Automation Fair 2007: 10 commandments of using Ethernet for control

Before making a plant-floor networking error of biblical proportions, here are 10 commandments for applying Ethernet for control applications, according to Scott J. Johnson, product marketing specialist, Rockwell Automation Logix/NetLinx Business, Mayfield Heights, OH. One of the most frequent errors, he says, involves Ethernet switch selection. And, among other advice: Thou shall not use hubs.

By Control Engineering Staff December 6, 2007

Before making a plant-floor networking error of biblical proportions, here are 10 commandments for applying Ethernet for control applications, according to Scott J. Johnson, product marketing specialist, Rockwell Automation Logix/NetLinx Business, Mayfield Heights, OH. One of the most frequent errors, he says, involves Ethernet switch selection. And, among other advice: Thou shall not use hubs.
Giving a Nov. 14 presentation of similar name at Automation Fair 2007 , Johnson drew nearly 200 attendees: a standing-room only, packed-around-the-edges crowd in a McCormick Place West conference room.
Johnson says the top 10 things to know about Ethernet for control are:
1. Understand system requirements, the working environment, from I/O control to information technology (IT) integration, expandability, security, performance, how much data will travel, how fast and how far, other infrastructure, media, switches, routers, firewall remote access needs. VPN access is great to see if someone has unplugged a device, so you don’t have to return to the plant at 2 a.m. The better you know system requirements, the better chances you will have a successful implementation.
2. Learn the system environment. A critical design decision is if the network should be integrated with corporate IT and management or CIP-connected (Common Industrial Protocol runs through the ODVA network family) through manufacturing operations or isolated. Integration with and from the corporate IT network has most significant impact on design considerations. Traffic, security, IP address consideration are among key issues.
3. Make those in IT are aware of plant floor needs and issues. I/O traffic is multicast data. Traffic on EtherNet/IP (EIP, an ODVA CIP-based Ethernet protocol) has input / output (I/O) implicit and messaging explicit. Who owns the network if EIP runs on the floor? Is there one pool of IP addresses? Who does what? Issues include industrial switch selection, when patches run and under what conditions, and ensuring IT understands that I/O traffic is multicast, at high continuous rates.
4. Segment networks properly. Proper segmenting simplifies network management, security, I/O traffic versus human machine interface (HMI) messaging. It can help with many challenges, including control-network availability.
5. Never use hubs. Hubs are repeaters that allow collisions and broadcast traffic to all ports. Switches eliminate collisions. It’s important for control to segment traffic within an IP subnet. Routers can be used to expand to various subnets.
6. Select the switch with proper features. Choose industrial rated equipment. Select carefully between managed and unmanaged, depending on needs. Only use unmanaged if there’s no potential to overload a device with traffic.

A Rockwell Automation white paper in PDF format has more information about Ethernet switch selection

. Managed switches help with port mirroring, Internet Group Management Protocol (IGMP, multicast) snooping, port diagnostics, Web browser use, diagnostics, troubleshooting, frame duplication. Ignoring required switch features could lead to problems.
7. Select and install the right physical media. For copper, CAT 5e and 6 cables are best for industrial applications. UTP (10 Mbps twisted pair cable) is generally recommended. STP can be used for metal conduits and noisy environments, but ground one end only, or a ground loop will be created. Fiber is best for noise immunity and distance; single mode is needed for high performance, multi-mode for general purpose. As for grounding and noise, the same basic guidelines apply as other network installations. See the Rockwell Automation EtherNet/IP media and installation guide (ENET-IN001) for more on media selection. Standard Ethernet rules apply: design carefully.
8. Understand end-device capacities. The bottlenecks of an EIP system are almost always in end devices. Traffic usually consumes very little infrastructure bandwidth at 100 Mbps (megabits/sec). This month a free EtherNet/IP system layout software tool is expected to be available to help with system layout, telling number of connections, packets per second, green, yellow, red, and other information. Not all end devices are created equally, Johnson warns, adding that Rockwell Automation publishes all its performance-related numbers.
9. Be aware of potential security issues. Follow common practices. Logically separate the IT / corporate network from control network. Severely limit logical and physical access to the control network to those with a legitimate need. Implement plant floor security processes, policies and procedures. Deny access by default, and permit by exception.

10. Don’t hesitate to get help .

Consult with Rockwell Automation experts. Time and money spent upfront will save a lot more time and money later, Johnson pointed out.
For more information from Control Engineering , please see a related story about the Rockwell Automation Ethernet reference guide: Rockwell Automation worked with Cisco to develop references for Ethernet use in manufacturing, which also links to those resources.
Mark T. Hoske , editor in chief
Control Engineering System Integration eNewsletter
( Register here and scroll down to select your choice of eNewsletters free. )