Twenty years ago, Ethernet was dismissed for industrial networking. But slowly, out of the rat's nest of front-office data networks, Ethernet has reached out and touched the plant floor, More than touched, in fact. Analyst firm ARC Advisory Group Inc. (Dedham, MA) forecasts that the use of Ethernet-ready products on device-level networks will explode., Explains Harry Forbes, senior analyst at ARC, "Ethernet is becoming an intrasystem interface." Forbes suggests that people are looking at ways to utilize Ethernet for factory control applications that push the proverbial envelope vis-a-vis its networking capabilities.

For a number of reasons, Ethernet is Enticing for industrial networking applications. It's tried and true; that is, it's popular, widely used, and used for years by information technology (IT) departments (and offices) worldwide. Moreover, compared to proprietary networks, Ethernet networks are easier to install and maintain; the networking technology is readily available, and far less expensive.

There's another enticement: The ability to standardize an entire enterprise--from the plant floor to the corporate boardroom--on one network. Such a network would greatly amortize the costs of installation, maintenance, and training. What's more, it would also promise greater access to production data throughout the enterprise, from anywhere around the world.

According to Hirschmann-USA (Pine Brook, NJ), a supplier of networking solutions, "Over the past five years, there have been many enhancements to the Ethernet standards, especially in areas of determinism, speed, and prioritization. There is no monger any reason why Ethernet cannot be used to build deterministic fieldbus solutions that are cost effective and open."

ETHERNET EVOLVES

When first adopted in the mid-1980s as the Institute of Electrical and Electronics Engineers (IEEE) 802.3 standard, Ethernet was considered unsuitable on the plant floor. The network protocol was (and still is) nondeterministic. Device response times could not be guaranteed because of data collisions and the delays in retransmitting data. Data throughput was slow. The network medium was subject to electromagnetic interference (EMI).

That was then; this is now. Nowadays, Ethernet now runs over shielded and unshielded twisted pair copper, coaxial, and fully EMI-resistant fiber-optic cable. (There's another medium. Hang on.)

Next, Ethernet operating speeds across even conventional wire cabling have increased at least an order of magnitude, from 10 megabits per second (Mbps) to 100 Mbps. Automatic switches that negotiate 10/100 Mbps are commonplace, thereby optimizing speed versus service, as well as letting you mix 10 Mbps (Ethernet) and 100 Mbps (Fast Ethernet) devices on the same network. The higher speed also reduces the probability of data collisions (thus lost data). The delays that do exist are so short as to be non-issues. For most factory applications, 100 Mbps Ethernet is deterministic enough.

Today, Gigabit Ethernet is available. This is Ethernet running at 1 Gbps, though the latest push is 10 Gbps. Gigabit Ethernet is primarily targeted lot enterprise-wide backbone networks; however, it is showing up in the backbone of distributed control systems in the process industries. Explains Forbes, "It's just for scalability. When users want to scale their [data] 'pipe' bigger, they don't have to do anything special; they just go buy a little more expensive switch."

DETERMINISTIC NON-DETERMINISM

The technology evolution most responsible lot minimizing Ethernet's nondeterminism are the advanced switching technologies that let multiple devices simultaneously transmit and receive data over multiple network loops. Unlike an Ethernet hub, which bridges all network ports into a common pool, an Ethernet switch lets users divide a network into virtual local area networks (LANs). This segments devices into logical workgroups, helping local data communications stay local. Ethernet switches also typically have a last internal backbone, which helps eliminate collisions among data packets and, therefore, lost packets. ARC points out that simply swapping a $400 Ethernet switch for a $100 Ethernet hub can make Ethernet more dependable. Even though both link devices and network rings together, fast switches can quickly swap network lines, thereby responding to anomalies quickly, such as miscommunications, power failures, and device failures.

For instance, the industrialized ED6008 8-port EtherDevice Server from Moxa Technologies, Inc. (City of Industry, CA) has redundant Ethernet ring capabilities so that when any segment of the network is disconnected, the server automatically recovers in 300 ms (with 120 nodes connected and a full load of network traffic). Plus, the server dynamically warns technicians when power fails or a port link breaks. It even sends warning emails when Ethernet traffic builds up.

Adding some intelligence-namely, software or firmware--to the switch improves quality of service (QoS) and adds queue management capabilities. According to officials at Cisco Systems, Inc. (San Jose, CA), "By assigning a priority to time-sensitive data, intelligent Ethernet switches can elevate that traffic above lower-priority data. This ensures that high-priority traffic always traverses the network even if the network becomes congested." Such capabilities are in the Cisco Catalyst 2955, a 12-port 10/100 Mbps switch [or linking programmable logic controllers (PLC) to factory floor networks. The switch has no fans, runs on a 24-volt DC current, operates at extreme temperatures, and can withstand extreme shock and vibration. Depending on the model, the switch can include two single-mode or multimode Fast/Gigabit Ethernet fiber uplink ports--for less than $3,600. The switches also have Qos capabilities that can classify, reclassify, police, mark, and even drop incoming data packets as application priorities require.