Design Article
Comment
jdub_innov
If I understand your proposal correctly, this approach is similar to that being ...
Mike Fahrion
Interesting concept! I'm a big advocate of this sort of system and node design ...
Industrial Ethernet—The basics
Mike Fahrion, B&B Electronics
8/1/2012 1:18 PM EDT
Ethernet and Industrial Systems
Ethernet’s simple and effective design has made it the most popular networking solution at the physical and data link levels, especially for general-purpose office networks. With high-speed options and a variety of media types to choose from Ethernet is efficient and flexible. Using inexpensive UTP cable and star topology, and CSMA/CD media access, Ethernet networks are easily designed and built. Nodes can be added or removed simply and troubleshooting is relatively easy to do. As Ethernet and related technologies have become prevalent in the general networking arena a large base of trained personnel has become available.
These factors, and the low cost of Ethernet hardware, have made Ethernet an attractive option for industrial networking applications. Also, the opportunity to use open protocols such as TCP/IP over Ethernet networks offers the possibility of a level of standardization and interoperability that has until now remained elusive in the industrial field.
However, the probabilistic nature of Ethernet is one characteristic that is a drawback for some industrial network applications. Historically, time critical networking applications have been handled using deterministic networks (using master/slave or token passing schemes). Utilization levels on industrial Ethernet networks must be carefully controlled as levels greater than 10% often result in inadequate performance. Still, as the overall cost/benefits of Ethernet have increased, industrial users have found ways to enhance Ethernet’s data transfer performance. One method is to segment networks using switches and routers to minimize unwanted network traffic and reduce utilization. Another is to use newer, higher level protocols that incorporate prioritization, synchronization and other techniques to ensure timely delivery of messages.
The result has been an ongoing shift toward the use of Ethernet for industrial control and automation applications. Ethernet is increasingly replacing proprietary communications at the plant floor level and in some cases moving downward into the cell and field levels.
Most major control system manufacturers now incorporate versions of Ethernet networks and higher-level Ethernet-related protocols into their product offerings. Often, several manufacturers and/or industry stakeholders have entered into cooperative efforts to develop Ethernet-related standards and products. Several other these now exist, though interoperability between them continues to be elusive.
Ethernet for Control Automation Technology (EtherCAT) is an open real-time Ethernet network developed by Beckhoff. It provides real-time performance, features twisted pair and fiber optic media, and supports various topologies. It is supported by the EtherCAT Technology Group, which has 168 member companies.
Ethernet Powerlink is a real-time Ethernet protocol that combines the CANopen concept with Ethernet technology. The Ethernet Powerlink Standardization Group (EPSG) is an open association of industry vendors, research institutes and end-users in the field of deterministic real-time Ethernet.
EtherNet/IP is an industrial networking standard that takes advantage of commercial off-the-shelf Ethernet communications chips and physical media. The IP stands for ‘industrial protocol’. ControlNet International (CI), the Industrial Ethernet Association (IEA) and the Open DeviceNet Vendor Association (ODVA) support it.
Modbus-TCP, supported by Schneider Automation, allows the well-proven Modbus protocol to be carried over standard Ethernet networks on TCP/IP.
PROFINET is Profibus’ Ethernet-based communication system, currently under development by Siemens and the Profibus User Organization (PNO).
The ongoing level of interest, activity and new product introductions of Ethernet-based equipment suggests industrial use of Ethernet will continue to grow for the foreseeable future.
About the Author
Mike Fahrion, director of product management at B&B Electronics is an expert in data communications with 20 years of design and application experience. He oversees development of the company’s rugged M2M connectivity solutions for wireless and wired networks based on serial, Ethernet, wireless and USB communication technologies. Fahrion has particular expertise in reliable connectivity solutions for devices deployed at the “edge” of networks in remote, harsh or uncontrolled environments. Fahrion is a speaker and widely published author, including his politically incorrect newsletter, uConnections, with over 50,000 monthly subscribers. Fahrion holds a BSEE from Iowa State University.


docdivakar
8/3/2012 5:47 PM EDT
Fairly high level description of industrial Ethernet. I hope there is a more detailed follow up to this article, one that describes how the different types of industrial control area networks can be unified by end users without costly tear downs or replacements. I would also like to see discussion on how long the fragmented industrial Ethernet can survive...
MP Divakar
Sign in to Reply
Mike Fahrion
8/4/2012 11:31 AM EDT
The days of proprietary industrial protocols did a lot of damage. Today, the idea of building any wide scale network with anything proprietary should strike fear into anyone. Off the shelf gigabit and a redundant architecture are more than enough to satisfy most applications. Mixing various network and transport layer protocols adds a degree of cost and difficulty but can typically be managed with a variety of gateways available from many suppliers. For precision timing applications where switched networks won't do the job protocols like EtherCAT provide an effective solution.
Sign in to Reply
I_B_GREEN
8/6/2012 4:06 PM EDT
Why not a 1/2 deterministic and 1/2 probalistic network topology.
Abuse your overusage and get thrown into the probabalistic that again is throttled back to some % say 1/2 of its own total to stay away from the anthing time zero = zero as you~pass the 3/4 full point.
Thus when the node is near ful capacity all the nodes are notified that determinisic BW is all used up and throttle back or take your chances on the probabalistic part of the BW.
It would allow higher throughput if messages are delayed rather than resend collision, esend collision...in the same stupid way we have done things so far.
Sign in to Reply
Mike Fahrion
8/9/2012 12:10 PM EDT
Interesting concept! I'm a big advocate of this sort of system and node design thinking. If nodes are aware of network conditions and able to react accordingly then the burden on the network can be reduced dramatically. A holistic approach like this can dramatically lower system cost as it can be quite expensive to build those 5-9's (99.999% uptime) networks, same is true for determinism.
Has anyone seen this type of network awareness built into automation systems today?
Sign in to Reply
jdub_innov
8/16/2012 11:25 AM EDT
If I understand your proposal correctly, this approach is similar to that being proposed for Audio Visual Bridging. In that scenario, participating nodes register for their required bandwidth. Resulting AVB packets then receive priority when traversing the network. This approach is dependent on several key technologies:
1. A system-wide time base/synchronization mechanism
2. A reservation mechanism
3. A prioritization mechanism
4. A system-wide queuing and forwarding mechanism (this involves traffic-shaping to smooth/distribute high-priority traffic and make optimal use of available bandwidth).
Obviously, this technology requires that both the infrastructure and node be aware of these mechanisms, but it doesn’t preclude network participation from nodes that are unaware of the protocol. They simply do not enjoy the level of determinism of participating nodes.
Clearly, this technology is applicable to broader markets, including automotive networks, and Industrial Controls.
Sign in to Reply