Don't Rip Out Your Legacy Network
While use of industrial network standards such as Arcnet, DeviceNet, Modbus, Profibus, and others continues to grow, we should not try to decide between them or Ethernet, but embrace their coexistence. As long as existing networks are performing as desired, it may be difficult to justify converting all sensors, controls, actuators, and related I/O modules so they can be connected to Ethernet.
While use of industrial network standards such as Arcnet, DeviceNet, Modbus, Profibus, and others continues to grow, we should not try to decide between them or Ethernet, but embrace their coexistence. As long as existing networks are performing as desired, it may be difficult to justify converting all sensors, controls, actuators, and related I/O modules so they can be connected to Ethernet. Additionally, industrial Ethernet installations require many considerations related to cabling and infrastructure devices (routers, firewalls, managed switches, etc.) that should be taken into account when determining the overall cost.
Many devices available to connect networks go by names such as gateway, bridge, router, device server, linking device, and media converter. With industrial Ethernet protocols or other networks or fieldbuses, these devices may act as slaves, masters, or a combination of the two, or support peer-to-peer messaging. Knowing these things in advance will help you design in the functionality that you need. Following are several examples.
A slave-master I/O gateway is a master to the end devices on network 1 and an I/O slave device on network 2. It transfers outputs from the controller on network 2 as outputs to the devices on network 1. It will also scan the devices on network 1 for inputs and transfer these as inputs to the controller on network 2.
CIP : EtherNet/IP is based on an underlying protocol called the Common Industrial Protocol (CIP), which is the basis for DeviceNet and ControlNet. Because these protocols share the same basic message structure, it is possible to route messages from an EtherNet/IP-based scanner to a DeviceNet device via a gateway.
Modbus : When considering connectivity between Modbus RTU and Modbus TCP, the core structure of a Modbus-RTU packet is the same for a Modbus-TCP packet. This simplifies routing Modbus messages from Ethernet to a legacy Modus serial network.
A slave-slave I/O gateway is a slave to the controllers on both networks. Outputs received from the PLC on network 1 are sent as inputs to the PLC on network 2 and vice-versa.
When information needs to be shared between two networks, the protocols may be based on different packet structures. In these cases, an I/O gateway may be used to pass I/O data back and forth. I/O gateways are designed to interpret messages and extract data from the Ethernet side of the gateway, then build a new packet and insert the data needed on the legacy-network side.
Slave-Master : Use of “slave-master” (also called parent-child) structure can occurs when you choose devices for their best-in-class performance for a particular application, but these do not support the industrial Ethernet protocol used by the rest of the control system. This problem can be solved by a gateway that acts as a slave device on the Ethernet side and a master to the devices.
For example, a PLC using Modbus-TCP may need to communicate with a motor control center that is equipped with Profibus devices inside. Outputs from the PLC would be written to an internal memory table in the gateway. The gateway would be configured to map these outputs to the various Profibus slaves. The Profibus master side of the gateway would be configured just as any other Profibus master using a configuration tool that uses the slave device's GSD files. The gateway continuously will scan the devices for inputs, and map these to the internal memory table to be read by the PLC via Ethernet.
Proxy : For Profinet, a specific type of gateway has been created called a proxy. A proxy device would also act as a master to a network of devices on a legacy network. However, it is based on a specification defining the way Profinet and the legacy network should work together. A proxy may be in the form of a stand-alone gateway or it may be integrated into a PLC. For example, a PLC may communicate with a series of actuators and sensors using AS-interface and be simultaneously connected to a Profinet network.
Slave-Slave : Some scenarios involve integrating two separate network technologies with separate controls, such as a machine with a fieldbus technology being installed in a plant using Ethernet technology. In this case, the PLC on the machine only may need to pass specific data to the plant network. A gateway acts as an I/O slave to both the machine and plant networks, allowing the controllers on each network to control the information being transmitted.
Communication between the gateway and the Ethernet PLC is based on the protocol being used. Communications between the gateway and a serial device can use a wide variety of protocols, either standard or proprietary.
Serial Gateways: Since it still will take some time before all devices have an integrated Ethernet port, there will be a need to communicate to some devices via their serial interface (RS-232/422/485).
One type of gateway on the market is called a device server or serial server. These are used in applications where the need exists to convert from serial protocols to Ethernet, but the underlying application data does not need to be changed. These are often used to allow programming of serial devices over a network and involve the use of software that routes serial packets to IP addresses.
Serial protocol converters
In other cases, serial devices may need to be connected to a PLC that is utilizing an industrial Ethernet protocol. This can be accomplished by a serial-to-Ethernet gateway that acts as a protocol converter. This type of gateway must be able to communicate using the native protocol of the serial device. There are a wide range of serial protocols. Some are open standards and some are proprietary. Some have complex message structures and others use simple ASCII strings.
In any scenario, the gateway should be able to interpret the serial message to extract the data. This data can be mapped to an internal memory table that would then be read by the controller via Ethernet. For example a bar code reader can send an ASCII string to the gateway after it scans a new bar code. This string will contain the data from the bar code plus some framing, identification and checksum characters. The gateway can strip off all extraneous characters and just map the bar code characters as inputs to the Ethernet PLC.
It is clear that legacy networks are here to stay for the foreseeable future. It is also clear that industrial Ethernet will proliferate deeper into manufacturing facilities. To achieve the best outcome, you should plan for this situation and design your systems to use multiple networks.
To do this effectively, look for gateways that have the features and performance you need, and are simple to configure, install, and maintain.
Related links and ONLINE extra on industrial Ethernet
More from the October 2008 Control Engineering supplement on Engineering-Drive Ethernet:
Don't Rip Out Your Legacy Network (this article)
Industrial Ethernet applications in real time with SERCOS III, EtherCAT
-SERCOS III Ethernet protocol has replaced the traditional network for direct control of I/O connections, monitoring subsystems, and other field bus devices on Rovema packaging machines. The open, standardized SERCOS III interface gives Rovema a universal drive and automation interface compatible with drives and devices from many manufacturers.
-Popcorn bags open better with EtherCAT, Beckhoff Automation:
Trevor L. Lang is the sales and marketing manager for the North American subsidiary of HMS Industrial Networks, which offers the Anybus brand of gateways and networking solutions. For more information, visit editorial director.