| ??? 02/11/07 15:19 Modified: 02/11/07 16:41 Read: times |
#132523 - Indeed! Responding to: ???'s previous message |
Richard said:
I think if the Vcc risetime is short enough, then things work "better" (forgive me for not having a precise characterization) than with the slow or noisilsy rising waveform on Vcc. Exactly! I remember an Intel's errata, which reported about lock-up problems of a certain derivative. The work arround was to guarantee that Vcc rise time is less than 1msec... Vcc rise time IS an issue with the 8051. It's a consequence of the original Intel's design and all following manufacturers of the 8051 reported about the same problems. In many situations, though, this "rise time" issue is more a "reset" issue. Means, if you can guarantee an absolutely proper working reset, even with the slowest Vcc ramps, even from Vcc=0V on, then the problem disapears. That's why I always use a MAX1232 in combination with an additional pull-up resistor on the reset line, like shown here: 
Keep your hands off the built-in power-on reset and brown-out detection. Many many of them are weak and make mistakes. Richard said:
Perhaps one should try a precision-comparator-based switchon Vcc, that supplies Vcc to the "downstream" circuitry only after there have been no variations above 50-60 mV for 100 ms or so, and then switch on that Vcc with a rise time of, say, 50-100 microseconds. Keep in mind, that reset chips need minimum Vcc rise and fall times, so that the internal reset cricuitry can work properly. By the way, the capabilities of a modern reset chip, with brown-out detection, etc, is totally underestimated by the most people. It's extremely difficult to design such a high quality reseter, which has to work perfectly under all conditions, even down to Vcc=0V... Kai |
