A topology, which is the pattern of interconnections among nodes, influences a network's cost and performance. While commercial networks move toward low cost solutions, industrial networks must concentrate on rugged, heavy-duty solutions for multiple environments.

This is achieved by using stronger signal integrity and heavier shields in the physical media. A topology determines what network characteristics are possible. What goes on top of that foundation, such as the physical media and transceivers, will determine network parameters.

Most common is the bus topology. Bus adaptations are called linear bus, bus with extensive branching, and bus tree. These adaptations came about with specified electrical properties that allow longer drops and drops within drops. With all bus topologies, communications are conducted on common conductors where the receiver and transmitter are connected to the same communication wires as all other network nodes. This allows the transmission from one node to be received by all others.

Parallel network wiring is a great advantage. In many networks the nodes can be hot swapped in a bus configuration, an essential feature to some industrial process applications. Other advantages include low cost and easy installation. However, bus segment lengths are limited, and a network will fail if the main communication wires are cut. Some networks use repeaters to increase network length, and redundant communication cables ensure that a broken wire will not bring the network down. Examples of networks that use bus topologies include DeviceNet, Profibus, ControlNet, AS-i, and FOUNDATION Fieldbus.

The second dominant form is ring topology, in which each node acts as a repeater. Transmission starts at a central station, usually the controller, and is sent to one node. That node receives the transmission, processes the information if needed, and sends it to the next node on the ring. Long networks are possible because each node reconditions the transmission; throughput time around the ring is predictable. When the ring breaks, communication is lost; hot swapping isn't possible. Interbus, Lightbus, and SERCOS are among networks using ring topology.

Bus and ring topologies are dominant in smaller fieldbus arenas. When a factory couples topologies together through gateways, routers, and bridges, the combination is called a tree topology. Large factories interested in keeping statistical data on processes will use a tree topology to gather information from different fieldbus controllers.

The Internet is the most common tree topology. Ways of using the Internet with control or supervisory systems are constantly being developed. Tree topologies interface with any topology as long a gateway exists, but that's also a disadvantage in many control systems. In gaining connectivity, the ability to determine the time it takes to receive information from a node is lost; generally throughput time is not a constant with a tree topology.

Star and mesh topologies, because of higher costs, are not used widely. Star topology has one central node with a direct connection to each individual node. Because the protocol is easy to develop, many private networks use it. The mesh topology connects each node with every other node, creating an isolated data path between each node.

The first question when selecting a network should be: "Will the pattern of interconnection among nodes allow the network to physically achieve what's needed? Manufacturers of the network components can confirm network parameters.

Send comments to mhoske@cahners.com.