AVR-Chat

--- In AVR-Chat@yahoogroups.com, "brewski922" <brewski@i...> wrote:
> 
> --- In AVR-Chat@yahoogroups.com, Valerij Avrealij <avrealij@m...> 
> wrote:
> > Hello 2AVR-Chat members.
> > 
> > I have interrupt based keyboard inquiry. How much time I need wait
> > after interrupt occurs and keyboard inquiry?
> > Thanks.
> > 
> > --
> >  Valerij Avrealij
> 
> It depends on the bounce time of the switches. Sometimes the 
> manufacture datasheet has the specs for the bounce, settling, time. 
> Most don't. Probably needless to say, but I'll say it anyway in 
case 
> there is folks that are not awhere of it. From the time that a 
switch 
> makes initial contact and a valid read is made of what switch 
created 
> the closure, the contact needs to settle to a continuous closure.
> 
> Anyway, for your case if you feel you are having problems with 
> contact bounce or something else and feel a time-delay is need go 
> ahead a insert a delay. Start with something big that WILL correct 
> the problem, say a 0.001 or may be 0.01 or even 0.1 and see what 
> happens. When the problem is corrected say between 0.001 and 0.01, 
> split the difference and try 0.05. If that worked split the 
> difference again. You will eventually get to a point where one 
delay 
> works and just a bit faster fails. Using the number that works 
> increase the delay a little, I usually use 10% slower. This gives 
you 
> a little insurance.
> 
> Mike

0.001 sec or 1/1000 or One One-Thousands of a second
0.01 sec or 1/100 or One One-Hundreds of a second
0.1 sec or 1/10 or One One-Tenth of a second

The theory is 
"Find a value that does not work."
"Find a value that does work."
Then pick a value 1/2 way between the two. Does it work?
Keep on dividing and testing until you get to a value, say 0.123, 
that works. A value of 0.122 on occasion fails. Add 10% to it.
0.135 sec = 0.123 + 0.012 or 135/1000 or 
delay_ms(135); //This seems to be a huge amount to me!
It is better to be slow and correct than fast and wrong.
Ten per cent is old engineering guess-work.
Mike