!
This is why Canopy employs the 900MHz, 2.4GHz, and 5.8GHz bands in its product. The 900MHz is their “last mile” solution. I believe this should address your “industrial world” comment.
Canopy (Wireless) @ Wikipedia
As for the design of my antennas, yes — the antennas I built in this post were flawed in design. The antennas I built in the following post are flawed, too! These posts, though, are more of a way to document how I learned to make cheap antennas the right way. I would rather spend US$20 to see if the link will work first with DIY yagis and then invest in a few hundred dollars later to buy better antennas.
My friend and I are constructing better yagis this time and phasing two together at each point. I have better instructions on building these yagis, so I should find out once and for all if this link *can* work…