Keep in mind that at the time, there was NO official explanation of what we were seeing. We knew a reset problem existed, but we were somewhat in the dark about this. At the time, Philips was "turning the blind eye" on this (or maybe they didn't know). As a result, our design was based on a LOT of testing and could probably be designed better knowing what we now know. My solution that got built into the product consisted of:

1) A reset supervisor. This was our initial attempt to fix the problem. As it turned out, it didn't do much for us. But it was left in the design. Based on experience, you might be able to leave this out, but I honestly didn't test it without.

2) A "pinch" circuit that was designed to IMMEDIATLY pull the reset line to under 0.2 volts when our +3.3v regulator dipped to a specific level. I think the typical school solution is a diode in reverse bias, but if I remember right, your standard diode won't "bleed" the reset line low enough. We used a transistor if I recall correctly. The idea here was to ensure that the reset line went to 0 volts when the regulated 3.3 volts dipped to a specified level. We had to "experiment" into this value and I don't remember what it was, but obviously it was at a voltage above when the brown out detector stopped working.

That is basically it. With the pinch circuit, we had no reset problems. Our application included a heavy duty DC motor driven thermal printer that would REALLY tax our system. Nasty regulator spikes were always a possible threat.

I'll try to dig up the paperwork from this project when I get home if your interrested.

Tomas