If you knew you could put 32 devices on a single two-wire bus such as Profibus PA or Foundation fieldbus, how much of that capability would you use? In their zeal to optimize the wiring savings of fieldbus, one end user had segments with more than 20 devices, but they had to make some serious tradeoffs in communications performance.
The cycle time or "macrocycle" of the segment had to be lengthened, and the scarce free time for lower-priority messages meant these methods could take a long time.
Both the Foundation fieldbus and Profibus PA specifications accommodate up to 32 devices per segment, although many hosts lower the limit to 16. If we were selecting conventional I/O for a new system, we might take our I/O count and add 20% or 25%, specifying enough cards for that number. So if a given host allows a maximum of 16 devices per fieldbus segment—per channel, if you will—is it wise to load them each with 12 devices, thereby providing the conventional 25% spare capacity? If you ask Michael Clark, president and chief consultant for Calgary, Alberta-based BusCorp, he'll tell you, "Just because you can, doesn't mean you should."
In fact, the latest release of the Fieldbus Foundation's "Systems Engineering Guide, AG-181" recommends users should aim for an average loading of 12 devices per segment. But Clark's methodology rolls back that limit another 25%. He recommends your initial install should have a maximum of nine devices per segment, and that you need never exceed 12 devices per segment. "Fence me in!" he says, because this conservative choice will result in an even more foolproof project.
Clark also recommends that we limit ourselves to a simple topology. The fieldbus physical layer has as many topology permutations as there are ways to connect 16 pairs to one another: You can have daisy chains, bus with drops, forked spurs and chicken's feet, and arrange them in any variety of combinations. The problem with creative topologies is that they create complexity and potential "gotchas" that manifest in the design phase, procurement, construction, commissioning and maintenance. Clark says,"Utilize the tree (a.k.a. chicken foot or star) topology for all segments."
One immediate result of planning for 12 lifetime maximum devices per segment and all "tree" topology is that your fieldbus couplers can all be identical 12-spur couplers. This simplifies design—all segments can share the same drawing template. Procurement is easier—the project need only specify and purchase one variety of coupler and in greater numbers. Construction benefits, as every junction box will have the same hardware and lend itself to mass production. Your electricians construct raceways of rigid conduit or tray that closely parallel traditional multipoint jobs. And maintenance will be happy it only needs to stock one flavor of spare part for all the segments.
Another fence that Clark erects is around spur and trunk length. The physical layer specs for both Profibus PA and Foundation fieldbus recommend that total network length—the sum of all spur lengths and trunk (home run) of any segment—should not exceed 1,900 meters (provided one is installing fieldbus Type A cable). In his recommendations, Clark says all spurs shall be 80 meters or less, and no trunk shall exceed 900 meters. So when he calls in Professor Einstein to do the arithmetic, he gets 12 x 80 + 900 = 1,860 meters, which is safely below the maximum.
What about segment calculations? With careful selection of redundant segment power supplies/conditioners and a little oversight of device current requirements, no calculations may be necessary.
"Freedom to Choose" has been a motto of the Fieldbus Foundation for a couple of decades. Imposing some conservative constraints may seem like you're fencing in that free spirit, but it's really a choice that can help free your project from complexity and stress.