All aboard! You might want to catch this train. Ethernet has rolled over the world, and now it's arriving at the last few pockets of resistance, or at least appearing on the horizon.
It's no longer a question of whether Ethernet can handle increasingly higher-speed motion; it's just a question of when and in what applications. But even though you can travel blazingly fast via some Ethernet-based networks, your data packets also have to stay on the rails and stop at the right stations to get where they're going. In machines, trains and on the plant floor, speed is nothing without accuracy. Ease of use helps a lot, too.
Jim Montague is the executive editor for Control. Email him at [email protected].
"We've been using EtherNet/IP for motion in our palletizers for six or eight years," says Kevin Davis, electrical design manager at Production Automation (PAI) in Montgomery, Ala. "Today, even in our high-speed areas running eight-axis with servos at 80 cases per minute with a half-second cycle time, it's still EtherNet/IP. We just make sure to isolate our high-speed motion from other Ethernet traffic by using an Ethernet module with a ring topology."
Joey Stubbs, North American representative of the EtherCAT Technology Group (ETG), adds, "The days of dedicated networks for motion, I/O and other tasks are effectively over, since all of these tasks can be serviced by one network simultaneously from one controller. Traditional motion protocols don't offer anywhere near the performance, diagnostics, ease of use, configuration and cabling of well-implemented, Ethernet-based protocols. This isn't just due to performance. It's also because machine-level communications now take advantage of consumer-based technologies such as Cat. 5/6 cabling, connectors, transceivers, standard NICs as masters, standard diagnostic tools, etc. This is instead of having to use special, costly hardware required by legacy protocols. This piggybacking of physical-layer components drives down the cost of systems and increases product selection."
PAI's palletizers have to quickly but gently stack and wrap cases of super-thin plastic bottles, and so they use robot arms and cranes on EtherNet/IP to replace traditional physical diverters.
So, how can machine builders choose the right type of Ethernet and related networking components to satisfy end users who need increased speed and throughput, better handling, increased flexibility and more open networking? Follow a good role model. There seem to be just as many Ethernet-based solutions as users have problems. You just have to get on the right train — and avoid the brawl in the club car.
Slice It or Stamp It?
Basic, vanilla Ethernet wiring and its pure TCP/IP came from IT and office realms, where data was blasted to all parts of the network and speed wasn't as crucial or possibly dangerous as on the plant floor. Ethernet's more recent industrial protocols adopted increasingly intelligent switches and addressing to achieve determinism, and then adjusted their communication methods and rules for increased speed by prioritizing and synchronizing how they transfer data. In general, tightly dedicated Ethernet networking permits greater speed, but means less flexibility. Meanwhile, less-restricted Ethernet enables more flexible communications, but typically runs slower.
Historically, there were two main schools of thought about manipulating Ethernet to serve in higher-speed motion: time slicing and time stamping. To prevent message collisions, boost speed and achieve determinism, time slicing ensures that critical devices and functions are assigned specified time slots to transmit information. A managing node on the network handles time allocation, so the others can transmit without interference when it's their turn. Time slicing is used mostly by EtherCAT, Powerlink, SERCOS III and Profinet Isochronous Real Time (IRT).
The second method, time stamping, is based on the IEEE 1588 standard's Precision Time Protocol (PTP), which defines a method for sub-microsecond synchronization of the clocks in sensors, actuators and other terminal devices on a standard Ethernet-based network or other distributed application. It's used primarily by EtherNet/IP with CIP Sync and CIP Motion. IEEE 1588's basic function is to have the most precise clock on a network synchronize all the others, and then time stamp each data packet moving on the network.
Not surprisingly, the different methods of altering Ethernet for higher-speed motion have led to some arguments. The main debate seems to be between the time slicers and the time stampers. The slicers say that time-stamped Ethernet is basically an Internet Protocol (IP) that's not dedicated enough to truly handle high-speed motion, and that it must run slower because its master device is always on and all data must run through it. However, the stampers counter that time-slicing might be quicker, but it risks losing data because its nodes can't run and communicate when the master device is talking.
Speedier Checks
Bickering aside, to give those food processors all the precise weighing data they need throughout their production lines, Friesen's of Detroit Lakes, Minn., designed and developed its Mach-Series and F-Series machines to serve in earlier food manufacturing steps and adapt to a wide range of user requirements.
Mach-Series can run at more than 100 parts per minute, and has two high-speed, precision checkweighers and a high-speed, in-motion checkweigher. F-Series machines run at less than 100 ppm, and have in-motion checkweighers. Its washdown-capable systems can be deployed throughout food production lines, and its IP69K-rated systems are ideal for USDA-regulated plants (Figure 1).
