Customer Feedback
Installation Report for an AW-900 Ethernet Bridge System

Richard Burgess IP Data Corporation Edgewater, FL Ret. US Coast Guard Lt. Licensed FCC RF Expert

After searching the Internet for any information on real-world installations of the new unlicensed 900 MHz NLOS (Non Line Of Sight) 802.x Ethernet devices (bridges, point to point, point to multipoint, etc.), I discovered there was very little information to be had. The technology is fairly new and I found more questions than answers. For this reason, I want to share my Avalan AW-900 installation details with you because you may find them handy someday if you want to bridge two offices at “near T1” speeds.

If you want to know about my education and experience with RF, data links, computers and electronics in general, there is a short biography at the end of this text. It may help you to understand some of the decisions I made that may not be obvious to you.

Avalan’s AW-900 seemed to be just what I was looking for to save money, reduce commercial bandwidth consumption, and provide good security over a link between two offices. The offices consist of our new primary office (a leased space that we just built out) and a secondary site where we copy important data to get it out of the primary offices, and also where we complete some projects. The secondary site was our old (smaller) office – we own the building. The link also gives us a way to access the Internet from the secondary site since we moved our T1s from there to the new office. We used to have our T1’s at the secondary site and we were going to leave one when we moved, but it didn’t look too good on the budget (that was too polite – it looked HORRIBLE on the budget at $7200/year).

Part of this has to do with the fact that a T1 is the only type of broadband link available in that location. It’s really “in the sticks” as far as the Internet is concerned. There’s no DSL available because it is too far from the phone company central office (CO), and no cable TV, either. These days, that is considered the Internet sticks.

Besides the cost (which isn’t an incidental at that price) it hurt us in another way leaving a T1 there. When we copied large amounts of data across to the secondary site, it ate bandwidth on one of our new T1s in the primary office to get over to the other T1. We sell data to clients all over the world at all times of day and night, and eating up client bandwidth is not a good idea (if we want to stay in business, that is). There is heavy competition in our industry and when a client is downloading a zipped up 35 megabyte file, 3 to 4 minutes looks much better than 6, 8, or even 10 minutes (depending on loading).

This is where the wireless 1Mbps link comes in. One Mbps is the advertised throughput, while the raw speed of the link prior to link level overhead (packet control and encryption) is 1.5Mbps (megabits per second). Divide by 8 and you roughly get the megabytes per second speed (that is about 120 kilobytes per second throughput).

HOW MANY TREES ARE “TOO MANY” TREES?

I had read about these devices, but no one really indicated just how many trees are too many trees before your 900 MHz signal is dead (or fairly useless, anyway).

I initially investigated 2.4 GHz systems, but everything I read and my own experiments with some indoor wireless units proved that it only takes a couple of dense trees and the 2.4GHz signal is as dead as a doornail.

The 900 MHz frequencies were reportedly much better to get through the trees, but I had to get through a solid mile of very dense woods between the two sites. I asked a lot of questions of unbiased people, and I got basically the same thing from everyone, which was: “No two sites are alike, you’ll have to experiment to find out what works.” A friend of mine up in New Jersey said “I am quite sure that you will have to get over the densest parts of the trees, but the tree tops won’t bother you at all, and the farther the trees are away the less affect they have.” He has been in RF as long as I have, and it agreed with my past experience, so I was pretty sure I knew what was needed.

Experimenting to determine if it could be done was really not an option. I had already decided that I was going to put in a 900 MHz link and it was going to work, no matter what. Even if I had to elevate the antennas on towers above the trees, I was going to have my own link and save about $7200/year plus leaving the bandwidth for my customers on my primary T1s.

When I Googled and found the new Avalan AW-900 Ethernet Bridge series I also tried to find some real world experiences about it in the News Groups or message boards dedicated to wireless experiences. I did find some information and a short movie on the Avalan site, but it showed 10 or 15 trees in the path of a pole-mounted camera going to a building, and compared to what I was going to try, their movie example was like being in the same building! No one seemed to have any real information about installation and operational experiences with a mile or more of dense woods using these new 900 MHz technologies. So here I am trying to fill that gap.

First, lets go over the area I had to cross and what it looks like on a map and in satellite photos. I will also describe the trees in detail that are shown in the satellite photos between the two endpoints. The distance between the two points is 1.85 statute miles (almost exactly 3 kilometers). The black line shows the path between the two buildings (it is a white line on the Satellite photo).

