but I'm not sure it fully addresses the reset issue. The main problem seems to be with the relationship between "sufficient" supply voltage as compared with RESET voltage. The Intel datasheet says it uses a schmidt-trigger. All that's required, then, is a reasonably slow reset. They recommend an 8K2 pulldown together with a 10uF cap. That will take a while to charge, while the supply should come up steadily but "be there" within a millisecond or three. These reset/supervisor IC's tend to incorporate circuitry that does all that comparison. I've never found it necessary, however. There are problems, however, if one has a switching power supply and lots of capacitance on the supply rails, as some switchers don't handle capacitive loads well during startup.

I personally prefer linear supplies.

RE