Optimizing Ethernet for industrial implementation
John Morse, senior market analyst, automation and control, IMS Research, answered questions regarding industrial Ethernet implementation.
What are key considerations to optimize an industrial Ethernet implementation?
Availability of skill set—If a user is switching from a fieldbus to Ethernet, an understanding of the modus operandi of Ethernet is essential, particularly regarding the addressing mechanism. Some fieldbuses are very easy to implement, and the Ethernet switch can be quite a change. Also, it is very easy to get blasé about Ethernet when used in a commercial environment where it has become plug-and-play for many users. There are more considerations when it is used in industrial automation.
Infrastructure—Is the cabling up to the job? This is particularly important if gigabit Ethernet is being employed.
Is nondeterminism an issue?
This used to be a big challenge but is less so in recent times due to the common use of switches to isolate sensitive parts of the network and the fact that many applications are unfazed by a small delay in node-to-node response times. Also, few networks are so busy that it’s a problem.
Why use Ethernet instead of another industrial network?
Four reasons: 1) Compatibility with the existing IT system enables seamless integration of data throughout the enterprise. 2) Common use of infrastructure network components can ease implementation. This is more relevant to enterprises that do not need industrially hardened products. In some cases, there are cost savings to be had as the price of components goes down and as demand goes up. 3) Speed helps. Even 10/100 Mbps Ethernet is fast! 4) Ethernet is fashionable right now.
What Ethernet protocol do you need and why?
There are two main types of Ethernet. General protocols, such as Ethernet TCP/IP, Profinet, and EtherNet/IP, and those designed to meet high-speed, deterministic and low-jitter application, such as EtherCAT and SERCOS III [also CC-Link IE and Profinet IRT].
Around 50% of industrial applications work fine with the standard Ethernet TCP/IP.
If determinism is an issue then an industrial variant must be considered. Reasons will include high-speed production (bottling plants are typical) and motion control applications where synchronization of motors/servos is critical. Where there are safety critical issues to consider, such as e-stops, the network must be able to react fast and reliably.
Can multiple protocols be used on the same Ethernet physical layer?
Theoretically, the answer is always "yes" if the protocol uses layers 1 and 2 of the OSI model. However, there are combinations that are not recommended, and it is best for users to check with the standard authority for the protocol before committing to a particular technology.
Should you use fiber or copper?
Typically, fiber optics are used for backbone with “drop-downs” using copper. Naturally, fiber-optic cable is immune to EMC but more expensive and harder to terminate reliably. Few companies appear to be using fiber-optic cable for other applications apart from those needing to transport data over a long distance. This is more typical in process industries where the sites are larger.
– Edited by Mark T. Hoske, CFE Media, Control Engineering, www.controleng.com.
Control Engineering webcasts include more on Ethernet optimization.