PC advances drive control technology

The driving force behind motion control for the past 15 years, and into the foreseeable future, stems from the rapid advancement of the computer's speed, power, and cost. The good news for all of us involved with industrial control is that the computing revolution is far from over. Computers continue to move ahead in accordance with Moore's law: Their speed and power will still double every 18 months.

As its capabilities grow, the PC continues to accelerate its expansion into the industrial automation marketplace. The result: It will coordinate a growing number of control tasks by the year 2000.

But questions remain. What are we going to do with all that computing power? How will engineers overcome the obstacle holding back computer-controlled, servo-driven machinery--the enormous amount of software required to advance state-of-the-art automation systems? In short, how can we most effectively use this technology to make our companies more productive and profitable?

Electronics take the stage. To answer those questions, it may be helpful to look back a bit. In the late 1970s, microprocessor technology achieved enough power to be useful in motion-control applications. Microprocessor-controlled servomotors provided many advantages compared with other motion-control alternatives: excellent torque/speed delivery, positioning accuracy, and the ability to limit torque. In a fully microprocessor-based, servo-control system, position, torque, and velocity are regularly monitored and, in some systems, they can be read at any time by the motion controller or a supervisory computer.

Design engineers increasingly used such systems to achieve greater flexibility, operating speeds, accuracies, and reliability than ever before. In factory-automation applications, servo-based electronic motion control steadily gained ground over more traditional mechanical means of power transmission, as well as "lower-tech" drive-system alternatives, such as stepping motors and ac and dc motors.

The rapid emergence of this technology resulted in a complex and often confusing array of choices for design engineers. Some suppliers chose to focus on low-cost products for use in high-volume applications. Usually, such systems sacrifice features and performance for stringent price objectives. Other suppliers decided to provide products that push the state of the art in one technological area at the expense of another.

At ORMEC, we specialized in multi-axis applications for high-speed, line-oriented automation. We felt that such systems provide excellent servo performance with features that permit ready integration into a variety of factory-automation environments.

Until now, most automation controls have relied on proprietary hardware and software. However, end users have a growing interest in open standards. The goals: to encourage use of commercial hardware and software, eliminate non-standard interfaces, reduce integration costs, and simplify diagnostics and maintainability of the equipment.

A few months ago, a group within General Motors, Ford, and Chrysler produced a white paper that describes the requirements for various elements of an open, modular architecture controller (OMAC). The needs of automotive manufacturing applications, the paper stated, can be examined under these headings: safety and liability, life-cycle cost, flexibility, connectivity, maintenance, and training.

From a hardware and software point of view, OMAC would "encourage use of commercial standards." The controller's hardware bus structure must be a de facto standard (VMEbus or some forms of PC bus architecture, such as ISA, EISA or PCI is preferred). The human interface must support commonly accepted, easy-to-configure graphical user interfaces. IEC-1131-3 standard programming languages must be used to program discrete I/O logic.

In that light, the lure of the IBM PC-compatible architecture is understandable. It is the de facto standard and, because of its market dominance, it also commands the world's largest base of software applications.

PC controller requirements. Although the PC has made many inroads on the factory floor, it has made only limited gains in actual control. Why? Because it lacks the "industrial features" needed in any automation controller. The additional speed and power of today's and tomorrow's microprocessors will, in part, make up for those deficiencies.

Computers used in industrial control should incorporate safety and liability features such as Watchdog Timers, Emergency Stop inputs, and No-Fault outputs. These features guarantee the user that if the computer stops operating correctly, or if the emergency stop circuit trips, the No-Fault output will be released, causing the machinery to be safely stopped.

In industrial controllers, smaller and harder is better. Standard packaging in today's PLCs is panel-mounted with front-loading, removable adapter modules. Industrial controllers generally can operate in temperatures from 0 degrees to 50 degrees C and withstand substantial shock and vibration. Industrial PCs must meet the same benchmarks.

MS-DOS is a single-tasking operating system; industrial control requires multi-tasking. Industrial controllers typically incorporate highly reliable, pre-emptive, multi-tasking kernels. They provide such features at reasonably low cost, while requiring relatively modest resources, such as RAM memory.

Because of the high cost of downtime, reliability and maintainability are paramount in industrial controls. Rotating components, such as fans and disk drives, are generally regarded as the least reliable components in the system and must be avoided when possible.

Non-volatile memory provides a convenience feature long enjoyed by PLC users. This allows machine setup parameters to survive power outages without special programming efforts.

Computer's future power. In some cases, industrial computers have such features, but the price premium is often prohibitive. The combination of Moore's Law and open standards, plus the availability of commercial hardware and software, has substantially reduced that premium.

Our new ORION Series is an important step for ORMEC, but we feel these new controllers are only part of an era where the IBM-PC architecture will make significant inroads. For example, they will no doubt continue the expansion of servomotor-based control into the industrial-automation control market.

We also feel that servomotor-based motion control will grow at a healthy 8-10% per year in the general-purpose factory-auto-mation market. It will continue to displace mechanical means of power transmission due to advantages in flexibility, speed, accuracy, measurability, reliability, and cost.

Servo feedback will continue to be split among encoder- and resolver-based systems. Due to their positioning accuracy and speed of response, encoders will serve 70% of applications in the factory automation market and will be required in 30%. Resolver feedback will be used in 30% of applications and required in 10%. This use stems from such environmental factors as extreme temperatures, shock and vibration, or high levels of contaminants in and around the motor/resolver.

Open digital communications standards (such as DeviceNet(TM), Profibus, Interbus-S and Ethernet) will gain widespread acceptance in factories. This, in turn, will weaken the dominance of proprietary factory networks running on large PLCs.

What all this means is that the distinction between motion controllers and other industrial computers will become further blurred and their cost continue to gradually decrease. More significantly, they will provide substantially more power, features, and ease of use. Within the next few years, the same industrial PC-based motion controller will run multiple tasks: automation control, motion control, human-machine interface, and factory network communications.

Along with this trend, software will need to grow geometrically to provide the increased functionality and ease of use demanded by the users. This software will be provided by multiple vendors who specialize in different applications unified by cross-vendor software compatibility. Independent vendors will support operating systems, programming standards (including IEC-1131), motion control, graphical human-machine interfaces, statistical process control, and open factory-network communications.

Obviously, open standards play a major role in our view of the future. The emergence of de facto standards that gain the broad support of multiple vendors is vital to creating an environment that allows this vision to emerge.

IBM-PC compatible architecture will drive the future of motion-control systems. And a single controller will handle more and more functions.

ORMEC Systems President and CEO Gordon Presher, Jr.: After receiving a EE degree from the University of Rochester, Presher served as a project engineer at Eastman Kodak. He founded ORMEC in 1982.