GET CAN IN GEAR Serial networks facilitate higher-level control, but some key issues are yet to be resolved. This article is reprinted with permission by Penton Publiching Inc. The article's text is reprinted as it appeared in the 09/12/96 issue of Machine Design. © Copyright 1996, 1997, Penton Publishing Inc. KENNETH J. KORANE Senior Editor To help their customers meet demands for higher throughput, ensure operator safety, and keep up with proliferating environmental regulations, manufacturers of off-highway equipment are taking a serious look at Controller Area Network (CAN) serial communications networks. Electronic control systems based on CAN provide fast closed-loop, real-time performance with excellent reliability. Developed by Bosch and Intel for the automotive industry, standard CAN microchips are widely available, inexpensive, and designed to withstand the harsh electrical and physical environment of a car. These attributes make CAN networks attractive for industrial machines. For instance, since the 1980s it has been used on the factory floor for numerous process and motion controllers, and discrete manufacturing production lines. Interest is now migrating to mobile equipment.   CAN BASICS CAN is a standard communications protocol (ISO 11898) that connects actuators and "intelligent" sensors in a serial network, coordinated by a host computer. On the machine, the information network interconnects every device - each with its own microcontroller - establishing a common communication path between them and the host. The network architecture has advantages compared with a standard design where every component is connected directly to a central processor. The primary benefit is serial networks have fewer wires. This reduces failures due to poor connections and broken wires, cuts the cost of wires and connectors, and can substantially reduce installation expenses. CAN networks also use digital technology that takes advantage of computers and microcontrollers. A digital controller's superior ability to store data and perform calculations makes it practical to add higher-level capabilities, such as automating machine tasks or optimizing fuel efficiency. CAN is also fast - 1 Mbyte/sec for a 40-m bus length. It also contains robust error-detection software and immediately retransmits if a message is corrupted. Finally, Intel, Motorola, NEC, Philips, and Siemens, among others, offer commercially available, low-cost microprocessors with built-in CAN controllers.   ENHANCING PERFORMANCE The use of CAN on mobile equipment is still in its infancy. One reason is simply that many designers are still learning how to use it. "CAN represents a new technology requiring a lot of rethinking," says Lars-Berno Fredriksson, president of Kvaser AB, Kinnahult, Sweden. CAN opens up options previously out of the systems designer's reach, he says. "But, unfortunately, you do not recognize the possibilities until you have gained some experience in the technology." As the basic understanding of CAN grows and the first CAN design tools appearing on the market, Fredriksson predicts a sharp growth in CAN systems over the next decade. According to Andrew Martin, product manager for mobile controls at Vickers, Troy, Mich., major equipment manufacturers have a number of development programs underway but few systems are in the field. Nonetheless, the company is marketing a new control valve that can be driven with CAN bus technology. Part of the reason, says Martin, is that future mobile controls will be very similar to those on automobiles, where devices located throughout the car send information to a central processor. "We see a similar direction for the mobile-vehicle control architecture, such that everything on the machine - fuel-injection control, steering, braking - will all be controlled from a central processor. Our valve fits in with that control philosophy," says Martin. The valve amplifier accepts CAN based signals from the controller (or it can be driven directly from a joystick) and sends milliampere control signals to the valve. The valve returns diagnostic signals to the amplifier. One benefit of this system, says Martin, is the digital amplifier has no potentiometers. "It's all set up from a PC, that is the deadband, ramps, and gain, and the PC can test the valve," he says. After the amplifier is set up for one machine, downloading the program to other amplifiers on a production line ensures they perform the same without the need for manual adjustment. Martin also adds that the high cost of CAN-compatible sensors - which can be 10 times that of conventional units - is limiting the full capabilities of CAN-based systems on mobile machines. For instance, he indicates the new EMV-611 valve could have more diagnostic capabilities if the sensors were less expensive. While a number of manufacturers are currently developing the low-level sensor chips, the question is how quickly they will become cost effective, he says. Parker Hannifin is another proponent of CAN-based systems, having recently introduced the MMC 1000 controller. "It's a quantum leap up, applying this type of technology to mobile equipment," says Anup Shetty, marketing and strategic product development