Click here.
ISAMarketplace | Member Resources | ISA News & Views | Fieldbus.ISA.org | Safety.ISA.org | Sensors.ISA.org
InTech Magazine | Industrial Computing Magazine | Events | ISAJobs.org | ISA Training | Educational Foundation
ISA Bookstore | Certification | Motion Control Magazine | Web Resources | Press Room | Reach Your Market
Education Services | Global Connections | Students & Advisors | ISA Transactions | Buy Standards | FAQs | ISA Directory
Home >> Industrial Computing Magazine >> Opinions >> Article
08 Jun 2001

 - ISA Member Login
 - Guest Login
 - Forgot Password?
 - Site Map
 - Join ISA
 - Sections
 - Divisions
 - Standards
 - Pressroom
 - Who's Who at ISA


 News
 Briefs
 Features
 Calendar
 Opinions
 References
 Products
 Subscriptions
 Editorial FAQ
 Media Kit


15 October 1999
Switches Keep Data Routed on Ethernet Track
by Stephen Woram

Many of today's industrial-control networks rely heavily on proprietary networks to transmit data between control network devices. With the proliferation of supervisory control and data acquisition and human-machine interface software, as well as the current sensor-to-boardroom network integration goals, the desire for communication on one network among plant-floor devices is fueling the need for a more open networking solution.

Making its way as the communication network of choice is Ethernet, with its easy accessibility in the commercial market and its open, multiprotocol ability. A concern in the plant, however, is that if you connect factory-floor Ethernet nodes on an Ethernet system, you run the risk of an arbitrary event, such as a network backup or a malfunctioning network interface card (NIC), causing havoc and reducing the speed and reliability of the plant-floor control system. With traditional shared Ethernet devices (e.g., hubs), these concerns are understandable; however, new advances in plant-floor Ethernet networking technology now address these issues. Ethernet switching hardware lets the plant-floor Ethernet network operate reliably at amazing speed while remaining open and easily connected to other company networks.

To monitor data traffic in many proprietary control networks, a signal (or token) is passed among the nodes on the network, which then lets each node transmit the data only during a controlled window of time. (In fact, each node in turn holds the token, even if most of the nodes have nothing to send). This collision-avoidance architecture ensures that collisions won't occur and that each node communicates in an orderly, albeit slow, fashion. These control networks, however, tend to be expensive, available from only one supplier, and slower than Ethernet, due to their proprietary design.

When an Ethernet network node wants to communicate, it first listens to its network connection to determine whether there is any traffic on the network. If there is no traffic at the instant the node checks the network, then the Ethernet node will transmit Ethernet packets. Similar to when you get on the telephone at home and realize someone else in the house is on the line, an Ethernet node will wait until the network is free of traffic. In this multiple-access model, however, two nodes that check the line and "hear" no traffic can simultaneously transmit a packet, causing the two packets to collide and the transmission to fail. When the two packets collide, all the nodes on the shared network halt communication and briefly wait before transmitting again.

Obviously, as the amount of network traffic and/or nodes increases on a shared Ethernet network, so does the opportunity for collision. Traditional Ethernet hubs, which allow multiple Ethernet nodes to plug into a central communications device, are simple sharing devices without the ability to sort or segregate data packets. If a data packet is sent from a PC to a hub, the hub will broadcast the packet to the NIC of every device plugged into the shared network, momentarily delaying communication among the other devices. This means it is possible for nonessential devices to delay the communication of two controllers connected to a shared network hub. Shared Ethernet networking using basic network connectivity tools, such as hubs or bus-style architecture, creates two possible problems for industrial use:

  • Potential slowdowns due to network unavailability stemming from too many nodes in a single shared network (known as a collision domain)
  • Slowdowns caused by the communication back-off algorithm, which each node runs after a collision

With the advent of low-cost, plug-and-play, industrialized Ethernet switches, however, the concerns regarding collisions and network traffic slowdowns can be relieved. An Ethernet switch can generally replace the traditional hubs used in the plant. Switches perform the following important tasks:

  • They automatically learn the addresses of the unique hardware devices plugged into each port. Plug-and-play node mapping in a switch directs traffic to each node only if the traffic is specifically destined for it. This feature ensures that each node can always get a "dial tone" unless a data packet is being sent specifically to it.
  • They enable multiple concurrent Ethernet transmissions. If each node is on its own dedicated port connection, multiple nodes can simultaneously get a "dial tone" and transmit their Ethernet data packets concurrently. The switch, which often operates at 100 MBps of backplane speed, will continuously route the different packets to only the designated nodes. An Ethernet switch can even buffer and forward packets destined for the same node that simultaneously arrive at the switch. Because the backplane of the industrial switch is often operating much faster than the nodes attached to it, multiple nodes can send large packets through the switch simultaneously, without fear of bogging down network communications speed.
  • Switches can segregate essential and nonessential traffic on the same physical network. Because switches can determine each packet's address, nonessential office traffic won't be sent to controllers unless specifically addressed, which allows plant and office traffic to interact when needed.
  • New industrially designed Ethernet switches support placing advanced Ethernet communications devices in the same field enclosures as programmable logic controllers and I/O racks. DIN rail mounting, redundant 24-VDC power, wide operating temperature ranges, and hazardous location ratings add to the flexibility of these powerful devices. Housing your Ethernet equipment in an environmentally controlled location is no longer necessary. A distributed network component architecture, as well as the ability to create redundant communications paths between switches, eliminates single point-of-failure concerns.

Today's Ethernet switch technology offers data-integrity safeguards that in the past could be provided only by wiring distinct, proprietary control networks. By connecting every node to a switched port to eliminate collisions and keeping control and office traffic from interacting unwontedly on one physical network, you can enjoy Ethernet's open architecture and massive bandwidth without compromising your control traffic integrity.


Additional Information

Author Information

Stephen Woram is president of Industrial Networking Solutions, a distribution firm in Dallas, Texas, that provides networking and communication products for industrial applications. Woram has more than 11 years of experience in the industrial controls market and has conducted numerous seminars across the U.S. on implementing Ethernet in plant-floor environments.



Click here.
Click Here
All contents copyright© 1995-2001. All rights reserved. Privacy and Legal Policies. Contact ISA at info@isa.org.
67 Alexander Drive, Research Triangle Park, NC, 27709, USA; Phone: 919-549-8411; Fax: 919-549-8288
ISA - The Instrumentation, Systems, and Automation Society