THE MAP

.

THE SATELLITE PHOTO

Note that the entire left side of the path is wooded and totally undeveloped.

PROBLEMS IN THE PATH

As you can see from the photo, almost one full mile is very heavily wooded (trees and dense underbrush). Oaks, palms, maples, and just about every other variety of tree you’ll find native to central Florida is in this wooded area nearest to the Secondary Site, and they are all mature trees. The area is not developed at all. It is original Florida “ Wet Lands” and woods. The very tallest trees top out at about 70 to 75 feet and the tallest trees appear in every direction at about 100 to 200 feet intervals. Another 25% to 30% of the trees along the path top out at about 45 feet with the rest at about 20 to 30 feet. The underbrush is also quite dense with short palms, ferns, saw grass, and several varieties of “I’ve never seen that before in my life” plants, and a good many of them thick are broadleaf and wide fronds - all excellent signal eaters, I’m pretty sure.

Sometimes you can use a directional antenna and go “under the tree canopy” for a half mile or more if there is little or no underbrush or lower branches, but that certainly was not the case here. You can’t see 10 yards in any direction in that swamp (and it IS a swamp – true Floridawet lands).

Adding to the “green” problem are power lines at about 50 feet and another set at about 35 feet that run along the entire length of Park Avenue. They usually don’t attenuate or interfere with the signals too much, but they can cause some serious interference if a 4,000 or 15,000 volt line has a bad (cracked) insulator near your location. Arcing in a cracked insulator can cause horrible wideband RF noise as high as 2 or 3 GHz. Listening to an AM radio can usually tell you if there is serious, undesirable noise being generated by the lines. A police scanner in the 900 MHz bands can also help you out listening for noise if you suspect there is an interference problem. Any of the Radio Shack police scanners made in the last 10 years can tune the band from 896MHz to 960MHz so you can even listen to see if the ISM band has traffic (an old Pro-34 model scanner can be had on EBay for 50 bucks or less). Just set it to scan and listen to everything from 900 to 928 and see what’s there. I had a clean shot except for something near the upper end. But the Avalan units automatically change and select channels to avoid interference for maximum signal strength and data throughput, so I didn’t worry about it. A very nice touch, indeed.

INDOOR & OUTDOOR AW-900 UNITS

There are indoor and outdoor (sealed) versions of the AW-900, and Avalan also has high gain Yagi antennas available for the 900 to 928 MHz ISM band (Industrial, Scientific and Medical) band which units use. There are several types of antennas available for this band and more being developed even as I type. Avalan has an entire “arsenal” of antennas for the 900 MHz band, some indoor, and many outdoor. Two of them are high gain Yagis, 11db and 15db. They also have an 8db omni which is intrigued me, but not enough to spend the money for a third antenna to play with, although I am now betting it would have worked fine in my situation (you’ll see why in a few minutes).

Yagi antennas are the lightest weight for the gain, easiest to mount and easier to work with than parabolic antennas, and their only negative aspect is that they may be a little more susceptible to interference from the sides and back quarters, when compared to a parabolic mesh or grid antennas with the same gain. If you have a choice, and you need the gain, go with the Yagis. They are inexpensive and easier to aim than parabolics and they weigh a whole lot less (3 lbs for the Yagi and 10 lbs for the parabolic).

When I purchased the Avalan units, I opted for the indoor version of the AW-900 that come with a 2.5db gain “rubber ducky” antenna on each unit to test out my ideas and to use in the final installation once I determined if I would need more than just the “basic system.” The outdoor sealed version is about $300 more than the indoor version ($995 vs. $695 – MSRP as of this writing) but I wanted to keep the electronics indoors. There was a time when I considered climbing towers and working on rooftops fun and good exercise. But not any more. I hit the big “FIVE O” and sanity is starting to finally take hold of me. I was also a little leery of having to aim them with the signal indicators at the top of a pole if I ended up with one of them up high and out of reach. And this was a “lucky guess” because one ended fairly high up off the ground.

You have to keep in mind that the amount of power the units produce is fairly small and you depend on the antenna for additional gain to go longer distances. But also be aware that the antennas increase the receive signal strength as well as the effective radiated transmitter power – a dual purpose mission.

