The task seemed daunting, particularly given the short timeframe, but it got done without a hitch. As Jeff Smith, technology lead at AAM in Detroit, detailed recently, his team launched four assembly systems (one with 28 stations and 800+ EtherNet/IP edge devices) over a four-month period flawlessly.
The switch to Ethernet does not have to be the daunting task that it once was. Though Ethernet was designed for a star topology — in which all devices connect back to a central hub — that is starting to change, making Ethernet networks more flexible and easier to deploy. In AAM's case, it was an EtherNet/IP network set up with controllers from Rockwell Automation, but the same is true for other Ethernet standards.
The key enabler for increased topology flexibility is the introduction of devices with integrated hubs or switches. Two connection ports in the device enable daisy-chaining into a line or ring configuration, as well as other setups.
Daisy-chaining into a line reduces overhead and complexity by running just one wire from the switch, then into node 1 and out, into node 2 and out, and so on for as many as 15 devices in the case of controllers from Wago. "With a star, you would have to bring all 15 nodes back to the switch," notes Tracy Lenz, senior product support for Wago's I/O group.
EtherCAT, the flavor of Ethernet from Beckhoff Automation, eliminates the switch altogether, which eliminates significant delay and jitter, says Joey Stubbs, North American representative of the EtherCAT Technology Group (ETG). Flexibility was built in from the beginning, so topology is not an issue.
Which kind of topology you want to use will depend on the type of machine or system being controlled, Stubbs explains. "If the thing you're trying to control is a conveyor line in a big sorting facility with miles and miles of conveyor lines, with EtherCAT you could easily just connect those together with a very long line topology out of the box," he says. "If you have a machine that's a cluster of devices, maybe a star works better for you."
Robert Muehlfellner, director of automation at B&R Industrial Automation, argues that a tree topology — a combination of line and star topologies — often provides the most flexibility, allowing for multiple branches from the root, or machine controller. "It branches out from there, and wherever it makes sense to have a junction point, you just place another hub," he says.
"A strict line configuration doesn't always work out that well," he adds. "Ideally, a network should be flexible enough to not make any compromises, to adapt to whatever needs you have. And it should do that without additional complexity."
Network topology flexibility becomes more important with increasing use of modularity and machine-mount I/O, Muehlfellner explains. "With the trend that we see to have many more machine-mount devices, the network architecture is becoming more and more important," he says.
A configuration designed to recover quickly when a single fault occurs on a network is ring topology. Although it requires more effort to configure, it can keep the network running seamlessly by providing an alternate path to route communications data to all the devices.
That ring capability can even be used for scheduled maintenance. "If you have an assembly in the middle of a redundant ring that needs to be taken out for maintenance, you can continue to run everything else after you take it out," Stubbs says.
"The downside to any ring topology is that you need to close the ring," Muehlfellner notes. "This kind of limits your flexibility in the network architecture."
Where ring topology makes most sense is in a slip ring application, involving a rotating turret of some kind, Muehlfellner adds. "The chance of having some intermittence in your network is a little higher."
But B&R doesn't use ring topology often because it's not well suited to modular machine designs. "You may want to attach or detach modules on the machine," Muehlfellner says. "A line or tree architecture is much more flexible than a ring."
