Plug into Industrial Ethernet Protocols
Industrial Ethernet protocols can be standard and unmodified, or modified to be industrially hardened. Perspectives from some protocol proponents clarify needs for various implementations.
Mark T. Hoske, Control Engineering
Industrial Ethernet, a growing force in industrial networking, can include hardware and software more suitable for a rugged environment. The physical layer of Ethernet can include hardened connectors, cables, and switches.
Likewise, the software protocol that runs over Ethernet can be TCP/IP or include extra capabilities on the application layer that can make Ethernet more useful for industrial applications, such as elements of prioritization or scheduling, which are considered appropriate for motion, safety, and other demanding implementations.
Ethernet use in industrial applications has increased significantly over two years. Graph shows percentage of survey respondents who currently use ( Control Engineering Product Research May 2008-top bar) Ethernet protocols, compared to current and planned use ( Control Engineering Product Research December 2006-bottom bar).
A standard since 1983, IEEE 802.3 Ethernet (originally designed for office use) continues to expand for industrial implementations. In fact, all protocols show increased use from late 2006 to mid-2008 (see bar chart). It's clear that an increasing number of industrial applications are plugging into Ethernet.
With this increase in use, Ethernet has effectively become the de facto local area network (LAN) standard for electronic communications, says Jason Chiang, senior technical marketing manager industrial business unit, Altera Corp. After starting as an unshielded twisted-pair cable network, it now delivers “highly reliable, low-cost, 10/100/1000 Mbps performance over long distances, with 10Gbps on the horizon,” Chiang says.
A side effect of Ethernet's increased use is that the word “Ethernet has become a generic term for a collection of “standards,” says Jeremy Bryant, Siemens Energy & Automation. “To some people, it even covers hardware like commercial off-the-shelf switches. The fact is there's no single 'standard' covering everything.”
The definition of “Ethernet” refers to the cabling system for transmitting data between connected devices. “On the ISO stack,” Bryant says, Ethernet is only the bottom two layers (layer 1 and 2) in the protocol stack. It's a bit like the RS-485 standard used by most fieldbuses—the key is what happens on the layers above. Without that, Ethernet is pretty dumb.”
While more devices and applications plug into a rugged Ethernet physical layer, the protocols can be unmodified or modified to be industrially hardened.
While a few industrial systems can pack data into standard Ethernet packets to take advantage of commercial Ethernet, Chiang continues, most industrial solutions require custom software at the OSI Layer 3 (and higher layers) and Media Access Control (MAC) hardware modifications to support multiple fieldbus and industrial Ethernet standards and real-time performance requirements. Chiang says, “Software can be ported to processor/DSP-based platforms or, more appropriately, to one FPGA-based platform supporting a choice of processor and DSP options and flexible Industrial Ethernet solutions.”
Bryant maintains that basic Ethernet “cannot meet the needs of automation. Therefore, all industrial Ethernet protocols are an 'enhancement' of the IEEE 802.3 standard. Put another way, all industrial Ethernet solutions (Profinet, EtherNet/IP, EtherCAT, etc.) are standards based (IEEE 802.3), but none can lay claim to be the industrial Ethernet standard because there is no recognized standards group for industrial networks.”
Software for hardened devices must support real-time applications in a predictable, reliable way, says Lee House, chief technology officer and vice president of engineering, GarrettCom Inc. Ensuring continuous uptime and reliability requires redundancy protocols, fault-tolerant software design, and very rigorous testing standards.
And, while commercial-grade equipment is targeted at wide-spread applications, industrial equipment also must support a wide range of mission critical legacy protocols.
Even so, there is no one-size-fits-all protocol for industrial Ethernet, says Irene Bearly, product marketing engineer, National Instruments (NI). Users must generally make a trade-off between interoperability or determinism and speed.
Ethernet protocols can be divided into three categories; non-real-time protocols, real-time protocols, and hard real-time protocols. Bearly says these are (see “Ethernet protocols, 3 ways”): 1. Non-real-time protocols provide open connectivity over standard Ethernet but have lower response times. 2. Real-time protocols use device broadcast configurations to minimize cycle times or packet prioritization techniques such as IEEE 802.1D/Q for Quality of Service (QoS) to reduce switch induced jitter. EtherNet/IP and Profinet RT are two examples. 3. Hard real-time protocols use custom hardware on the devices and either require special Ethernet switches or eliminate Ethernet switches through daisy chaining of devices. Hard real-time protocols can perform high-level motion control. EtherCAT and Profinet IRT are two examples.
These developments represent a huge leap from just a few years ago, says Mike Hannah, product business manager, NetLinx, Rockwell Automation, when Ethernet's collision detection and lack of determinism made it unsuitable for plant floor use. “Technical advancements like higher-speed throughput, message prioritization, and industrial switching hubs have made Ethernet a viable network throughout the entire enterprise,” Hannah says.
