Feature Archives | Feedback

Musings on the Year 2000 Problem in Embedded Systems

by Arnie Berger
Applied Microsystems Corp., Redmond, Wash.

On New Year's Eve, 1999, I plan to do what I do every New Year's Eve. My wife and I will snuggle down with some videos, play Scrabble, kiss when the TV says to kiss, and then go to bed. I will also say a short prayer for anyone traveling by air. Finally, before New Year's Eve, I will make sure that my generator is working and full of gas and that I have lots of batteries, candles, and foodstuffs.

Why the fuss? Everyone says the millennium problem will only affect the big information systems. Embedded systems, so I'm told, are pretty much immune to the problem. I'm not so sure. In the Microprocessor Report (Dec. 29, 1997), Jim Turley says that embedded systems are largely immune to the angst surrounding the new millennium. He goes on to say that the problem is largely due to shortsighted chip designers who designed their real-time clocks with only one byte of storage for the year. He also makes the argument that the problem, in most cases, will be benign. The most obvious manifestation of the problem will be a miscalculation of the day of the week in systems, such as elevators, traffic lights, and air conditioning, that need to differentiate between weekday and weekend operational modes.

I believe that Jim Turley is missing an important point. The issue is not, "If a date gets mangled, so what?" The issue is "What happens to that miscalculated date as it bubbles up through the rest of an embedded system?" As an example, suppose the date is used to tell a central control when the last time a pressure sensor was calibrated. What will the control system do if the sensor hasn't been recalibrated in 100 years?

My Jameco catalog has an ad for a single-chip controller device that includes a software patch for the Y2K problem. Obviously, another one-byte real-time clock bug. This got me to thinking about how pervasive embedded systems have become. I remember reading that 90% of the dollar volume of embedded systems is concentrated in the 8051 microcontroller and all of its variants. How many of these devices are "deeply embedded" in other devices, which are deeply embedded in other systems?

Concern for Y2K-induced failures in embedded applications has been growing lately as people begin to do more in-depth system analyses of the problem in their particular environments. I'm sure that they will write articles and we will be able to gauge the magnitude of our problem from their experiences. Unfortunately, if we wait to hear how they resolved the problem in their environment, it may be too late for the rest of us. This is one schedule milestone that we can't let slip.

Arnie Berger is the director of research and development (R&D) at Applied Microsystems Corp., of Redmond, Wash. Before coming to Applied, he was the development tools marketing manager at AMD and an R&D project manager at Hewlett-Packard. Prior to working in the electronics industry, Arnie was a research physicist at Argonne National Lab, specializing in the study of crystalline defects in metals. He received his BS and PhD degrees in materials science from Cornell University. Arnie's e-mail address is arnie.berger@amc.com


		Click here to get your listing up.