According to IMS Research fieldbus–based communications systems accounted for 75% of new industrial automation network connections in 2011, three times more than those using industrial Ethernet. IMS does, however, say Ethernet is growing faster and will become the dominant industrial networking technology within 10–15 years. The IMS study labels all non-Ethernet industrial networks as fieldbus, but for this article we'll use terminology more common to industrial automation. We'll define fieldbus networks as those used in process applications to connect instruments and analyzers to the main control system, typically a distributed control system (DCS). Leading examples are Foundation H1, HART and Profibus-PA.
SEE ALSO: Ethernet: The Promised Land?
We'll identify device-level networks as those that connect discrete devices such as sensors, switches and motor starters to controllers, most typically PLCs and PACs. Some of the most popular are AS-i , CANopen, DeviceNet, IO-Link, Modbus and Profibus-DP.
Despite the Ethernet onslaught, device and fieldbus networks are still widely used in industrial applications. This article will explain why, using real-world applications, these networks are preferred over Ethernet in particular applications. We'll also gaze into the future, and speculate on the future of device, fieldbus and industrial Ethernet networks.
Simpler and Cheaper
When it comes to connecting field-level equipment — such as transmitters, control valves, motors, proximity sensors, encoders and similar monitoring and control devices — to control systems, device and fieldbus networks are often the simplest and cheapest way to go.
"Process controllers, temperature controllers, small ac drives and similar equipment typically have built in RS-232/RS485 ports," explains Paul Johnson, senior electrical engineer at CMD Corp., Appleton, Wis., which manufactures converting equipment for blown film (Figure 1). "Since most HMIs and PLCs have the same ports, they can support general-purpose protocols such as Modbus or DF1. We choose Modbus to save money. Ethernet support in many cases is an added expense. In some cases, Ethernet doesn't meet application requirements. The issues include too high a bandwidth, static noise, and its inability to interface to legacy equipment."
Carl Henning, deputy director at PI North America, the association for both Profibus and Profinet, agrees. "It makes no sense to put an Ethernet connection in a proximity switch or any other simple sensor or actuator," he maintains.
Ethernet can get complicated, too. "Ethernet typically calls for external hardware, architecture design and IT department involvement that industrial control engineers want to avoid if possible," says Bryan Sisler, product manager at ABB Low Voltage Drives.
Keeping the IT department out of the picture is easier with device and fieldbus networks. "Modbus can be handled by currently embedded protocols, no additional hardware is required, and speeds are adequate to handle most system requirements," Sisler says. "CANopen is simple to connect and use, is really open, PC-friendly, and great for low node-count systems. DeviceNet provides determinism, plug-and-play system components, and a hardy transceiver that can survive transients that Ethernet transceivers cannot." These, adds Sisler, are all reasons to avoid the expense, time consumption and complications brought on by more sophisticated networks, "especially those that might require the involvement and training of a new group of non-engineering folks."
Mallard Creek Polymers, Charlotte, N.C., a maker of latex products, also uses Modbus. "Device networks tend to be preferred in many cases due to their broad installation base, determinism and familiarity," says engineer Matt Bothe, who explains that MCP uses Modbus to communicate with several smart, mass flowmeters in order to extract flow and totalized measurements.
"Device networks generally possess advantages over analog/discrete systems, namely far less wire, ease of calibration, access to more data, ease of troubleshooting, and ease of installation," Bothe adds.
Determinism and Distance
One difference between device and fieldbus networks and Ethernet is determinism. Proponents argue that device and fieldbus networks guarantee a response time, while Ethernet can't, and this can be vitally important in certain applications.
"For position-dependent operations (Figure 2), deterministic behavior is a must," states Nick Hunt, manager of automotive technology and support at robot builder ABB Robotics. "For example, when a dispenser is told to turn on exactly one second before the robot arrives on point, it needs to be extremely consistent. We can even live with the fact that one second actually turns out to be 1.32 seconds. If it's consistent, it easily can be tuned in by including the latency in the robot controller's I/O config file."
And consistency seems to be the key. Ethernet, as fast as it is, isn't deterministic, so it can't provide the consistency needed for factory automation when timing is critical. "Speed is part of it, but speed really doesn't address the determinism issue, only the packet transfer rate," Hunt notes. "Determinism is addressed by the way collisions are handled, which they're not with Ethernet." Hunt goes on to say the raw speed of Ethernet can overcome most of the determinism concerns, but not all.
Another issue is distance. Wired Ethernet is limited to about 100 m, but device and fieldbus network distances can be much longer.
"We worked in a gypsum board mill during a process speed increase, and were increasing the number of drives in the system from 40 to 70," relates Brian Radichel, business development manager at Purvis Industries, Dallas, Texas, a manufacturer of material handling and mechanical systems, of his experience at a prior company. "The backbone was Profibus-DP, and in a gypsum plant the distances between drive nodes and CPUs can be long. The drives were clustered in many cases, but one application required distances of 200-300 ft between each drive. Profibus easily handled the increased devices and distances with minimal network changes. The installation was slightly more complicated due to the daisy-chain topology of Profibus, but once the network was set up and addressed, the commissioning proceeded smoothly."
Getting Ethernet to cover the same distance with fiberoptic cable or routers can get expensive, particularly in a daisy-chain configuration. "In outdoor, long-lead-length applications, the cost of running fiber for Ethernet makes Profibus-DP or other high-speed serial networks attractive for both performance and cost of installation," he adds.
Device networks have been around for a long time, and engineers have found ways to use their advantages to solve problems that Ethernet can't. Johnson had just such a problem with networking on CMD's machines. "We have applications that require devices to be connected through slip rings," he explains. "10 MHz/100 MHz Ethernet doesn't transmit very well through slip ring assemblies. We had to use a low-speed DeviceNet running at 56 kHz to avoid the problem."
