CG0901_Fieldbus
CG0901_Fieldbus
CG0901_Fieldbus
CG0901_Fieldbus
CG0901_Fieldbus

Fieldbus on a Shoestring

Jan. 12, 2009
Use the Wire You Have. Unless You’re Really Challenging the Limits of the Physical Layer, Ordinary Twisted/Shielded Pair Will Work Reliably
By John Rezabek, Contributing Editor

If federal bailout money is not in your project budget’s future, how can you minimize the installed cost of your measurement and control system? While “cans on a string” might be interoperable, shoestrings won’t work for fieldbus, and they probably cost more per foot than premium “Type A” fieldbus cable anyhow. So what to do?

Why not start with the wire? I remember a test at Emerson Process Management’s headquarters in Austin, Tex., where the BP team was trying hard to break Emerson’s implementation of fieldbus. One of the tests called for exceeding the maximum length of an H1 segment, and so our host pulled out a bail of some ugly, small-gauge non-“Type A” cable and hooked it to our test segment. It was too long by a 1,000 meters or more, but that “wrong” cable still worked. The segment functioned without errors, at least in the lab. While I was at BP, this same test was repeated at Invensys, Honeywell and Yokogawa with similar results, and I’d bet a new pair of Nikes that the same would be true at any of the leading host suppliers.

One option I’d exploit is to use the wire you have on hand. Unless you’re really challenging the limits of the physical layer (1,900 meters total length, approximately 120 meters per spur), I’ll bet ordinary twisted/shielded pair will work reliably until the economy improves. I would stage a test in the shop, and hook up lengths in excess of what you’re anticipating for the field to reassure yourself that all will be well. If you’re like me, you’ll roll the dice with what you have, and address deficiencies if and when they arise, assuming your projects and business demands will tolerate a little back-peddling.

What about terminations? Physical layer suppliers such as MTL/Relcom, Pepperl+Fuchs, Turck, MooreHawke, Phoenix Contact and others have outstanding products with features like short-circuit protection for every fieldbus spur. This is a good feature that I would routinely exploit, but it isn’t free. These weren’t available commercially when our first fieldbus segments were installed, and we used essentially simple terminal strips. Foundation fieldbus and Profibus PA will function happily on ordinary terminals—even wire nuts or butt-splices. I’ve spent years lashing the backs of our maintenance folks about how much I abhor the thought of butt-splices in form-7 fittings, but in times of hardship, I’ve been known to compromise. If you have excellent maintenance discipline, I contend that you could consider a system of ordinary, unprotected terminations. You can always replace these down the road when money is flowing easier. On segments where the consequences of a failure are minimal, this could be your default choice. But users who are striving to comply with Intrinsic Safety requirements may find that the reduction in devices-per-segment is not worth the savings afforded by “unprotected” simple terminals.

Back in the host system, what about “letting go” of redundant H1 cards? When my DCS went in, redundant H1 wasn’t supported by the system. We rationalized that an H1 card failure was a “soft” failure—loss of view only. We also parsed our segments, so no two “critical” segments were terminated on the same card. I’d argue that H1 card failures are rare, and when they do occur, nothing “fatal” happens. Loops solved in the field keep on keeping on as long as a backup link-active scheduler (LAS) is configured, and the default action of control valves—backup LAS or not—is to “hold last position.”

Also, look at field devices. Maybe your favorite supplier’s top-of-the-line transmitter is overkill for a significant fraction of your applications. Many of our favorites have second-tier offerings at more competitive prices. Also, see if you are paying for FF function blocks you don’t need. Every transmitter I buy comes with PID blocks in it, but I can count on one hand how many actually are implemented.

Our suppliers would rather sell us something than nothing, so let’s work with them to find solutions that meet our requirements without exceeding them.