RISE AND FALL AND RISE
Once again this spring, I made my annual pilgrimage to the IPC show, which used to be the International Programmable Controller show. I can still remember the vary first one, at least 25 years ago, with only half-a-dozen exhibitors in the cellar of the EDS building. Over the next 20 years, the show grew until, each year, for the better part of a week is occupied a very substantial portion of Detroit's Cobo Hall. The show's growth was a reflection of the fact that an entire industry had grown up around the programmable logic controller.
The programmable logic controller (PLC) is to factory automation what the PC is to office automation. It is a microprocessor-based device that stores instructions in programmable memory and implements logic, sequence, and timing functions based on inputs from limit switches, push buttons, thermocouples, and other devices. Introduced in the automotive industry to replace relay boards, today even the smallest PLC's may be equipped with serial communication and analog control capabilities for loop control of flow, level, temperature, pressure, and other variables.
Now, 25 years later, the PLC industry has gone though several waves of consolidation. although still in wide use, the PLC is more of a commodity item and there are fewer programmable controller manufacturers than there were several years ago. More and more, the PC is to factory automation what the PC is to office automation.
These facts, too, have changed the IPC show, which isn't quite what it used to be. In fact, next year it will be called the International Automotive Manufacturing show. The pace of technology change is now such that one can not only see something new appear on the face of the earth, one can also, in that same lifetime, see it disappear.
While at the conference, I sat on a panel of experts discussing the current status of the PLC. As the pundits argued endlessly about various standards, flow charting, Microsoft Windows, and laptops, global competition and technological turbulence were eating our lunch. It seemed to me that the panel, and the attendees, should be concerned with the future instead of the past, and I said so. Actually, what I said was that Detroit seemed mired in technology's third world. I gotta do something about my temper.
Now, a more appropriate subject would be something like the impact and direction of computer chips. Processor performance improves by a factor of ten every five years. Or a factor of one hundred over the next decade. Right now, RISC computer chips are running at about 150 megahertz. By 2002, they will be running at 1.5 billion hertz. And in 2006, 15 billion hertz is predicted.
Object-Oriented software development promises that in the future fewer lines of code will be needed for equivalent functionality. But to enact object orientation, a corresponding increase in computing power and compiler know-how is required. Now, powerful chip will be needed because of the new kinds of software.
Applications of the new power sound like things out of a world of science fiction: speech translators, PLC Web platforms, disbursed databases, sentient behavior, and transparent use of control. In general, software models of "real" systems will be used to monitor and control those systems.
One company I'm associated with, Flavors Technology, builds hardware and software for rule-based agents that have emergent properties: Initially, we had "only"680xx chips available for use. Up to 128 of these chips were used to get 16,000 agents all running in parallel at 60 hertz. The boards were 22 inches long. Now we do the same with standard VME cards. Our Asian customers want even more power - up to a million agents all resolved in the same 16milliseconds. This will be possible over the next decade.
Applications, in general, will move off the desktop an into real time. Word processing have gotten about as good as it needs to get. After all, we can only type so fast. But planes, trains and automobiles, scheduling, and data processing can all make use of the possibilities inherent in the convergence of the Worldwide Web, expert systems, LAN legacy support, and emergent capabilities.
We need to gain a greater understanding of the new software technologies. We need compiler experts in parallelism and agents. We need semiconductor resolution at the sub-micron level - at least down to quantum effects. We need chip with memory bandwidth. What we need is an I/O fire-hose to make the CPU chip work. Look for a future, not of computers, but of appliances.
As appeared in Manufacturing Systems Magazine July 1996 Page 148
References - Table of Contents
|R. Morley Incorporated
614 Nashua Street, Suite 56
Milford, NH 03055-4992 USA
Tel: 603-878-4365 FAX: 603-878-4385