Having decided on the indoor units, and also knowing that I may have to elevate one or more of them, I began to hunt down extremely low loss cable with the right connectors. Back in the old “pre Internet” days, you built your own cables 99% of the time – not anymore!. Now I just get online, click some buttons and someone builds them for me using the nifty new compression type connectors that an elephant can’t pull apart, and in 3 or 4 days you have your cables (and, at a decent price!). I didn’t even have to talk to a human, not that I have anything against humans, mind you, but it was so easy to get exactly what I wanted. They even have close-up pictures of the connectors so it is difficult to make a mistake – and this is a good thing because the industry has become confused and they have made it difficult if not impossible to tell a male from a female connector! Is this science imitating life instead of art, maybe? Telling a male from a female is something you learn at a VERY young age (at least where I came from). The source of confusion in this area seems to stem from the fact that some people look at the threads and assume that if the threads go outside, it is a female connector. Looking at it this way, your only right half the time. With 32 years in the field of RF electronics, I can assure you that the deciding factor of whether a connector is male or female is decided by the center pin and not the threads or hood of the connector. Take a Type –N for instance, this is one where you have the female outer hood that goes over a smaller threaded part, but inside the smaller threaded part is the pin, and this pin is inserted into a center pin with a hole in it. The center pin with the hole that receives the pointed solid pin is the female even through it is smaller and goes inside the outer hood that turns to tighten it. HOWEVER, somehow, the wireless industry has decided to invent something called a Reverse Polarity Connector. Thanks for nothing! They are not any better impedance-wise, and they serve no additional purpose except to make it more difficult to buy the right connector. I’m guessing that this was a company trying to make sure that you buy THEIR antennas, or at least the right antennas. So, Caveat Emptor (buyer beware).

The important thing is not that you know MY definitions (and what I have known for 35 years), but instead to make sure you know to get the correct connector genders. You must make sure you use the pictures to order your connectors. If you don’t, you will get the wrong gender about 30% of the time. Here are some pictures to help you out:

Now look at the original TNC connectors (before someone who possibly took a lot of drugs in high school decided to “reverse” them):

Yes, the center pin on the “Female” RP-TNC is solid and it fits into the hollow center pin on the “Male” RP-TNC. In my humble opinion, this is where they got it wrong, and this is obviously what the “Reverse” in Reverse Polarity means. Compare this to the original TNC (above) and you will see what I mean.

What it boils down to is this: Iif you intend to use cables between the antenna and the unit, make sure you get the right connectors. Make sure you SEE the pictures and that you know what comes with the antenna. In fact, and antenna I ordered said it came with an N-Type Male. It showed up with an N-Type Female (which is what I was going to install, anyway). This was from someone on EBay, and I hope that the manufacturer would have gotten it right. Here is where I ordered the cables and yes, they have pictures of the connectors: http://www.wirelessnetworkproducts.com/index.asp - Just click on the cables menu. They had the cable lengths I needed with the correct connectors already on their list of pre-made cables! They have hundreds of different type listed and they have decent prices, too.

I wasn’t sure if I would need high gain antennas, but I already guessed that I would have to get up above some of the trees and my first round of experimenting proved me correct (or so I thought).

I had one 900 MHz ISM band 15db gain parabolic antenna that I bought from a friend that was working with another 900 MHz ISM application. The sealed AW-900 units are designed to mount outside an office window or at the top of a pole (a traffic light pole with a camera, for instance), but I wanted to see the signal indicators up close and personal, and not while hanging out of a bucket truck in a Florida afternoon thunderstorm. This is just my preference, mind you, but I also wanted to consider future maintenance. Granted, antennas and cables require maintenance, but if they are initially installed and sealed properly they can go for years with no need to even think about them.

I also had to think about cable loss which is I why I got the LMR-400 low loss cable – it is made by Time Microwave and not very flexible, but the specs are excellent. The loss is less than 3.5db per 100 feet at 1000 MHz. For easy calculations you can figure that you lose half of your power for every 3db of loss. Likewise, you double it with every 3db of gain. This is an approximation, but it is fairly close.

The active elements on the antennas (the parabolic dishes) are at DC ground. Even with the elements at ground level, I recommend you purchase and install lightning arrestors for the cable. They plug inline right at the pigtail of the antenna and are clamped to the tower. In all of my years, whenever a tower has taken a direct hit, the gear connected to the antenna at the top was almost always fried. Most of the time it was a duplexer and/or a receiver, but the arrestor will help some. And make sure you have adequate grounding if you put up a pole or a tower that extends above surrounding structures.

