I don't buy that debouncing code for a second. 

posted by mrflip at 02:20AM CST on January 23

HHhmmm.... to trust the anonymous website or the known experienced coder. I think I'll go with the experienced coder on this one. Interested in trying to build one? Maybe PoblanoMacaroni would like to jump in, too? 

posted by javelina at 12:36PM CST on January 23

Just to be clear: I bet the whole thing works for the most part -- but the author admits the device is occasionally flaky, and I am highly suspicious of the debouncing strategy (which is simply to poll every 10ms). I don't see how that keeps you from sampling during a bounce. Also to note -- the device is an old-school analog cell phone, not a modern digital SprintPCS/Cingular/TMobile/Whatevs. 

posted by mrflip at 12:55AM CST on January 25

While it is not that fancy, I don't think the debouncing strategy is necessarily as bad as you think. The way the while loops are structured, the 10 ms delay is synchronized to the change in logic state, so it's not as if there is a 100 Hz polling routine that may or may not hit in the middle of a bounce. However, the real question is how he came up with 10 ms. You obviously have to choose a time here that is longer than the bouncing time but short enough that you don't miss a '1' or '0' entirely. A little characterization of the bouncing with an o'scope goes a long way here. 

posted by habcous at 11:59PM CST on January 25