Years ago in the mid ‘70s I was working at Tektronix in their calibration lab. Tek used a lot of purchased power supplies from name brand manufacturers. Our lab also had a repair shop for non-Tek equipment, so we got to look at all kinds of items. One day I got a small 5-10W power supply in for repair. The output voltage was out of range. After analyzing the circuit, I found that a feedback resistor had changed by close to 20 percent from its nominal value.
I replaced the resistor with the same size and value and the supply worked again. A few months later, I had another of these supplies show up with the same problem. After analyzing the circuit more carefully, I found that the precision 1/4W resistor was passing more like 3/8W of power. I replaced the resistor with a 1/2W part and notified the area that they had a problem and to send in all the power supplies for a modification. Once these were all modified, I never saw one again while working in the lab. It was such a simple circuit, but it was not completely reviewed before it was released to manufacturing. Maybe this company was building-in calculated failures for replacement a few years down the road.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.