Industrial networking applications require a high level of determinism to ensure that data is reliably transferred on a consistent, repeatable basis, says Chuck Lukasik, director, CC-Link Partner Association.
“Without such deterministic data transfer, the processing or manufacture of quality products is not possible,” Lukasik says. “Control engineers and quality engineers rely on real-time, accurate data exchange within their control systems and between their control systems and plant information systems. Although there have been efforts to apply commercial unmodified Ethernet to the plant or factory floor, the inherent non-deterministic nature of commercial Ethernet and non-ruggedized commercial media do not deliver the required level of high-speed reliable data exchange needed for today's manufacturing systems.” Unmodified Ethernet allows data collisions to occur, detects them, and provides for retransmission, so determinism cannot be guaranteed without the use of additional hardware (Ethernet switches), which complicates network design, Lukasik says.
To ensure high-speed real-time data exchange, CC-Link IE, a Gigabit speed Ethernet, has been “optimized to ensure deterministic data exchange without the need for Ethernet switches but still maintaining the Ethernet physical layer. For continuous data exchange without interruption or data errors in the industrial environment, noise-immune fiber optic media should be used. CC-Link IE guarantees determinism by using a token-passing communication technique. A typical network of 32 stations (industrial controllers, computers, HMIs, etc.) can each transfer 4 kilobytes of data for a total of 128 kilobytes in only 60 microseconds. All network stations are connected by a fiber optic loop rather than a star configuration. Up to 120 stations can be connected on one network. The protocol stack is shown (below). Also see Gigabit-speed industrial Ethernet protocol: A look inside .
CC-Link IE, a Gigabit speed Ethernet, ensures deterministic data exchange without the need for Ethernet switches.
Profinet: layers of performance
Just as there are environmental factors that influence the hardware selection for industrial Ethernet devices, various performance factors influence how Ethernet itself is used, acknowledges Carl Henning, deputy director, PTO, Profibus and Profinet North America.
Profinet uses these four steps that build on each other: TCP/IP, real-time, bandwidth reservation, and scheduling.
These steps collectively are all part of the current Profinet specification and not separate versions. They all work on the same network, representing increasing levels of performance. Profinet uses TCP/IP for configuration and diagnostic information transfer. To avoid the delays and variations inherent in TCP/IP and UDP/IP, Profinet real-time skips these layers. Like all serial fieldbuses (including Profibus and DeviceNet), Profinet real-time uses layers 1, 2, and 7 of the ISO 7-layer stack. Profinet real-time skips layers 3 and 4. This Profinet technique allows sub-millisecond response times for IO. When sub-millisecond response times are not fast enough and no jitter is acceptable, such as for motion control, additional measures must be taken. Profinet uses bandwidth reservation and can use scheduling.
ODVA's approach to industrial Ethernet has been to rely on standard, unmodified Ethernet for its industrial Ethernet network solution, EtherNet/IP, says Katherine Voss, executive director of ODVA. EtherNet/IP is the adaptation of the Common Industrial Protocol (CIP) to standard the TCP/IP Suite in conjunction with the IEEE 802.3 standard and a standard Ethernet infrastructure, she says.
The onces and zeros that comprise the bits in an Ethernet frame are assembled in this way. Source: Control Engineering; ODVA publication "Ethernet/IP-CIP on Ethernet Technology" at www.odva.org ; and IEEE 802.3 spec (2005, section 1, 3-1)
“The flexibility of deploying standard Ethernet allows the end user to take advantage of the full range of capabilities possible with today's Ethernet and Internet. If the application calls for Web browsing, for example, it can be easily incorporated. By relying on standard Ethernet, the user can be assured that an investment is protected and well positioned by taking advantage of future advancements in commercial-of-the-shelf technology, as well as in EtherNet/IP.
“No special hardware or software is required to build an EtherNet/IP device, allowing other protocols using standard, unmodified Ethernet to co-exist in the same device (such as Modbus TCP).
Approaches incorporating proprietary and non-standard technology require extra hardware, software, or both for the device to function correctly and co-exist with standard Ethernet TCP/IP traffic. EtherNet/IP makes it possible to integrate control networks with the standard Ethernet networks used for quality, MES, or ERP systems in the plant, saving users money and allowing them to achieve the proven benefits of Ethernet and Internet technologies,” Voss says. End-users or OEMs can review ODVA test data to determine how rugged or commercial a product needs to be for the application, she adds.
Read more on industrial Ethernet at www.controleng.com.
Mark T. Hoske, is editor in chief of Control Engineering . Reach him at MHoske@cfemedia.com .