Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


Multimode Fiber Provides Option for Microwave Data

A solution to Cat-5 limitations

A 90 micron fiber with some of the buffer stripped off. A 1/4-watt resistor is shown for scale. Tiny stuff, and it will draw blood! Since 2003 or so, I have been experimenting with alternative STL systems, and have written about some of my experiences in these pages. On a recent project, however, I took the opportunity to go a different direction with part of the link.

These “alternative STLs” have, for the most part, taken the form of some kind of 802.11 unlicensed link, a Part 101 licensed microwave system or both. All provide some quantity of Ethernet bandwidth between one end of the link and the other, and we have employed various codecs to transport audio from end to end in either or both directions.

Because they carry data, each of the links is fed with a piece of Cat-5 or Cat-6 cable. That is good in one sense because Cat-5 and Cat-6 cable is a lot cheaper than coaxial cable or waveguide, but it’s bad in a couple of others because there is a 100-meter length limit on such cables, and it’s tough to get them across the base insulator of a series-fed AM tower in AM station applications.

Those limitations led me to a relatively inexpensive solution: employing a short-hop 802.11 link to get from the top of the tower to the transmitter building with the data. There are many options commercially available, but my favorite is the Ubiquiti NanoBridge M5 link. It’s inexpensive (less than $200 for a complete link), reliable and has a lot more bandwidth than any long-haul microwave link you’re likely to use. This configuration requires power on the tower very near the microwave and 802.11 link antennas, which is probably not a big issue or expense, and it also requires a small Ethernet switch at that location.


Over the decade or so that I have been using this kind of link, those Ethernet switches have been the number one point of failure. They don’t fail often, but let’s face it — in this application, they’re operating in a tough environment. These simple consumer-style switches are mounted outdoors, probably in a poorly ventilated NEMA enclosure, with ambient temperatures ranging from 10 below zero to 140 above … and don’t forget about the occasional magnetic field spike from a lightning strike on the tower! It’s amazing they work at all!

Contents of the Corning fiber-optic toolkit. In my new project, I was looking for a way to do away with the on-tower Ethernet switch and the 802.11 link, thus reducing possible points of failure in the system by three or so active devices. Because I was working with a series-fed AM tower, part of a 50 kW array, I had to deal with the issue of crossing the base insulator without adversely affecting the impedance of the tower. My options were few, but one kept jumping out at me: multimode fiber.

Except for the short piece of fiber in the old Continental “Power Rock” series of transmitters, my experience with fiber-optic cables was limited to gimmick Christmas decorations. I knew it was possible to pass a lot of data over a long piece of fiber, but I knew nothing of how to convert Ethernet to optical, put on connectors, make splices or any of that. It was time to do some investigating.

So I started talking to people, surfing the Web and learning all I could about it. Multimode fiber is certainly nothing new to the folks who work with data. I found some of my amateur radio friends who regularly use it as a transmission medium within large data infrastructures. As such, there is a lot of information on fiber-optics out there. That’s the good part. The bad is that none of the information fit my specific application.

Rear of Dragonwave 11 GHz transceiver showing power connection (top) and fiber connection (bottom). That’s where we engineers get to do our thing — figure out how to apply different technologies to meet the needs of particular situations.


The microwave radios I planned to use were made by Dragonwave, and a brand-new option at the time was for multimode fiber inputs and outputs. It was so new, in fact, that evidently none had yet been shipped.

Dragonwave provided me with the information I needed — the type of fiber, connectors, etc. required; and with that in hand, I began searching for equipment that I could use on the other end of the piece of fiber to get the blinky red light back to Ethernet ones and zeros.

That search turned out to be short. Cisco and other manufacturers of high-end gigabit switches offer units with GBIC (gigabit interface converter) ports, and for under $40, there are any number of GBIC fiber-optic transceivers available. Buy one, plug it into the switch’s GBIC port and you’ve got yourself a gigabit switch with a fiber port on it.

The next part of my education came in learning how to put connectors on multimode fiber. That was quite a learning curve. On the recommendation of a friend who works with fiber, I purchased a Corning fiber-optic connector installation toolkit, something that set me back some $1,400. It included all kinds of tools used to work with fiber, the connector jig, crimp tool and continuity tester with several different connector adaptors.

I had to learn first to be very careful, very gentle. Fiber is incredibly robust, but nick it and it’s extremely brittle. Removing the “buffer” (what you and I would think of as “insulation”) from a 90-micron fiber is a delicate operation that can easily end with a broken fiber and ruined installation.

