Ethernet has dominated wired computer-networking technology for decades, evidenced by its wide adoption in office, business, data-center, and home applications. Only more recently, though, has industry discovered the benefits of Ethernet. The industrial arena uses include factory automation, process control, building automation, and equipment that connects programmable logic controllers (PLCs), human machine interfaces (HMIs), motor controllers, pumps, valves, robots, sensors, and actuators.
However, Ethernet has some limitations when it comes to industrial applications. To correct these shortcomings, multiple application protocols have been developed. This article identifies the pros and cons of Ethernet for industrial uses and summarizes the most popular application protocols.
The good news is that the widespread usage of Ethernet has led to readily available components and equipment that’s very affordable. Standard cable and connectors, interface cards, hubs, switches, and other gear are often usable in an industrial setting. Special “hardened” versions of all these items also can be found for use in dirty and hazardous industrial settings. The Good News and the Bad News
Different speed versions of Ethernet (10 Mb/s, 100 Mb/s, 1 Gb/s, 10 Gb/s, and others) make it adaptable to many applications. Furthermore, an industrial Ethernet network can easily connect to the standard office network or the Internet.
The bad news is that Ethernet has a problem with timing and latency. Ethernet’s access mode (CSMA/CD) is a probabilistic method that doesn’t fit the needs of industrial control. Industrial applications require determinism that can accurately predict or schedule the arrival of a control packet or signal. Often, shorter reaction times are needed to meet the monitoring or control aspects of the usage. Standard Ethernet responses are usually about 100 ms or more when at times they’re required to be less than 1 ms. Synchronization may also be an issue.
Latency can be reduced by using a faster network or by partitioning the network with switches that isolate segments of the network to reduce message collisions (see “Enhancing Ethernet for Industrial Applications”). As for implementing true determinism, a different networking method and equipment are necessary.
Industrial Ethernet Protocols
Industrial networking manufacturers have recognized the limitations of Ethernet for years. Wanting to take advantage of the Ethernet’s benefits, these manufacturers developed special protocols that overcome the shortcomings of Ethernet. Some of these higher-level protocols are effectively application layers of the OSI model. They encapsulate the Ethernet packets and lower layers, but add features like a master/slave topology, time multiplexing, prioritization, synchronization, or other techniques. Some other protocols work at the MAC layer.
There are over 20 different protocols, each unique to a specific manufacturer or group of manufacturers. Each protocol is complex and a story unto itself, and too much to cover here. The following is a brief summary of the most widely used protocols and their basic features:
• EtherCAT: EtherCAT, developed by Beckhoff Automation, is supported by the EtherCAT Technology Group (ETG). It uses a master/slave network and employs either a bus or ring topology, each of which capable of supporting up to 65,535 slave nodes. EtherCAT implements a standard Ethernet frame, so it works at the MAC layer. Each node carried in the frame is assigned a sequential time slot to carry process data. A standard Ethernet controller is used for the master, but special slave hardware is necessary to read and process its own time slot. A distributed clock synchronized to the master is used to provide determinism. Timing resolution can drop down to 100 µs.
• EtherNet/IP: Known as International Electrotechnical Commission (IEC) standard 61158, Ethernet/IP is the creation of automation company Rockwell. It’s supported by the Open DeviceNet Vendor association (ODVA). EtherNet/IP is an application-layer protocol that operates above TCP/IP and the MAC layer. It incorporates the Common Industrial Protocol (CIP) as the top three layers. CIP provides a standard set of messages and services, as well as timing synchronization, to handle most automated systems. Timing resolution down to 1 ms is possible. Standard Ethernet interface cards and switches can be used with this protocol. Also, improved performance is possible with special MAC-layer hardware and software.
• PROFINET: The PROFINET protocol, developed by Siemens, is part of IEC standards 61158 and 61784. The standard is supported by the British PROFINET International and the German PROFINET Nutzerorganisation. The protocol uses standard Ethernet topologies and hardware as well as TCP/IP. Moreover, it implements a master/slave architecture. PROFINET comes in several versions for different levels of determinism. Class A provides a cycle time down to 100 ms and uses a standard MAC controller. Class B, called PROFINET RT, gives a cycle time down to about 10 ms. Class C is known as PROFINET IRT for isochronous real time. It uses special hardware to provide a cycle time of less than 1 ms.
• Sercos III: Sercos III is the third iteration of this Serial Real-time Communication System. It’s similar to EtherCAT in terms of its master/slave arrangement, but in a ring topology. Up to 511 slave nodes can be accommodated. The protocol is time-multiplexed and cycle time down to 31.25 µs is possible. Sercos III can handle almost any industrial equipment.
These four protocols are by far the most widely used, but others also get play. These include POWERLINK, CC-Link IE, Modbus/TCP, and other entrants, each usually associated with some manufacturer or group of manufacturers. Selecting a protocol is based upon the application, the equipment, and the degree of determinism needed.
Implementing Industrial Ethernet Protocols
Most of the protocols discussed here cannot use a standard MAC controller in the slave nodes, because they’re simply not fast enough. Instead, most hardware uses an application-specific integrated circuit (ASIC) or an FPGA developed by the equipment manufacturer. One key feature is called a cut-through. A cut-through reads and processes the Ethernet frame on the fly rather than reading the whole frame first before processing. This feature allows a port-to-port delay of less than 1 µs.
One alternative to an ASIC or FPGA is to use one of Texas Instruments’ Sitara ARM-based processors that integrates a programmable real-time unit and industrial communication subsystem (PRU-ICSS). This is special firmware that allows you to easily build the cut-through and other features to get the required determinism.
Looking for parts? Go to SourceESB.