Network Determinism: Not Just for Motion Anymore
Determinism is also important in a factory where land systems and specialty vehicles are manufactured. Singapore Technologies Kinetics (www.stengg.com), a $1 billion company based in the city it’s named after, produces an array of systems and vehicles that are used in defense and other high reliability environments. This equipment is produced on a production line that’s managed by a large-scale condition monitoring and management system. This networked system is connected via an eight-slot National Instruments 9144 EtherCat slave chassis that can connect up to 65,535 devices. This EtherCat controller lets the condition monitoring system monitor and control more than 150 analog and digital input/output (I/O) signals in each location.
Not just for motion
In years past, “it was typically motion-centric systems with coordinated motion or CNC-type applications that required determinism,” says Joey Stubbs, the North American Representative of the EtherCat Technology Group (www.ethercat.org). “Today, however, users are recognizing that a deterministic system can eliminate waste, result in better measurement, improved production and increase the overall efficiency of almost any size of system, from the simplest PLC system to the most complex processes.”
When production lines are being set up or revamped to increase speed, developers must be sure that all aspects of the system are improved. A fast network can’t overcome the limitations of slow equipment, and the best equipment can’t run at peak levels if commands don’t arrive on time because of a slow network.
It’s not always easy to ensure that all operations run at optimal rates, however. Avoiding bottlenecks is always a challenge in manufacturing, but it’s more of an issue as the trend to outsourcing reaches ever further. Most companies now call on outsiders to help them set up production lines. These specialists have a lot of knowledge about the latest equipment and they know how to deal with many of the issues that arise unexpectedly, but they often don’t have a thorough understanding of a facility’s operations. If they don’t work closely with experienced operators and managers, problems can ensue.
“More system integrators now come from the outside, they don’t have the intimate knowledge of the entire application that internal people have,” Wozniak says. “When they’re only working on one segment, they may not see what happens with the full process. You can end up with one part of the process running at 50 mph and a related process that’s only running at 20 mph, so you have bottlenecks.”
Whether or not bottlenecks have been eliminated, developers need to set up architectures that allow for the inevitable problem. Cables may be broken, critical nodes or servers may crash. Planners can avoid this by choosing an Ethernet network architecture that will bypass the fault. Performance may be degraded slightly, but lines won’t be shut down unless safety or quality levels are threatened.
“There are a series of architectures that ensure that if you break one wire you can continue your operations,” says Steve Zuponcic, applications engineering manager at Rockwell Automation (www.rockwellautomation.com) in Mayfield Heights, Ohio. “Redundant stars can be set up and ring topologies can be put in place.”
Good diagnostics needed
When failures do occur, good diagnostic programs will alert managers so they can quickly dispatch maintenance personnel. A wide range of commercial tools can provide solid diagnostics that will suffice in many installations. Companies that provide real-time networking technology also have connections with diagnostic tool providers that focus on industrial applications. This type of equipment is far broader than than the type linked to commercial networks.
This ability to connect many types of equipment and carry many different protocols is perhaps the biggest benefit of moving to Ethernet. New industrial users often start with limited communication, but Ethernet traffic is then rapidly expanded to carry other types of data, including safety and front office communications.
“The technology is there to handle more on the media; it can be used for multi-discipline communications,” Zuponcic says.
Click here to read more about the broadening of deterministic networks.
Ethernet also has the speed and distance to enable installers to place equipment where they need it. That’s a big improvement over many of the proprietary industrial networks of the past, which sometimes required users to locate demanding nodes close to each other so signals could arrive on time. Ethernet minimizes the need to tie location to the network. It also makes it much simpler to share all types of data.
“You shouldn’t have to put all your real time equipment in one location,” Zuponcic says. “You should also be able to transmit everything on the same wires. You can put your HMI [human machine interface] on the same network as your motion control system. That gives you optimal flexibility.”
This ability to handle different data types also makes it possible to utilize a single cable for all types of communications. Safety networks, for example, once required dedicated hardware. But even basic Ethernet links have enough bandwidth to carry multi-discipline signals, letting users eliminate single-function networks.
“This functionality used to exist only in stand-alone solutions, such as functional safety, condition monitoring, and data acquisition systems,” Stubbs says. “For the user, this [ability to combine communications] cuts costs and reduces programming effort while increasing the overall system’s diagnostics capabilities. Now these subsystems can be simply implemented as software modules in the main controller and all variables, diagnostics and states can be freely communicated, instead of having separate hardware add-ons for each subsystem.”
While moving to a deterministic network adds speed to this flexibility, observers note that adding deterministic capabilities won’t guarantee significant improvements in performance. As with other technologies, merely selecting the fastest and most expensive products doesn’t ensure that they will provide performance that’s near the peak rates touted by marketers.
For example, a deterministic network will not necessarily provide a high scan rate or offer better device control than its non-deterministic brethren. Vendors note that the deterministic segments of a network must have more processing capability than non-deterministic sections. “If a network has a lack of raw scanning speed, it is still all but impossible to close the motion control loops to even one axis using this network,” Stubbs says.
Competitors often criticize deterministic networks that require extra hardware to handle messages in a timely fashion. This moves the network away from the ideal that’s often touted for those who still use proprietary networks: Ethernet provides universal connectivity.
However, management teams that want to gain the benefits of determinism will find that they’re casting their lot with a specific architecture. The additional processing needed to handle precisely timed messages doesn’t occur without some additional hardware or software, so it’s difficult to intermix deterministic products that come from suppliers of different deterministic technologies.
“With all real time versions of Ethernet, you’re pretty much in line with a specific vendor,” Wozniak says. “If you stick with basic TCP/IP, everything can talk easily. But when you add determinism, that changes things.”
Read "Understanding Ethernet Speed and Determinism" to see how commercial Ethernet is adapting to the needs of industrial applications. Visit http://bit.ly/awfeat058
Add new comment



























