The first thing that the majority of responders pointed out was that there was no "RESET IC." However, it turned out that the problem, if there even was one, aside from failure to read the datasheets, was improper coding.

How many other examples of such utter foolishness would surface if people actually investigated the causes of their "problems" rather than assuming that they're this or that and the using the "scattergun" approach to fix them? Doesn't anyone ever perform rigorous failure mode and effects analysis?

I've repeatedly asked for specific symptoms and, instead, gotten "stories" about how adding a supervisor "fixed" the problem. How do we know what the actual problem was? How do we verify that it has been fixed? Is it sufficient that there are no more complaints from a given customer? How do we know that customer simply quit using the product and substituted a competitor's product that actually works properly?

I know it's easier to go to the boss and say, "Well, we've got to add this little supervisor, since the RC reset doesn't work properly," when the real issue is something entirely removed, but of which both designer and boss are unaware. Yes, the problem eventually gets fixed, but not until after there are several changes, and the real cause of the problem is never documented, and seldom even properly investigated.

Up to now, nobody's even offered information indicating that they tried a heftier power supply, one that produces a, say, 1 ms rise time on Vcc rather than 20 seconds. I doubt there'd ever be an oscillator startup problem under such circumstances.

A rigorous investigation would yield information that pointed out the cause of the so-called "RESET" problem has never been found. I recognize that it's probably about the stupidest thing one could do, using a positive-supply's changing voltage as a reference for so important a signal as RESET or INT, but Intel did it, and they can't be wrong, can they?

Until someone comes out with a precise description of the problem, stating that Vcc stayed at +5 within 50mv, and the reset button was pressed, causing the RESET, and the supply remained unchanged, yet the observed reset failure occurred, and then described the precise signal sequence that shows that there was a malfunction, and that the simple addition of a supervisor circuit without any changes in code or other circuitry caused the malfunction to cease, I'm going to continue to suspect that there is no RESET malfunction attributable to the conventional RC reset.

It's not easy to track down a system failure. It's certainly not easy to do it rigorously. So far, I've read lots of stories, totally inconclusive ones, to which people have responded by adding circuitry, but of which no one seems ever to have performed a serious analysis. I think it's a shame that nobody has even attempted to isolate the fault by using a reset switch across the 10 uF cap while monitoring the power supply. I think it's a shame that no one has apparently attempted even to put a man-sized power supply on a small board just to prove that it wasn't an underdesigned power supply.

Just how many of the designers in this forum actually measure the power consumption of their circuits? How many document the power supply characteristics on startup? How many even measure it? How many even know how to do those things?

RE


F