> >It sound very noisy button. I have not seen 30 ms bouncing signal
> >coming from a button. I assume you did not see it under a
> >osciloscope. My avagare bouncing lasts for less than one ms. I use
> >three successive stable readings to validate a keystroke within 12ms
> >periodic polling to read matrix key strokes and it works just fine.
>
> I've got a truckload of those buttons  :(
>
> I shouldn't really complain because it has allowed me to come up with some
> pretty efficient routines for debouncing.
>
> If you're polling the button (no interrupts) the algorithm is:
>
> DebounceLoop:                   ;  Come Back Here to Wait for the keypress
>   if Button_Up goto DebounceLoop
>
>   Counter = 20 msec delay       ;  Wait a solid 20 msec to assume good down
> DebounceWaitLoop:               ;  Button in this loop
>   Counter = Counter - 1
>   if Counter == 0 goto DebounceEnd
>   if Button_Up goto DebounceLoop  ;  If Button goes up, start over
>   goto DebounceWaitLoop
>
> DebounceEnd:                    ;  Button is down and Debounced
>
> Note that I continuously poll the button and if it ever goes back up I
> restart the process.  Note that the same thing should be done for the button
> release.
>
Since nobody's mentioned it yet, there is the debouncing system that uses
parallel counters to debounce buttons.  It uses simple boolean logic to
low pass filter 1 to 8 inputs.  It has the advantage of requiring only
one counter (spread over two bytes) and can debounce up to 8 inputs
simultanously.
I found it on 'n PIC related web page a while back - I cannot remember
the URL or the original author anymore, but it really works well.

Niki