Motion Control Resources
Ethernet Pulls Motion Together
by Kristin Lewotsky, Contributing Editor
Motion Control & Motor Association Posted 10/29/2007
Ethernet shows great promise for motion control, provided the underlying transport layer is maintained. Is it the right solution for your application?
There was a time when every element of a motion control system had to be accessed separately, but no more. Just as Ethernet allows us to simultaneously browse the web, listen to Internet radio, and send e-mail, so Ethernet allows motion control components to be networked together more effectively than ever before, simplifying control, diagnostics, updates, and more.
“In the past, you saw a network for safety, a network for motion, a network for discrete control, and a network for enterprise integration. Now you’re seeing that consolidated into a single network infrastructure,” says Matheus Bulho, product marketing manager for motion control business at the Kinetix Motion Control group at Rockwell Automation (Mequon, Wisconsin). “That means fewer parts to buy, fewer parts to install, fewer parts to configure, and so forth. The life-cycle cost of an integrated system tends to be a lot lower than a fragmented type of solution.”
And it’s not just about cost. Integration makes the whole far greater than the sum of the parts. “The huge strength of Ethernet is that it’s a transport layer. It doesn’t care what the application does with the information,” says David Greensmith, product manager for Ethernet motion products at Baldor Electric Co. (Bristol, UK). “It’s just this powerful communication layer.”
The application defines the meaning of the data -- device profile specification, object dictionaries, what needs to be exchanged for the application, and so on. This flexibility of device-to-device data exchange, combined with the real-time performance and increased bandwidth of Ethernet compared to fieldbus systems, blurs the boundaries of devices. A system becomes a pool of the device resources on the network.
Consider, for example, encoder interfaces. “We quite often have drives which are capable of hosting an additional feedback interface but don’t necessarily make use of that capability,” Greensmith says. “There might be another device on the network that is physically limited, however, and doesn’t have an encoder input of its own. Now, we can make that encoder available over Ethernet Powerlink and use it locally as if it were physically connected to that device.”
The networked nature of Ethernet allows access to any device in the system from a single point. Gone are the days of needing to be physically wired in to a device to access it, which is an enormous advance, according to Greensmith, a former applications engineer. “I can remember the days when you would have to carry a large number of communications cables, one for each device you were going to work on. That was a bit of a nightmare,” he says. “And of course, you had to physically walk to wherever that unit was, sometimes crawl under machine spaces and plug in to be able to program it and configure it.”
With an Ethernet system, it’s possible to download software to all devices on the network from a single point. A host PC in the system can store the configuration files for each device on the network, which makes setting up the machine as simple as running a preprogrammed sequence to download parameter and firmware files as required.
Such capabilities provide enormous flexibility in system operations. It’s possible to set up a drive to toggle between two positions automatically, and then immediately run it on a series of torque commands. In today’s manufacturing environment, the ability to change on the fly is paramount, and Ethernet allows systems to do just that.
The Ethernet Variations
There’s an even bigger promise to Ethernet -- that of connecting the motion network to the enterprise network. Such an approach would bring new transparency to process control and supply chain management. “We have the opportunity today to leverage standard Ethernet in an integrated manufacturing environment,” says Bulho. “You [could imagine] recipes being downloaded from the enterprise system based on an order that came through, and configuring equipment at the shop floor to build that product in real time.”
That’s the idea, anyway - the reality isn’t quite so simple. The biggest concern that Ethernet posed initially was its use of TCP/IP protocol; as such, it passes information on a first come, first served basis. That’s fine when it’s used to connect a dozen office computers to a single printer and fax, but in a motion system, which generally requires high-precision synchronization, that approach can be challenging.
As a result, most versions of Ethernet for motion control are variations of the basic protocols. The problem is that the further the code departs from standard Ethernet, the more difficult the realization of enterprise-wide integration becomes. “The vast majority of enterprise systems are not designed to operate on proprietary Ethernet,” says Bulho. “Standard unmodified Ethernet delivers full compatibility with off-the-shelf technology.” The addition of quality of service (QoS) technology, full duplex operation, managed-switch technology, and the better than 200-ns synchronization accuracy provided by the IEEE 1588 standard makes the unmodified Ethernet approach compliant with the deterministic requirements of motion control, he notes.
