| ??? 08/10/07 06:08 Read: times |
#143034 - I never doubted that, Kai Responding to: ???'s previous message |
Kai Klaas said:
Lynn once told, that the flash writing hardware needs a minimum supply voltage to produce the programming voltage. If the reset chip properly resets the micro down to this minimum voltage, when decaying, then there should not be any harm to the flash. Yes, that's what he said, more or less, but, what he also said was that the RESET is not a true RESET, but, rather, a sort of NMI, which ensures that certain values are restored to their initial state. He also stated that the oscillator has to be running in order to effect this, so it's not hard to envision a scenario in which, while Vcc is decaying but not yet sufficiently to stop the oscillator, that the internal logic of the MCU could, momentarily run away. The likelihood that it will accidentally stumble into the flash writing routines is a mite small, but it is finite. I cannot see any source of trouble. And we had no, so far. But we had with this stupid RC-reset.
Kai You are, of course, a thorough and rigorous investigator of faults in your circuits. If you had trouble with that "stupid RC-reset" you surely made some observations that persuaded you that it was, indeed a RESET problem that you were trying to solve, didn't you? I've done a little experimenting with RESET and with Vcc fall times. If I use a really hefty linear power supply, one which, with the on-board ~2750 uF of capacitance still produces a rise time on Vcc of about 140 microseconds, when switched on, I see no observable flaws in the RESET behavior. However, when I switch off the power, regardless of whether I use a DALLAS, Philips, Signetics, Siemens, or Intel MCU, it frequently (about one time in 6) corrupts my BBRAM, which is mapped into both data and program space. It does not exhibit this behavior on a pushbutton RESET. If I omit 2300 uF of the capacitors, it doesn't seem to happen at all. That, in my opinion, is not a RESET problem. The decay time of Vcc down to less than 2 volts is well in excess of 2 seconds with all that capacitance. I've not measured it without ... yet. I fear that many such problems don't receive sufficient study. This "RESET problem" seems to have arisen concurrently with the adoption of small "wall-wart" supplies and small inexpensive switchers, both of which may provide a slow enough Vcc rise time to cause problems. They also demand a substantial amount of capacitance on the board to minimize brownouts, and that suggests that the decay time of Vcc will be quite long. I doubt that it's at all a mystery that, at about the same time, and under such circumstances, internal FLASH memories started to behave badly from time to time. That's not a RESET issue, though, is it? After all, what the MCU does once the Vcc falls below a certain point is pretty much undefined, is it not? RE |