Interior of the NEMA enclosure on the tower containing the surge suppressor, –48V power supply and loop-through for the fiber. The –48V power supply employs a coaxial-type connector. The coaxial m-f connection was secured with electrical tape by the tower rigger to keep it from vibrating apart. I also had to learn that everything had to be very clean. You clean the bare fiber with a special cloth and cleaning solution before putting on the connector. And you don’t cut fiber, you “cleave” it, essentially scoring it at the desired cut point and then breaking it. Weird.

Once I got the hang of it, though, putting on the connectors wasn’t so hard. My third or fourth effort worked (the connectors I wrecked were $12 each!). For a test, I used a pair of Cisco switches with GBIC fiber transceivers to connect my laptop to the office network. I was amazed when I was able to connect to the Internet and pass data over that piece of fiber! But that experiment proved to me that it would work, and it showed me that I could successfully put the connectors on.


I had the Dragonwave transceivers shipped to my office in Colorado and my next step was to configure them and connect them with fiber to the LAN.

I got that done and tested, and then programmed in the frequencies to the Dragonwaves. Next, I ran the power all the way down and actually pushed data through the microwave link and fiber, just to make sure it would work (it did). Better to confirm all that on the ground than try and troubleshoot it 300 feet in the air!

So I shipped the Dragonwave transceivers, power supplies and a 1,000-foot roll of duplex multimode 90 micron fiber-optic cable to California for the installation. The tower crew installed the antenna, a four-foot dish, on the tower along with a NEMA enclosure for the –48V power supply for the radio. The enclosure provided a pass-through for the fiber-optic cable to the radio.

Fiber-optic cable splice employing LC-connectors and a dual coupler inside the ATU. The duplex cable, with buffer and outer jackets, resembles zip cord. My initial plan was to use a solid run of duplex multimode fiber from the transmitter building all the way to the top of the tower, and indeed, that would be the ideal way to install it. But after much thought and discussions with the tower crew, we decided it would be too difficult to pull a cable with delicate connectors installed in a piece of 3/4-inch rigid conduit up a 300-foot tower. If a connector were damaged, it would all have to come back out — there’s no way I could install a connector hanging from a harness at the top of a tower!

The plan, then, changed a bit. I opted to terminate the fiber-optic cable from the transmitter building (some 500 feet away) inside the ATU where it would be spliced to the cable coming down the tower. The conduit from the transmitter building terminated into the bottom of the ATU cabinet, so that part was easy. For the run from the tower, I set a 6x6x4 PVC J-box right on the transom plate of the tower, terminated the vertical conduit run from the top into that box, then ran a piece of Seal-Tight to the lighting control box on the tower to pick up 120 VAC.

To cross the base insulator with the fiber, I used a length of 100 percent PVC non-metallic Seal-Tight, connecting one end into the J-box on the transom plate and the other end into the underside of the ATU cabinet. I had some reservations about this (there would be 15 kW or so across that base insulator and thus across the piece of Seal-Tight), and I expressed them to Bobby Cox of Kintronic Laboratories. Bobby replied that KTL has been using PVC to manufacture isocouplers for years with no issue. That made sense, so I stopped worrying. So far so good with all the RF!

PVC J-Box on transom plate. The conduit coming in on the top left is from the NEMA box on top of the tower and contains both 120 VAC and fiber-optic cables. The top right conduit carries the 120 VAC power conductors to the lighting control box on the tower. The PVC conduit on the bottom contains the fiber cable and crosses the base insulator to the ATU. With the splice made inside the ATU, I applied power to the Dragonwave transceiver on the tower, and just like that, the GBIC port lit up with a valid gigabit connection and we were immediately able to push management data to the transceiver. In short order we had the path aligned with the predicted signal strength at each end, and we were pushing data between studio and transmitter in both directions.


The link now carries STL audio, remote control, security, video surveillance and telephone to and from the site. I sure feel a lot better without an Ethernet switch on the tower and an unlicensed RF link between tower and transmitter building. The removal of those additional possible points of failure changes everything.

Going forward, I plan to make fiber a primary design element in my AM and FM STL links. It solves both the base insulator and length limit problems with Cat-5/6. And besides that, I think it’s pretty cool.

Cris Alexander is director of engineering at Crawford Broadcasting, a longtime RW contributor and a past recipient of SBE’s Broadcast Engineer of the Year award.