manager at the company's Hydraulic Valve Div. in Elyria, Ohio. Much of the mobile machinery used today has no electronic controls at all, he says, instead they rely strictly on mechanical systems. "Some have PLCs and other control devices, but they're not very sophisticated. We developed the controller because we saw CAN as up-and-coming technology," he says. With the automotive industry and major players such as Allen-Bradley and Honeywell adapting CAN, inexpensive, robust chips suited for mobile equipment are now available, says Shetty. He indicates their digital-based controller offers some significant benefits. One, the controller, valve modules, I/O modules, and transducers all communicate via the CAN bus. "So, instead of running up to eight wires from every valve back to the central processor, a single cable connects each," says Shetty. CAN SHARPENS TIMBER CUTTING CONTROLS The Timberjack 3000 is a measuring and control system for single-grip harvesters. It controls option of the harvester head, calculates optimized cut points based the size and shape of a tree, monitors production, and communicates with external computers. Forest machines have used computer-based control systems for more than 10 years, but traditional systems feature a central computer inside the cabin and extensive cabling to sensors and actuators. The Timberjack 3000 is a distributed system based on the CAN bus. According to system developer CC Systems, Alfta, Sweden, this reduces cabling, enhances optimal quality from remote sensors, improves fault finding and diagnostics, and is easily upgraded. The components include a host computer that contains the system database, performs optimization calculations, and manages data communication. The harvester head module controls 19 valves on the head and measures length, diameter, and saw position. It translates high-level commands into specific valve movements enabling highly automated machine operation. The keypad control module translates pushbutton and lever signals into CAN messages, and the system also includes a LCD display. The controller also brings added intelligence to the machine because the CAN-bus system simplifies getting data from remote locations. "Obviously, you can get information from a sensor whether you're on a bus or not," he says, "but now the valves can report back, tell you where they are and what they're doing, and the I/O modules can provide higher level information. So we get diagnostic capabilities that let you know if a valve malfunctions, a benefit that you don't have with a traditional control scheme." Because the MMC 1000 is a machine-level controller, it also takes in information from other sensors on the machine, reading engine rpm, oil temperature and pressure, and the like. Last and most significantly, says Shetty, machine functions can be automated. One example involves monitoring demand on a trencher. As the draw on horsepower from trenching increases, the controller slows the vehicle, maintaining a constant speed on the trenching wheel and preventing system overload. This produces a consistent depth and straightline cut, rather than allowing the operator to bog down the machine and have to back up and restart. It also reduces hydraulic pressure spikes, so components last longer. Automating tasks now lets less-skilled personnel do what once required highly trained operators, while improving the speed and productivity of the machine. The higher-level controller can also improve safety, such as ensuring that a manlift cannot operate outside a predetermined, safe envelope. In terms of costs, Parker's experience proves that on complex machines such as rock drills and large manlifts, the savings on plumbing and wiring offset the cost of implementation. In smaller, less-complicated mobile equipment, added functionality typically increases costs. "But in today's marketplace, users want and need features that improve the performance of their machines, so they're willing to pay the price," says Shetty.   PROTOCOL DEBATE Part of the reason few CAN systems for mobile equipment are on the market is that there is no standard, higher-level protocol. CAN takes care of the low levels of a communication protocol but a network also requires higher-layer protocols, such as DeviceNet or SDS (Smart Distributed System) used on industrial systems. For a designer to assemble a system from various manufacturers' off-the-shelf components, a fully open higher-level protocol is needed. Such a standard, universally accepted protocol for mobile equipment appears to be a ways off. In the meantime, options abound. For example, researchers at Vickers found 31 different CAN protocols currently in use, with none emerging as the front-runner for a global standard. Their solution, says Andrew Martin, was to develop hardware that supports virtually any protocol, and program it based on customer requirements. "We're using SAE 11939 because the first customer that came along asked for it," says Martin, adding that this standard appears to be the best, at the moment, for North America. Sauer-Sundstrand, which currently supplies CAN-based systems for paving machinery, takes another approach, according to Dick Heiser, an engineering