"In the past, individual adaptation and scaling of the controls could be tricky tasks," says Kari McAllister, Friesen's product development director. "In my view, some of the major controls vendors in this space are too proprietary in nature, and fail to provide interfaces for third-party systems."
To eliminate these roadblocks, Friesen's adopted Beckhoff Automation's embedded PC with TwinCAT software and EtherCAT I/O terminals and servo drives to automate, accelerate and simplify its checkweighers. "We sought to process product weights faster and communicate to our reject systems at higher speeds," McAllister says. "EtherCAT can achieve update times for data from 1,000 distributed I/Os in only 30 µs, including terminal cycle time. Up to 1,486 bytes of process data can be exchanged with one Ethernet frame, which is equivalent to almost 12,000 digital inputs and outputs. The transfer of this quantity of data only takes 300 µs."
As a result, EtherCAT and open PC-based controls allowed Friesen's to decrease machine build time by 25%, cut installation time in half, reduce startup time by 30%, and reduce downtime by 22%. "Also, since Friesen's began using EtherCAT and embedded PCs, checkweigher system speed increased by 31%, which means we give our users much higher throughput and really differentiate ourselves, too," McAllister adds.
These gains are no surprise to ETG's Stubbs. "The highest-performance Ethernet-based protocols meet or exceed the capabilities of even the most demanding real-time motion control tasks," he states. "The best ones have leftover bandwidth to integrate other functionality, which previously had to be implemented in separate systems, such as ‘non-motion' I/O, functional safety, data acquisition, and servicing of standard IP devices on the network such as for web thin clients. EtherCAT is capable of scanning 100 servo axes in 100 µs in one network, which sets it at the highest end of the performance metrics, while undercutting costs, eliminating the need for special cables or special infrastructure components, and allowing users to integrate functions over the same network, such as condition monitoring, functional safety and supporting other protocols."
Figure 2: Brückner's linear motor-driven simultaneous stretching system (LISIM) machine pulls and stretches plastic film at 6.6 m/s between two opposing, ring-shaped rail lines, all coordinated by Powerlink, which reduced cycle time to 400 µs.
Source: Brückner
Synch Lots of Axes
Besides enabling traditional speed, Ethernet-based protocols also need to accelerate the coordination required for multiplying numbers of axes, drives and other components. For instance, Brückner Maschinenbau in Siegsdorf, Germany, uses more than 700 clips to pull, stretch and heat set plastic film at 6.6 m/s through its simultaneous stretching system (LISIM) on two opposing, ring-shaped rail lines (Figure 2). The clips are pulled by a moving magnetic field, similar to the railway cars on a magnetic levitation train, which is generated by linear motors with 728 windings.
Current for these 728 zones along the 65 m long LISIM machine is supplied by 384 single- and dual-axis inverter modules in conjunction with 14 power supply modules and drives from B&R Industrial Automation, and they're all synchronized via Powerlink.
"LISIM's machine concept is based on linear motors, and we invented it more than 15 years ago, but it's experiencing a renaissance now because of increasing demand from the packaging and flat panel display industry for film with special properties," says Günter Oedl, Brückner's electrical engineering manager for automation and development. "However, the drive technology that we used up to now was getting on in years, and a new proprietary solution would have been difficult and costly."
Oedl adds, "This solution was possible because Powerlink allows precise synchronization of hundreds of network nodes, and simultaneously provides high data throughput. On one hand, we were able to reduce the cycle time significantly — it's now only 400 µs. On the other hand, we were able to move large chunks of software from the drives to a central drive controller. This also simplified servicing and software maintenance." B&R maintains that controlling 728 axes with a 400 µs update set a world record.
All 398 of the Acopos modules, both power supplies and inverters, in LISIM are synchronized by 12 of B&R's industrial computers. Each is equipped with three Powerlink cards, which control up to 13 modules. Using another Powerlink card, the computers communicate with each other or with a higher-level industrial computer that runs Brückner's motion control software. The plant control system, which is responsible for controlling LISIM's oven, is connected to this industrial PC via a Profibus interface.
Finally, Powerlink's short cycle times and minimized jitter allowed Brückner to position the zones along LISIM very closely. "The individual zones are grouped in a very homogeneous manner," Oedl says. "So the error tolerance is significantly less than a millisecond, which was stipulated by the application."