THE FIRST TESTS

I stuck one of the 2.5db gain flexible antennas of one unit out of my office at about 17 feet (a hole in the second floor wall), and I got an inverter for my car and plugged the other unit in and started driving west on Park Avenue (see the map) to see how far I would get at ground level with the supplied antennas. I got just short of 1 mile (right past the airport headed west on Park) and the signal dropped to zero from the 5 or 6 lit LEDs it had been on since I got in the car at the office. There are 6 signal level LEDs on the device to indicate a combination of relative signal strength and data loss. I got out of the car and tried to move the antenna around, but no dice. I was at the maximum in this direction using the little flexible antennas. A mile is actually quite good at almost ground level! There was a large steel building between me and the building at this point, so I wasn’t going any further in this direction (which is approximately the direction I needed to go).

THE NEXT TEST

A friend of mine had a 15db gain parabolic antenna from a failed experiment he tried and I was lucky enough to get it for a song ($50). He tried to extend a cordless phone over 1 mile, and I could have told him it wouldn’t work (for reasons I won’t get into here), but he was sure he could do it – now, I had one very nice antenna!

I hooked up the 15db gain parabolic dish antenna and put it in the back of my Honda and headed west from the office again. Periodically I stopped and took out the antenna and aimed it at the office. This time I had good signal (5 to 6 lights) at the point where I last lost it. Also, aiming the antenna was very critical and not easy holding it over my head (not to mention some mighty strange looks from people passing by). it has a 22 degree beam width at the advertised maximum gain, but it really didn’t seem to be 22 full degrees. I then turned north on Airport Road (it crosses Park Ave going north and south just west of the airport – see the map). I drive up to the next corner (not more than 250 yards) and tried to aim the antenna to get a signal. There was a thick strip of trees and underbrush now between me and the office. I got absolutely nothing. Even with this 15 db gain antenna, I found no signal at all. That was depressing. I only had one of these antennas so I ordered another one. The antennas are made by Pacific Wireless, are well made, and well worth the $75 price I paid on Ebay for it.

I would probably have been just as well off with the 15 db gain Yagis from Avalan, but I already had one of these same parabolic dish antennas and I wanted the same type of antennas at both ends (chalk this desire for symmetry up to experience). The Yagis are lighter and easier to work with and have the same gain, so if you need high gain antennas, I recommend that you go with the Yagi (unless you have a friend that has one or two antennas he is not using at the moment that are good for the right band).

TOWER TIME

I wanted a small tower for other reasons at our original building (the old office site), so I was intent on installing one. I wasn’t going to fool around and take any chances on failure – I WANTED THIS LINK WORKING! So, up went five 10 foot sections of Rohn model 25 triangular tower next to the building. We guyed it at the 35 foot point. The first dish went up on the tower mounted on a 10 foot 1.25” pole insert. The top section of the tower is designed for an antenna rotor which also allows you to insert and clamp a single 5 to 10 foot pole through a hole at the top. Below is a picture of the tower with the dish at the top. The picture is deceiving because most of the trees are higher than the antenna (including the tree directly to the left of it in the picture – the antenna is much closer to me).

Antenna on tower at secondary site (about 55 feet up)

The second dish we installed on three sections of galvanized TV mast (courtesy of the local Radio Shack). Using the maps and a GPS unit, we aimed the disk antennas, and even with 15db gain antennas, one side at 55 feet and the other side at about 26 feet... we had nothing. Nada, zip, zero, zilch, flat...

Antenna at primary offices (about 26 feet up)

NOT THE PROBLEM I THOUGHT IT WAS

It was depressing! I immediately thought: “The tops of those trees are eating all of my signal!” However, as I was standing there holding the Avalan unit at the base of the 50 foot tower, I saw the signal lights flicker. I moved the box again and they flickered again. So I grabbed the antenna cable in one hand and the box in the other and bent them slightly in opposite directions, and sure enough – first I had 4 signal lights, then a few seconds later I had all 6 lit up rock solid. Finding this problem was pure luck. And yes, it was almost “Miller time,” too!

It turns out that with all my movement of the cable and the box, I had loosened a solder connection inside the unit where the center pin of the RF connector is soldered to what I assume is the transceiver circuit board inside the unit. I don’t think that the Avalan engineers originally planned for someone to hook up thick, solid LMR-400 cable to the connector on the end of it, then drive it around in a Honda for a few hours. They aren’t really that delicate, and the end of the box is metal, but the end is seated in a groove in the two plastic halves of the small box, and even with the box sealed, the metal with the connector can move ever so slightly (almost imperceptible movement) – but it was enough to break the solder connection where the RF connector is attached to the printed circuit board.

