Hallo Michael,

datasheet tells, that when sensing low level at INT0# or INT1# input wake-up is invoked and with release of interrupt signal (level change from low to high again) wake-up is exited. Means, after falling edge mcu should not do anything before oscillator has started-up correctly. Then, when oscillator is working correct again, after at least 10msec, signal at INT0# pin should be set to high level. Only then mcu executes normal program. Means, if you have a flip-flop, which starts wake-up, mcu will not be able to reset flip-flop, because mcu awaits a low to high transition at INT0# pin. Only when this arrives mcu resumes normal program, which could deliver some reset signal for flip-flop.
Hope, that my english is not too confusing. But, wake-up is a two stage process: 1. High to low level change at INT0# input, which starts oscillator, and 2. Low to high level change at INT0# input, which let the mcu leave wake-up condition and causes the resumption of mcu.
That's how I understood datasheet, but may be I overlooked something...

I have drawn a little schematic for the case that Javier thinks about the use of monoflop.

By the way, your solution with CD4082 in the other thread was much more elegant than mine with LM393. It demonstrates, that working with linear circuits often results in drastically increased amount of electronic parts. Here a filtering, there a level shifter, etc.

Here is the monoflop circuit: It gives a pulse width of about 100msec and is retriggerable. Of course, also CD4082 could have been used, then feeding it's output to B-input of CD4538. But I wanted to use Schmitt-trigger input at A-input.


Regards,
Kai