CD1505FiberOpticsCheck-list

Do You Know Your Fiberoptics Checklist?

May 2, 2014
For Any Network, the Designer Needs to Ponder a Few Fundamental Questions: How Far? How fast? How Critical Are the Communications?

Where I live there's no cable, as in coaxial cable for cable television or broadband Internet access. So the local phone company blasts Internet connectivity to my home using some pretty dicey, direct-burial, copper UTP (unshielded twisted pair) phone cables. You've heard of DSL, or digital subscriber line, through which the phone company can deliver bandwidth well into the 10 megabaud range over a few miles of ordinary UTP analog phone lines. Newer technologies are being deployed to use similarly questionable copper conductors—your home power wiring, for example—to carry high-bandwidth, digital signals such as Ethernet.

For any network, the designer needs to ponder a few fundamental questions: How far? How fast? How many concurrent users/messages/signals? How critical are the communications?

Consider the latter. How critical is it that messages get delivered in a timely manner? In industrial networks, many architectures lack distributed control and rely on a centralized controller or logic solver to ensure the factory remains productive or that the wind farm stays in sync. Ordinary feedback control relies on sampling rates that are six to 10 times faster than the dominant (analog) time constant, but also are deterministic; that is, the sampled data needs to be transmitted at a constant frequency (sampling interval) with minimal latency. While even 10-MB Ethernet is likely to be many times faster than needed for most feedback control (PID, for example), we need to consider the worst case when perhaps an alarm flood, network malfunction or maintenance activity inadvertently might consume much of the bandwidth.

The other shoe of criticality is the reliability of the physical layer. Redundant networks have become commonplace, and some manner of geographic diversity (redundant media following separate paths) is a good idea. But many—perhaps the majority—of devices have copper/Cat 5 network interface cards. Therefore, we need fiber-to-copper media converters, and these also need to be redundant. We might be compelled to use UPS power for the same.

Even though it's slightly more complicated, fiber easily wins the bandwidth battle, and that bandwidth itself could be considered more-reliable. Copper media is arguably more vulnerable to stretching and kinking. Terminations such as RJ45 jacks can be a challenge to install reliably. Fiber is totally non-incendive and is unaffected by radio or electromagnetic interference and crosstalk. If copper switches degrade to 10 MB due to cable issues, we might not detect it. And it's also well-known that copper Cat5 UTP isn't rated to operate at distances of more than 100 meters, even with a clean installation.

How far you can go with fiber is tied to "how fast." Our brethren in the telecommunications industry run almost exclusively single-mode fiber—a more expensive, finer and more precisely manufactured media that can go 200 km and maintain data rates of many gigabits per second without repeaters or amplification. For most of the distances we encounter in industrial networks, multimode fiber is more than adequate.

How far and fast you can go with multimode depends on your media converters (whether external or part of a switch). Some media converters use LEDs to transmit the signal, but converters that use semiconductor lasers not only can transmit farther, but also have higher bandwidth capability. Since these optical transmitters need to traverse the same distance each way, you need the same species for both ends. I've mixed and matched some different vendors' modems for experiments and low priority/low bandwidth networks, but it's also advisable to use uniform network hardware (same species/same vendor) for the most critical parts of the network.

You can find Cat 5 cable and connectors at the local hardware or builder's supply, but you won't find fiberoptic cable or termination kits there. There's a reason for that: Reliable and optimum bandwidth fiber terminations require some specialized skills and an investment in tools and testers. There are some new products that have some promise to make fiber terminations more accessible to novices, but we've used these with mixed success. For the relatively low cost of bringing in someone who terminates fiber for a living—the phone man, for example—it's a good investment.

For comprehensive guidance and checklists for fiber usage and installations, users can consult the Fiber Optics Association online.