Well, I was chugging along researching this issue's cover article, "Alive and growing," about how industrial Ethernet is basically taking over plant floors and process applications, when I hit a snag. Of course, Ethernet is saving cable, digitizing networks, aiding wireless, enabling cloud and virtual computing, and paving the way to the Industrial Internet of Things (IIoT). However, I think all this sunshine pumping about Ethernet created a vacuum in my head, which pulled some reality back in, along with the voices of several cooler-headed sources reminding me that Ethernet is not always as terrific as advertised.
Sure, there's always the possibility of poor connections, unshielded cables, wiring runs exceeding Ethernet's 100-meter limit with no repeater, unmanaged switches or misdirected antennas causing interference, data collisions and dropped communications. Unfortunately, my particular snag came up when I asked about interoperability between devices and communication protocols running on Ethernet.
It had been awhile since I covered industrial Ethernet directly, so I figured all the up-and-coming, speedy, inexpensive, microprocessor-driven gateways and switches with software-based translating functions must have finally broken through the walls preventing Ethernet protocols like Profinet, EtherNet/IP, Foundation fieldbus High-Speed Ethernet (HSE), Ethernet PowerLink, CC-Link IE and others from talking to each other. They must have succeeded by now, right?
Not so fast. Despite the gains in speed and capabilities provided by faster, cheaper chips and software, I learned Ethernet is still hindered by old, artificial language barriers between its protocols, which are still propped up by efforts to preserve old-style market shares.
"Ethernet is now used in every kind of distributed control system (DCS), but each DCS has its own proprietary Ethernet version. For instance, some may be called 'fieldbus' or have almost all the functions of a protocol like Foundation fieldbus HSE, but they aren't HSE and don't comply with it," says Richard Caro, CEO of CMC Associates and an ISA-certified automation professional. The third edition of Caro's book, "Automation Network Selection," is due to be published by ISA this month.
"Everyone can get on Ethernet, just like we can all get on the same phone lines. However, even though my phone can reach and 'interoperate' with Saudi Arabia, I still have to speak English with whoever I'm talking to there. Likewise, Profinet has its own protocol and data frames that ride on top of Ethernet, while EtherNet/IP uses its CIP protocol, and so different stations with either protocol still can't talk to each other and can't interoperate."
Caro adds that some gateway devices, typically with two Ethernet cards, can cross-translate between protocols by receiving a data stream in one protocol, use application software to regenerate that information, and then reframe it in the second protocol's frame structure. This microprocessor-based method is more than fast enough for most process applications, but the latency it introduces is likely too slow for high-speed machining or running an encoder. Plus, this added complexity still serves the purpose of pushing users into accepting single-vendor Ethernet solutions, and aligning with one protocol, instead of truly being able to accessing the whole patchwork of devices and networks that most of them actually have onsite.
"Suppliers could get organized and do the physical programming that a unified Ethernet protocol would need, but there's no real market for it because most users are still tied to one primary vendor's protocol or the other," explains Caro. "Everyone could communicate and interoperate on the same Ethernet network, but only if enough major and regular customers demand a mixed network, and specify a certain level of network interoperability at the application layer for all future system purchases. I still think the push for one network is coming."