Fieldbus in Process Automation
Fieldbus networks are widely used in the process industries and offer several advantages over Ethernet. Larry O'Brien, global marketing manager at Fieldbus Foundation, doesn't see Ethernet encroaching on process automation networks at the fieldbus level. "Process field devices require power, and Foundation H1 provides digital communications and power over standard twisted-pair wiring," O'Brien says. "Ethernet does support power over Ethernet (PoE), but the primary use of PoE is for phones, panels, access points and cameras — not field instrumentation."
Process industries also have other requirements that are not met by simply implementing Ethernet at the physical layer, such as operation in hazardous areas. "Foundation H1 is a two-wire, twisted-pair field level network that can be installed safely in a hazardous area. It uses simple screw terminations, which are comfortable to a device installer."
Diagnostics specific to process automation and control in the field are other key advantages of fieldbus networks. "Foundation H1 provides sophisticated, diagnostic, data-management capabilities, and a block structure that allows end users to implement function blocks in control valves or field devices for the purpose of implementing control in the field," O'Brien explains. "There is evidence that control in the field has an 80% increase in meantime between failures compared to traditional DCS control.
"A good example of the value of control in the field can be found at the Shin-Etsu chemical plant in the Netherlands. At this facility, the interface card in the DCS failed, which meant communication from the DCS to the field devices was no longer taking place. Instead of a plant shutdown, the control-in-the-field functionality of Foundation H1 enabled continued operation via direct communication among measurement devices and valves."
The 100-m distance limitation of Ethernet-based systems is also a factor. With Ethernet, O'Brien says, "distances are significantly shorter; it has no multi-drop capabilities; and it's more susceptible to noise that can degrade overall network performance. You can use fiberoptic cable with Ethernet, but that's even more challenging to install."
While proponents talk of Ethernet's ability to support wireless, the same capability is available with WirelessHART. In fact, since hundreds of thousands of process instruments already have HART installed, connecting to them via WirelessHART is simpler than with Ethernet.
Future Device Networks
ABB Robotics' Hunt says Ethernet is moving into automation at the device level. "Ethernet-based protocols like EtherNet/IP and Profinet have been widely accepted as viable in robot end-of-arm tooling and PLC communications," he points out. "The original concern with using Ethernet-based networks was fear that the level of determinism was not suitable for communication between the scanner and the device nodes. The determinism debate still rages, but doesn't appear to be cause for concern. Device networks, though respected and considered as robust, proven and deterministic, are being replaced by Ethernet."
Henning seems to agree. "Ethernet will work its way further down into simpler and simpler devices," he opines. "If devices are available with industrial Ethernet connections, use them. Industrial Ethernets are faster, have greater bandwidth, unlimited node counts, improved diagnostics, easier upward integration, can use standard wireless, and have more topology options. We actually spend some time in both our Profibus and Profinet one-day training classes on the advantages of Profinet over Profibus."
O'Brien also sees the value of Ethernet in industrial applications; that's why Foundation HSE, is based on high-speed Ethernet. "Running at 100 Mbit/s, Foundation HSE is designed for device, subsystem and enterprise integration," he explains. "It supports the entire range of fieldbus capabilities, including standard function blocks and device descriptions, as well as application-specific, flexible function blocks for advanced process and discrete/hybrid/batch applications."
Joey Stubbs, North American representative of the EtherCAT Technology Group, adds that, "Actually, there is no reason why a device network can't also be Ethernet-based for industrial applications. Obviously, trying to replace a simple serial network on legacy devices wouldn't be attractive if this results in a complex network with many switches or routers, and suddenly requires IT support. However, not all Ethernet fieldbus systems are created equal. EtherCAT, for instance, is appropriate for high-speed, high-precision devices, and also for enabling simple devices to communicate at lower scan rates — even with low-powered embedded controllers with less processing power. It's difficult to find controllers without on-board Ethernet ports, which is the only hardware requirement for an EtherCAT master."
"EtherCAT is the standard network for our control solution, with approximately 99% of our controls and I/O devices having EtherCAT connections," says James Wood, director of engineering at Advanced Blending Solutions, Menominee, Wis., a manufacturer of blending and extrusion systems.
Mike Rasner, Advanced's president and CEO, says the biggest factor favoring EtherCAT was its easy connectivity to CANopen and networks such as DeviceNet, Profibus and EtherNet/IP via OPC. "We use bus couplers to tie different networks to the controller via EtherCAT, and this provides quick and seamless connectivity," Rasner eplains. "In our business, we need to interface to a variety of networks from our main controller, so connectivity is very important."
Similarly, XCS Systems, a system integrator in Worcestershire, U.K., specializes in high-speed process control in the packaging industry. Using EtherNet/IP, the company recently integrated more than 160 motors and drives into a new can conveyor line, simplifying equipment design, configuration and commissioning. In the past, XCS Systems relied on a dedicated motion control network for motor and drive synchronization. Today, they rely on a single network infrastructure for discrete and motion control.
"The EtherNet/IP network does all the interlocking between machines, improving machine synchronization," says Paul Croad, systems integrator at XCS. "It allows engineers to program any machine from anywhere in the line."
Ethernet also can serve the needs of both automation and process control at the same time. Gram Equipment, Vojens, Denmark, a manufacturer of ice cream equipment, uses EtherNet/IP for discrete, motion, process and safety control, as well as HMI and enterprise-wide information management.
Scott Bivens, director of electrical engineering at R.A Jones & Co., a packaging machinery OEM that was recently bought by Coesia Group from Oystar Group, views the future of industrial networking by saying, "Everything is going Ethernet. We are migrating to Ethernet on pretty much everything because Ethernet is easier to design and commission, and speeds are faster."