Networks and Communications

Networks and Communications

Getting on the Right Bus

Choosing the Right Network From the Device to the Internet

It may be coincidence that industrial networks share the 'bus' moniker with the Greyhound crowd, but in reality their functions are quite similar. For instance, both transport things - data packets versus people - from one place to another. In addition, many of the same issues that face today's bus-taking traveler are the same problems plant managers confront when choosing a network.

Those who ride the bus, for example, want the most direct route for the least fare. And they want a bus that fits their specific needs, such as convenient departure and arrival times and sufficient storage space for carry-ons. When it comes to industrial buses, parallel concerns are present - and have already been addressed. With today's array of options, engineers can pick the networking bus that offers the most efficient data 'route' and best meets their application needs. This may involve choosing a bus that provides intrinsic safety, low overhead, or a specific data packet size.

The fact that users need several networks in one plant (e.g. process, discrete, control-level, device-level, etc.) presents a new twist to integrating the entire architecture. The current factory situation is akin to taking different modes of transportation, depending on whether you're travelling from city-to-city, state-to-state or country-to-country. Most will agree that taking a bus is economical, but hardly appropriate for going overseas. The same trade-offs must be considered for the plant floor, depending on whether you are transferring data from device-to-device, controller-to-controller or business-to-business.

In the plant, a device-level network such as DeviceNet is generally the most appropriate choice for carrying information from device-to-device. It is cost effective and fast, but it is optimized for small data packets and limited distance. This is one of the many reasons engineers add another mode of transportation to their plant's networking architecture - in the form of a control or information-level network.

Here's where the traveling gets tricky. Because most open networks in the market have different protocols and application layers (and speak different languages as a result), transferring information from one to the next requires the use of expensive gateways to translate and transfer protocol information. Not only do gateways slow data delivery and add complexity, programming and cost, they make networking from the device to the Internet a burden.

To solve this puzzle, Rockwell Automation introduced NetLinx, a communication architecture that uses open networking technology for seamless top-floor to shop-floor integration. The networks that use the NetLinx architecture - ControlNet, DeviceNet and the recently introduced Ethernet/IP - share a common application layer, CIP (Control and Information Protocol). This provides a universal set of networking services while physical and media access methods are very different. Because the NetLinx architecture provides the common language, information can be transferred and understood between all three networks using routers - an inexpensive and less complex option compared to gateways. The benefits are obvious: reduced cost and easier access to the information that users need to improve productivity.

What isn't so obvious is the role that each of the networks play in the factory. Where is DeviceNet typically used? What are the differences between ControlNet and EtherNet/IP? Each network has its own unique characteristics and will, as a result, be more appropriate for certain applications. The remainder of this article will help explain the primary roles of each network.

The Ever-Popular DeviceNet

DeviceNet is an open communication network designed to cost-effectively connect factory-floor devices such as sensors, push buttons, motor starters and drives, to control systems. As the market leader for device-level networking, DeviceNet delivers data at speeds from 125 to 500k bit/s. The message length for DeviceNet is 0 to 8 bytes, which is ideal for simple devices with small amounts of I/O data that must be exchanged frequently. To configure the system or send longer messages, DeviceNet supports message fragmentation allowing the data to be contained in several packets.

The market space for DeviceNet is clear: device-level communication. No other type of network transfers small packets of control data and information as cost-effectively. While it is technically possible to implement Ethernet/IP at the device-level, a significant cost barrier remains. Here's why: DeviceNet uses a multi-drop cable configuration. This means that you could, for instance, connect 50 devices to a single DeviceNet cable. On Ethernet, to be viable for control applications, you must use a switch (point-to-point) configuration, and you'd have 50 pairs of wires running between a switch and the devices. The DeviceNet design is based on CAN, a communication technology that is used in many industries including high volume automotive production. It is relatively simple and does not require significant memory in the device or controller. The result is a very low cost component (as low as $1) and availability from many chip manufacturers. The CPU may even include the CAN protocol, and no special communications component is required. This use of simple and low cost commercial technology is the key to cost-effective products and systems.

The Newest Mode of Transportation - EtherNet/IP

In the spring of 1998, ControlNet International, the open standards organization for ControlNet, applied CIP - the open, published and widely accepted application layer shared by ControlNet and DeviceNet - over Ethernet. ControlNet International and other open network organizations introduced EtherNet/IP, the Ethernet standard for factory automation. The open application layer provides Ethernet users with both information and control services.

