Determinism and Network Wars
Pioneers in software-based machine control had to create a software ‘tick-rate’ that was consistent, regardless of the magnitude of the machine control program, at least up to a point. Early scan rates of 20 msec were adequate for many machines, but not for all, and not for long.
With new developments in ASICs, DSPs, FPGAs, and other CPUs, processing speeds improved dramatically to predictable 20 µsec rates in multiaxis synchronization cases, and routine update rates of 1 msec on HMI and PLC software control packages for more timely access to networks. Determinism in networks became, and remains today, a huge differentiator between various suppliers of fieldbus and Ethernet networks.
First there were the fieldbus hissy-fits, then the Ethernet flavor skirmishes, and now we enter the newest era with mesh and wireless network solutions that might or might not turn into opposing camps of standards-think. Back then, we examined how Ethernet cable was slithering from the office into the factory. In our “Tomorrow’s Technologies Identified” cover story, Feb. ’98, many were giddy that “Ethernet PC boards sell for only about $50.” We also reported that “many PCs today include an Ethernet interface and Windows NT has built-in Ethernet drivers.”
Regarding wireless, we wrote in that same story, “It’s a stretch to think your customers will be telling you to interface your machine with their wireless network in 1998, but the technology may be emerging, although it might take time to gain standardized application strength.” Business Week called wireless one of the most over-hyped technologies of the ‘90s. Pretty accurate, though today it’s getting closer to fulfilling its potential.
Keeping in mind the roll out of Profinet in recent years, in “Profibus Supporters Ponder Life with Ethernet,” Oct./Nov. ’99, we reported that one of the vexing issues discussed at PTO’s annual meeting was what to do about Ethernet. “All agreed that Ethernet was moving closer to the factory floor.” But, they continued, “the main sticking point on Ethernet right now is the lack of a standard at the application level. A proprietary protocol on top of Ethernet still is a proprietary solution.”
A working group made up of representatives from Schneider Electric, Wago, Softing, Kloeckner Moeller, Hirschmann, and Siemens envisioned a 10-year process that “would culminate in complex field devices connected directly to Ethernet.”
The stakes were raised in “General Motors Standardizes on EtherNet/IP,” Nov. ’03, when “GM standardized on EtherNet/IP for its vehicle manufacturing control systems to provide real-time communications between GM machine controllers, robots, and process control equipment, and provide information to higher-level business systems.” GM said it was giving the suppliers to more than 60 factories around the world until Jan. 1, 2007 to make their products EtherNet/IP-compatible. That got everyone’s attention about the inevitability of Ethernet-coated networks.
Connectivity
Real connectivity remained elusive in “Connectivity–Past, Present, and Future,” OEM Insight, May ’99, in which Nick Jones, then CTO at ODVA, reported, “When the software goal of total hardware independence is attained, system designers will be inclined to select the control software first and then select a bus and remote devices that support distributed applications and provide suitable performance.”
An Industrial Networking story, “Cast in Ethernet,” Winter/Feb.’04, related how Rimrock Corp., a die-cast machine builder, improved its customers’ performance while introducing them to the benefits of Ethernet-enabled controls. “I talk to customers and they ask, ‘Why do I need that? I’m not going to network my equipment unless the big boss says to do it.’ I tell them we’re providing the capability whether you need it now or not,” explained Dave Woods, Rimrock’s product development manager. Woods actually was reluctant initially to incorporate Ethernet. “I was a fieldbus snob, but the more I learned, the more it made sense.”
As real-time networks came to the forefront, our Winter/Jan. ’05 Industrial Networking article “Real Choices for Real-Time Networks” reported, “The selection process for real-time network performance starts by understanding network architecture differentiates.” Industrial network guru Ian Verhappen noted that, “Because the life cycles of most plants go beyond 15 years, pick a network that can stand the test of time.”
With movement to distributed control underway, our Spring ’06 Industrial Networking story, “Training, Techniques, Tips,” emphasized the importance of training and cooperation between the factory floor and the IT group. “Without [them] the prospect of building a dependable data network is just an illusion.” This level of cross-training, we said, raises vital awareness of the issues on both sides of the firewall. “IT should be aware that the control network might be sensitive to heavy traffic. Do any control devices require a hole punched in the firewall so that it can be remotely accessed via the Internet?”
As networks proliferate, security issues grow as networks extend. Industrial Networking tackled this in “Fence Me In; Common Sense Security” cover story, Fall ’06. “There are two kinds of network security, real and imagined, and it can be hard to tell them apart, especially if you don’t know how they work and the problems they’re supposed to solve.”
We also noted the work that lay ahead: “We’ve simply had a lot of traditional neglect of security, so now we have a lot of confusion,” said Rich Clark, information security analyst for Wonderware. “Most users and their companies either don’t have security policies and procedures, or their procedures are far out of date.”