I took the unit back to my work bench and heated up my trusty 30 watt soldering iron and had it fixed in less than 5 minutes (including disassembly and reassembly). I don’t recommend you try this unless you want to invalidate your warranty, and besides, after talking to Avalan, they are already taking steps to prevent this from happening in the next batch they produce. That was very good to hear. I may need one or two more of these things in the not too distant future, and good support is important, and rare these days!

THE FINAL TEST

So, just as a test we lowered the antenna at our new building and got it almost to ground level before the signal decreased. The tower on the heavily wooded side is probably 10 to 15 feet higher than it needs to be, but I’m sure I had to get it over most of the trees to get maximum signal strength all the time, and we do have 6 solid lights, 24 hours a day. I had originally planned on 20 to 30 foot poles on both sides, but I’m glad I put the tower up on the heavily wooded side. I know that if it had worked with two 20 foot poles, I would have left it, even if I had less than maximum signal strength.

Below are pictures of the main Avalan unit in the equipment closet in the back of our computer room at the main office. You can see the rounded curve of the LMR-400 cable going into the bottom side of the unit. The LMR-400 is not very flexible and in fact, you do not want to kink this cable at all. Every minor kink and bend that is tight is just like a speed bump to the signal.

As a final test, I measured the bandwidth by copying a 10 megabyte file across the link and it took 84 seconds. That is excellent! That is one megabyte approximately every 8 seconds, or right at 1 megabit/second (maybe just a wee bit less). That is almost exactly as advertised. If you test speed on a link like this, don’t copy a file using the graphical user interface copy function. If you are using Windows XP or 2000, use named pipes on a command (e.g., Copy \\RemoteComputer\C\datafile.dat). If you are on a Unix-like system, you’ll use the command line, anyway (won’t you?). This is because the GUI copy function adds a second or two for no apparent reason and skews your results (I have no idea why. Ask Mr. G. - He might know.).

BOTTOM LINE

I now have a rock solid one megabit link between two offices almost 2 miles apart (including a mile of dense woods) and it has been running for almost 8 weeks 24 hours a day. I had to reset one of the units after a very bad thunderstorm one afternoon when we took a hit close to the tower, but other than that it has run continuously and flawlessly, and I am absolutely thrilled with it (not too mention the fact that I am saving $7200/year with this link). The units work as advertised, and you probably won’t have to go near as high as I did with the antennas, either!

ABOUT THE AUTHOR

I am a retired US Coast Guard LT with 10 years enlisted experience as an Electronics Technician (4 years as Chief). I’ve been working with RF (radio frequency) communications equipment and various RF applications, digital data transmission, and telecommunications in general for 32 years. I’ve been assigned as the senior technical officer at a USCG LORAN transmitting station in charge of the operation and maintenance of a 1.25 Megawatt (peak) 100 KHz transmitter that emitted a tightly controlled digitally formed pulse (along with additional digital data) that covered a good portion of the northwestern Pacific ocean. This was on the island of Iwo Jima, Japan while I was an Electronics Warrant Officer in the US Coast Guard. I was originally trained in electronics maintenance and engineering when widespread use of digital ICs (Integrated Circuits) was in its infancy (the pre Intel 8080 days) – Yes, I’m dating myself pretty accurately. I’m also a General Class Amateur Radio operator, and I held a 1 st Class FCC License (“a commercial broadcast ticket”) some years ago when they were actually worth something and you sweated over a 5 hour test with a slide rule and the “then brand new” 8 digit 4 function calculators were taboo (they didn’t have square root functions, anyway). I now hold a General Class Commercial license with a ship’s Radar Endorsement. I’ve set up, aligned and maintained everything from 10.2 KHz Omega equipment up through 10GHz Radar including radio and packet data links on almost every amateur and military frequency from the 2.0 MHz Marine bands up to 2 GHz Microwave links. RF was always my favorite engineering toy to play with.

Back in the mid 1990s I wrote a book that was published by MacMillan (SAMS) named “Developing Your Own 32 Bit Computer Operating System” about a multitasking operating system I developed with help from a friend. You can now get the code for the operating system in the public domain.