Robert Muehlfellner, B&R's automation technology director, adds, "The three fastest Ethernet-based protocols are Powerlink, EtherCAT and SERCOS III, and they're all fast and reliable enough for high-speed motion. They all can achieve fast update times between sub-500 µs and sub-1 ms, and they have high synchronization and minimal jitter. There are some technical differences, but their end results are the same, and so they're all good for 95% of machine applications."
Closer to Machining?
Aside from axes, one type of motion that Ethernet hasn't taken on yet is in the CNCs used in machining centers and related tools. Though Ethernet isn't practical for the critical tasks of high-speed machine tools, such as controlling spindles at thousands of rpms, it can help coordinate axes, load and unload parts, manage material handling, and perform other support tasks — essentially surrounding the core machining functions.
"Fieldbuses and Ethernet aren't fast or secure enough for CNC motion, and so we use our dedicated, fiberoptic Fanuc Servo Serial Bus (FSSB) communication line, which runs at 150 MHz, to control servo and linear motors," says Paul Webster, engineering manager at Fanuc. "Machine tool builders can use Profinet or EtherNet/IP to reach PLCs and control auxiliary devices, such as robotic loading or conveyors, or to reach the enterprise, but not for CNC."
However, though most CNC functions in machine tools remain as separate islands, Webster says they still added Ethernet ports and fieldbus links over the past 10 years; they download programs with Fanuc's open CNC automatic programming interfaces (APIs); and they use the XML-based MTConnect standard to communicate with PLCs, monitor machine performance, and even download G-code files that the CNC machines use to perform projects.
Figure 3: Winema's RV 10 Flexmaster CNC-controlled rotary indexing machine for small, 2–23 mm parts uses SERCOS III and a CNC with decentralized intelligent electric and hydraulic drives.
Source: Winema
Similarly, Winema Maschinenbau in Grosselfingen, Germany, introduced a CNC-controlled rotary indexing machine in a market usually dominated by cam-controlled equipment. Designed to handle small, 2–23 mm diameter workpieces, RV 10 Flexmaster employs an IndraMotion CNC-based system with decentralized intelligent electric and hydraulic drives from Bosch Rexroth, which control the machine's 54 CNC axes, including 27 IndraDrive spindle drives. It basically consolidates CNC and PLC tasks into one IEC 61131-3 compliant module (Figure 3).
RV 10 Flexmaster's core consists of a vertical indexing table with 10 clamping stations. In each cycle, the motor-powered table rotates the workpieces to the next station. The machine's rotary indexing principle allows it to process nine workpieces in parallel in each cycle and achieve high throughput.
RV 10 Flexmaster uses SERCOS III, which provides profiles for smoothly integrating both drive technologies, and helps accommodate frequent changeovers. "Previously, cam-controlled machines were unbeatably fast for large-scale series production," says Eckhard Neth, Winema's managing director. "However, we've been able to top this speed, and combine it with the advantages of CNC technology for increased flexibility. And combining the HNC with SERCOS III and IndraMotion has allowed us to improve our output by 20% compared to the previously available solutions."
Warp Speed WrappingFigure 4: Production Automation's Gantry Hybrid Palletizer uses integrated motion via EtherNet/IP to control a dual-head gantry crane and two robot arms with eight-axis servos to quickly, but gently, grab, position and build layers of cases of water bottles for palletizing in a 0.5 s cycle time at an overall rate of 80 cases per minute.
Source: Production Automation
Fast and Familiar Programming
To wrangle those increasingly thin and delicate plastic water bottles and build layers from widely varying case sizes, PAI developed its Hybrid Gantry Palletizer in 2009, and began building it last year, Davis says. Unlike a regular robot arm, this machine and its robots and crane can be thought of as a "Cartesian palletizer" that moves in the usual x, y and z planes, he says (Figure 4).
Besides maintaining high speed and providing softer handling, PAI found that using the eight servos via EtherNet/IP to move, manipulate and rotate the cases also eliminated a lot of mechanical assemblies such as diverters, case turners and conveyor rollers.
"However, the problem was that the initial programming for this new palletizer was more like using G-code and a teach pendant, which was hard for our customers to learn because they're more familiar with the ladder logic that palletizers usually employ," Davis says. "So we looked for a better way, and decided to use RSLogix 5000 software because we could program it so our users could understand it better and work in a more familiar environment. For example, customers say they can use logic modules in the software's DMAT toolkit to bring in tags for the servos, and easily set up and commission them without having to go to a second software program. Also, the servos' communications are isolated by the Ethernet module, which plugs into the Control Logix PAC, and creates a private network for that EtherNet/IP ring. We also knew this isolation was important because, in the past, we had a problem with a more open network and some users and devices trying to use the same IP address and being unable to communicate."