manager at the company's Electrohydraulics Div. in Minneapolis. l They use their own proprietary protocol, called S-Net. The reason, says Heiser, is that scant information on protocols was available when they began development work, forcing them to develop their own. The lack of a standard protocol is a substantial issue, says Heiser. "Once you start getting locked in, it's pretty i hard to change," although he predicts they will likely adapt to meet their customers needs. Some manufacturers, he says, are using proprietary protocols for their subsystems and then building a bridge to the rest of the machine. "That's not an ideal situation either, but it's happening because standards aren't keeping pace," says Heiser. He adds that the leading standards effort is probably taking place under the auspices of the Equipment Manufacturers Assn. (EMI), which is working on an enhanced version of 11939 suited for off-road vehicles. EMI is a trade association of agricultural, construction, and off-road equipment manufacturers, and is the administrator of international activities in the agricultural arena. While not a standards-writing organization, EMI participants serve on SAE committees. According to spokesman Darrin Drollinger, EMI members are rapidly embracing the J1939 concept, although some issues involving physical connections, testing, and chip-level technology are yet to be resolved. However, he expects a version of J1939 will soon emerge for agricultural and construction equipment. SAE J1939 for Bus & Trucks is a communications network designed to support real-time, closed-loop control between electronic-control devices throughout a vehicle. J1939 uses the CAN protocol which permits any device to transmit a message on the network when the bus is idle. Every message includes an identifier that defines the message priority, the device that sent it, and the data it contains. In J1939, addresses for the units are defined in advance. Critics of the protocol note that this can limit the freedom of system designers. For example, predefinition of addresses can cause problems if a change in priority on the bus is necessary. CANHUG EMBRACES NETWORKS CAN Hydraulic Users Group is a society of CAN users involved in the hydraulics arena. The group has two goals. First, to form guidelines so different modules from different producers can be connected into a system, and that the guidelines will be accepted as a de facto industry standard. The second goal is to grow large enough to form a sizable for standard CAN-hydraulic products, initially for mobile use. Working areas include physical layer, higher layer protocol, system architecture, and safety issues. Members include module producers such as valve and joystick manufacturers, equipment producers, including manufacturers of trucks, rock carving and drilling machines, excavators, and forklifts, and consultants. Many of the 35 current members, from seven countries including the U.S., have in-depth experience in CAN, and some have CAN products on the market. The association is open to companies and individuals working with industrial and mobile hydraulic products, including CAN systems or modules. The group holds three meetings per year, the scheduled for October 17-18 in Schopp, Germany, and nonmembers can participate. For more information, contact: CAN Hydraulic Users Group Kvaser AB Box 4076 S-511 04 Kinnahult Sweden fax: +46 320 15284 Internet: www.kvaser.se e-mail: lbf@kvaser.se Another candidate for a mobile-hydraulics protocol is CAN Kingdom, supported by the CAN Hydraulic Users Group (CANHUG). CAN Kingdom takes a different approach than J1939. The software of each module contains protocol primitives that a system designer can use to build a final protocol. The advantage of CAN Kingdom not being a protocol but a set of protocol primitives, says Lars-Berno Fredriksson, president of Kvaser, is that especially for real-time systems, the system designer can choose the topology and bus-access management best suited for the application. Although CAN is specified as the master protocol, the final protocol can be of another type, for example a logical slotted ring or a token bus, depending on the system designer. CAN Kingdom also allows DeviceNet, SDS, and J1939 nodes on the same network. According to Fredriksson, there is an important difference between the principles behind J1939 and CAN Kingdom. With the latter, a network is tailor-made to the needs of the machine system. With J1939 (as with other conventional protocols such as DeviceNet and SDS), the qualities of the network are given and the machine must be designed according to the requirements of the network. Until the dust settles and a single standard emerges, the variety of available protocols may be a hurdle to systems development as manufacturers follow different courses. "Proprietary CAN systems with proprietary modules are relatively simple to design," says Fredriksson. "Designing systems with modules from different vendors can be difficult, especially if high-speed data handling and fast, error-free machine response is required." As appeared on pages 50-53 in the September 12, 1996 issue of Machine Design © Copyright 1996, 1997, Penton Publishing Inc.