So what are the differences between the Ethernet that has been around for 20 years and the new EtherNet/IP? The primary difference is that in order to select products from multiple vendors and have them operate together on the plant floor, the products must be able to speak a common language. For the past decade, vendors have been applying their own application layer to the Ethernet products they have introduced to the market. As a result, Vendor A's Ethernet product hasn't been able to understand the data sent to it from Vendor B's Ethernet product. This is not an ideal situation when trying to take advantage of today's information rich, diagnostic-carrying products.

In addition to interoperability, EtherNet/IP also provides users with control capability, data acquisition, device-configuration and diagnostic capabilities. It employs the standard TCP/IP protocol suite allowing information and control messaging.

EtherNet/IP has evolved thanks to the existing architecture, commercial availability of components and users experience with it in plants. In terms of cost, it is important to note that a commercially available Ethernet PC card may only cost $20, but an Ethernet module used in an industrial controller may cost more than $1,000. This difference is due to the fact that a central processing unit and memory must be included in the module to perform the PC tasks. In addition, industrial products must be built to withstand higher ranges of temperature, humidity, electrical interference, shock and vibration than most off-the-shelf Ethernet products were designed to handle.

The low cost benefit associated with Ethernet, therefore, might refer to manpower and training. For years, IS and IT departments worldwide have been using Ethernet. Such long-term exposure to the Ethernet technology has produced an expansive knowledge base and unparalleled resources. In addition, Ethernet is fast. Based on data rate alone, Ethernet has a considerably higher data transfer rate than many other networks.

Real-Time Control Using ControlNet

If users can achieve real-time control using EtherNet/IP, why is ControlNet still a popular selection? ControlNet was introduced to provide a high-speed deterministic link. ControlNet meets the demands of complex control systems such as coordinated drive systems, weld control, motion control, vision systems, complex batch control systems, process control systems with large data requirements, and systems with multiple controllers and human-machine interfaces. ControlNet is ideal for systems with multiple controllers interlocking between all types of controllers including robots, welders and PCs.

Most users who have selected ControlNet use it because it was specifically designed for complex control systems. For instance, ControlNet uses passive taps, which means the network is unpowered, and loss of any one node will not result in a network failure. In contrast, Ethernet has powered switches. A loss of power results in loss of the network. Users will continue to select ControlNet because they want the benefit of out of box media redundancy, intrinsically safe cabling options, and the many other features associated with ControlNet.

Applications wanting the following advantages are most likely to use ControlNet:

Engineering Productivity with Deterministic Performance. ControlNet configuration software helps eliminate configuration conflicts and device errors by monitoring network bandwidth usage and device communication rates and by alerting users if problems arise. Here's how this comes in handy. Users can configure a sensor, for instance, to update every 5ms and the drives to update every 10ms, and confirmation will be given that there is adequate bandwidth. In operation, the sensor and drive data will always be updated as scheduled without any interruption due to other network activities such as configuration or messaging.

Complete Solution due to Product Availability. Due to the open standard, the maturity of ControlNet and customer requirements, users are able to select from a wide range of products. ControlNet International provides a complete list at www.controlnet.org. This list continues to grow in variety and number of choices.

System Uptime with Media Redundancy. ControlNet offers physical media redundancy as standard. This feature provides extra security against a cut or damaged cable. When implemented, each device is connected to two cables and identical data is sent in parallel on each cable. Alarms are sent to the management system if either cable is not functioning, but the operation continues without any interruption or loss of communication. Furthermore, no manual intervention is required.

Designed for Process Control with Intrinsically Safe Cabling. ControlNet is designed to support intrinsically safe cabling options. This is available due to its bus architecture with passive network components. Specially designed cables, I/O and other components are available for use in the hazardous environment. In addition, fiber optic cabling is available and utilized with barriers that isolate the hazardous area. The system design and use of ControlNet provides a very cost-effective system without unnecessary complexity.

Multiple Media Options. ControlNet cabling components provide flexibility when designing for an application. The trunk cable is the central backbone of the system and is available in coaxial and optical fiber. Users select media based on the environment of their system. In addition, there are special use cables that meet the need for high flexibility, extra or special material for insulation, and many more. A BNC connector is used to connect taps and devices to the network. Repeaters, both fiber and coaxial, are available to extend the length, create ring, star or tree configurations.

Have You Selected the Right Bus?

Choosing the appropriate mode of transportation can be the ticket for efficiently transferring data from the device to the Internet. As frequent travelers enjoy the freedom of selecting between a myriad of travel options - buses, planes and trains to name a few, plant managers can also take advantage of this freedom. However, no one knows better than you whether the right network for your application is DeviceNet, EtherNet/IP, ControlNet or a combination of these networks.

A-B Journal, June 2000