The dual advantages of treating every data acquisition node as an IP address or web page along with the extended reach and data rate of Ethernet are making it a good communication platform for Industrial customers.
The first requirement for Industrial Ethernet is reliable and predictable delivery of data. Here we look at some of the OSI Layer 1 (Physical) and Layer 2 (Data Link) issues and solutions encountered in an Industrial Ethernet deployment. In addition to Layer 1 and Layer 2 solutions, there are also network standards such as Modbus TCP/IP, Ethernet I/P and Profibus that address predictable data delivery using Layers 3 and up in the OSI network model. Ethernet can be considered a deployment means for most of these standards, with the end application determining the best protocol.
When Ethernet is introduced into the Industrial setting, an existing Ethernet switch or hub is often used to distribute Ethernet in a star configuration from the multi-port switch to many destinations. Customers sometimes need more nodes than are available on a multi-port switch or have a need to locate distributed data-acquisition nodes at distances greater than the 100m from the switch. Ethernet over copper (10BASE-T, 100BASE-TX) specifies a signaling distance of 100m. 100BASE-FX over optical fiber is sometimes used to address greater distances and can typically extend reach to 2Km.
Figure 1 shows an installation using a traditional star configuration versus the addition of a daisy-chained network of Ethernet nodes. Use of copper networks to extend reach by daisy chaining nodes together outweighs the cost associated with installing a fibre based system.
When Ethernet is daisy-chained, there are two problems that need to be addressed. The first problem is delay of data through multiple switches. For Industrial applications, delay and deterministic delivery of a packet can be augmented with higher-level protocols such as IEEE 1588. IEEE 1588 uses UDP (User Datagram Protocol) packets over IP on the Ethernet network. Sensing nodes typically use the IEEE 1588 clock in one of two ways. One, to generate a timestamp at the moment data is acquired. Two, the generation of an acquisition trigger by comparing the time of the IEEE 1588 clock to a specified trigger time. More information can be found at (http://ieee1588.nist.gov).
Overall, 100Mbps Ethernet provides adequate throughput to deliver messages on a timely basis, so delivery can be theoretically guaranteed.
The second problem that arises in daisy-chained Ethernet are the loops that can be introduced during field installation. Loops or multiple paths can cause network failure due to duplicate packets resulting from redundant message paths. Figure 2 gives an example of a network with a loop.
Loops or redundant paths are not allowed in traditional Ethernet networks. However, if the redundant paths are present, most of today's intelligent switches use a protocol known as Spanning Tree to address redundant paths and assist in directing data in the proper direction. If Spanning Tree is implemented on each of the daisy-chained nodes, a very robust and redundant Ethernet network takes shape by allowing for and even capitalising on path redundancy. This enables Industrial networks to have multiple available paths with only one path active at any time. Spanning Tree protocol blocks a selected redundant path, but is able to quickly change paths if a cable is disconnected or cut; overall, Spanning Tree at the node level allows for reconfiguration of the path between the nodes.
An implementation of daisy-chained Ethernet nodes can be realised using Micrel's KS8993M as shown in Figure 3. This device consists of three MACs (Media Access Controllers) and two PHYs (Physical Layer Transceivers) along with a fully non-blocking switch in a single 128-PQFP package. The part is connected to a host controller via the third Ethernet MAC as shown in Node 3 of Figure 3. When used in this way an Ethernet packet intended for the third port is dropped to the processor, while the remaining data is forwarded to its ultimate destination. Spanning Tree protocols would be performed by the host controller and are read from the KS8993M MIB counters to determine optimal data flow.
Micrel's 3 and 5 channel Industrial Ethernet switches also support features such as QoS (Quality of Service) and VLAN (Virtual LAN) in addition to Spanning Tree. QoS allows for classes of service and controls egress queuing and priority of packet departure. QoS is useful in prioritising the class and flow of critical data. VLAN allows traffic to be segmented while coexisting on a single cable. For example, in Figure 3 Nodes 1 and 4 could belong to VLAN No1, while Nodes 2 and 3 could belong to VLAN No2. This reduces the processing burden of the local controller on Node 3 by passing through data not intended for VLAN No2. The use of VLAN's facilitates the grouping of node types ie alarms, timing, motion, temperature, etc.
The KS8993M also supports a feature known as Auto-MDI/MDI-X that allows the use of straight through CAT-5 cables or crossover cables. The Auto-MDI/MDI-X feature eliminates the possibility of improper installations that can occur when field installers improperly connect a crossover cable where a straight through cable is required. Auto-MDI/MDI-X allows routers, switches or laptops to easily connect to the industrial Ethernet node without concern for use of crossover or straight through cables.