In all the years I’ve used Ethernet switches in various control systems, I don’t recall a single failure. Ethernet switches have been replaced through upgrades and obsolescence, and as security concerns spurred more industrial options. However, there’s never been an instance where a network switch had a detectable hiccup, let alone a complete loss of function. For our relatively small control networks, it might be surmised that we’re just not stressing them very much.
Despite their reliability, our process control network employs Ethernet switch redundancy because of the many-eggs-in-one-basket consequences—loss of view for the operator, loss of inter-controller communications and loss of the alarm subsystem. It also connects to any centralized database, trends and process history. A process shutdown would likely be invoked rather than risk a calamity. The non-redundant network switch becomes a single point of failure for several critical systems, though field devices remain powered, and many or most of the control loops in “auto” continue to function, albeit at their last setpoint.
Get your subscription to Control's tri-weekly newsletter.
Fault tolerance, robustness and single points of failure will weigh on our choices when designing a Foundation Fieldbus (FF)-to-advanced physical layer (APL) migration. With FF and Profibus PA, field networks were considered robust because the two-wire, power-plus-communications physical layer could be free of active devices—microprocessors executing code—from the field devices back to the host interface cards, which could also be redundant. Redundant power conditioners for segments further hardened individual networks against single points of failure.
What new challenges await with APL? Available products require either fiber or copper, and conventional Ethernet backhauls from the field switch to the host system. The hardware for using FF infrastructure for APL—twisted-pair Type A cable—still appears to be under development.
If we use existing FF trunk cable for APL, a device for coupling two-wire APL to conventional Ethernet will be necessary—somewhere. Last year’s FF-to-APL migration concept (FCG TR10365) proposes that redundant APL “power switches”—I imagine something akin to PoE power injectors—can be redundant. While consuming two twisted-pair trunk cables, they’re an improvement over dual-redundant FF power conditioners, which typically use a single twisted-pair on the field side. Unlike relatively passive FF power conditioners, APL power switches are active Ethernet devices. Some redundancy firmware must be running to determine which of this pair is actively online.
Read last month's On the Bus column: Incentives to migrate fieldbus to Ethernet-Advanced Physical Layer (APL)
Out in the field junction box, the APL field switch is taking the place of the FF (or PA) coupler. Users preserving a strictly star-network topology will have an advantage because migrating a multidrop or hybrid “star plus multidrop” will have more hurdles. Common FF or PA couplers have few or zero active components, and there’s no silicon running firmware. Aside from short-circuit protection and (sometimes) accommodations for extending spurs into hazardous areas (Zone 1 / Division 1), the coupler is effectively a terminal block.
Having consumed two FF trunks for power redundancy, the FF-proxy APL field switch must integrate into many devices. Though uncommon, fieldbus segments can have 16 spurs (some systems support 32), so strategic combinations of heavily and lightly loaded segments are required. So far, the most spurs I’ve seen on an APL switch is 24. If you take your total fieldbus devices divided by the number of segments and it’s greater than 12, you might be pulling in more twisted-pair cable if you want redundant power from the house/rack room.
Another factor will be the power consumption of field devices. There’s only so much current you can push out on 18 AWG conductors. APL devices promise to be considerably more power-hungry.
Finally, how do you feel about up to 24 devices sharing a single point of failure—the APL field switch? It’s unclear whether the power to field devices will be maintained, for example, during a firmware flash. If diagnostics indicate an issue, how would you exchange it for a replacement without all of the attached field devices powering down?
One hopes that APL field switches that are fault-tolerant, redundant and hot-swappable are on someone’s roadmap. Otherwise, end users serving large continuous processes may be shy about switching” to APL.
