??? 06/19/09 13:52 Modified: 06/19/09 13:58 Read: times |
#166257 - Long rise times of Vcc are critical! Responding to: ???'s previous message |
That you should use a proper reset chip with brown-out detection like the MAX1232, I usually use, is evident. But, not widely known, many micros do not like long rise times of Vcc! You need not to be Richard to be worried about this issue, there are hard facts. The built-in Pierce oscillator can stay dead when Vcc rises very slowly! The reason for this is, that a standard Pierce oscillator needs either lots of noise or a steep Vcc ramp to start to oscillate. Manufacturing tolerances of the transconductance of oscillator-FET and temperature drifts can really make that your Pierce oscillator refuses to start properly. Then, even the best brown-out detector will not help.
Long Vcc rise times can also corrupt the sophisticated internal timing of a microcontroller. Think of the internal supply voltage generation for the substrate and flash programming section, for instance. Micros are built for a fast Vcc rise time. If I remember correctly, then I have seen a specification of "no more than 10msec" in the original 8051 datasheet from Intel. Also, some Cygnal micros had a "no more than 5msec" specs, if I remember correctly. Micros having their sophisticated internal timing corrupted often quit this with a dangerous lock-up, a state where the micro seems to be dead but becomes abnormeously hot! Only a full power-down followed by a proper power-up can make him become alive again, provided that it has survived the lock-up! Kai |