Maintaining the Ethernet transport layer is key, Greensmith agrees. Without it, you lose the ability to perform trans-network, and even trans-world, control. “You don’t want to destroy those underlying protocol mechanisms in order to implement your motion bus, which is, sadly, what some of the motion buses do -- they break those rules, they modify the protocols and behavior so significantly that it’s no longer really a compliant Ethernet network.”
It’s a good point. On the other hand, striving for a universal solution carries challenges of its own. “Some of these networks are trying to be a single network for all capabilities,” says Greensmith, “and while that’s good -- it has some definite benefits -- it can’t really be optimized for motion control and can easily become overly complex. One of the things that we always look for in any technology is what is easiest for the user. How’s the user going to understand the technology to apply it?”
The Open Source Challenge
Currently, there are nearly 20 different flavors of Ethernet-based motion control software on the market. Several of the entrants are open source, chief among them EtherCAT, Ethernet IP, Ethernet Powerlink, and Profinet. Ultimately, the market will determine the survival of the fittest. In the meantime, as with encoder protocols, it’s a challenging market for the end user -- not least, ironically, because of the open-source nature of some of the protocols.
Open source software offers end-users the choice of multi-source component or system solutions, while allowing vendors to participate without high-cost licensing fees or entering into contracts. The result is a wider selection of devices and greater cost competition. All good, right?
Mostly. There can be a downside, it has to be acknowledged. While some open standards maintain strict certification processes, others leave a greater amount of freedom in the hands of the vendors implementing the technology. As a result, some manufacturers may choose to skip fully implementing the standard, or may resort to adding vendor-specific functions with their own parameters. When such a component is plugged into the network, the software tools provided by the host device may be unable to understand what those parameters are, how to configure them, and so forth. The promise of plug and play, of network-wide control, is suddenly gone -- in a multi-vendor network, not any one single software suite can set up all of those devices.
Rosser Pryor, president of system integrator Factory Automation Systems Inc. (Atlanta, Georgia) weighs in: “Open Ethernet is still, in our opinion, a promise that is as yet unrealized with servo applications. So far, we haven’t experienced that it’s brought ease to integration. We find that each [Ethernet protocol] has some peculiarities but actually the ones that are proprietary or semi-proprietary work the best.” The downside of a proprietary solution, of course, is that it locks users into a specific vendor, for better or worse.
Greensmith is quick to note that the standard conformance issue is not new and that the industry has learned many lessons on this front from past fieldbus communities. “It’s one of the key reasons for adopting existing, well-proven device profiles, such as those of CANopen, and ensuring conformance testing and closer collaboration between vendors.”
Ethernet for motion control is enabled in large part by the economies of scale driven by Ethernet chips for the computer industry. The question is what happens when the computer industry moves on from Ethernet to Gigabit Ethernet and even to 10 Gigabit Ethernet? Will motion control still be able to leverage such increased speed? Or will vendors find themselves faced with a choice between paying for capabilities they won’t use or paying a premium for slower chips whose cost basis is now driven by the motion control industry’s demand for millions of units versus hundreds of millions?
Nipun Mathur, motion control and mechatronics product manager at National Instruments (Austin, Texas) doesn’t see the speed jump as a cause for concern. “That’s a really good problem to have,” he says. “There are lots of applications that need high speed synchronization between axes. People are going more and more toward I/O on the same bus as motion, so as you add more bandwidth to the motion bus, you can actually do high speed streaming analog I/O with motion control.” He points to the increasing use of diagnostic information gathering and transmission to yield smart machines capable of sophisticated safety and fault-monitoring operations.
As to the future? “It’s whole new world,” says Greensmith. “What we’ve had in the last three years is this evolution of the underlying technology that is Ethernet for motion control purposes. What’s happening now is it’s creating this platform for evolving from an application point of view. What people can do with that product is a step change in the capabilities. I can now sit at my desk and, via VPN connection to a machine in Australia, have a look at what’s going on through that device -- and that’s all through the underlying power of traditional Ethernet.”