AN: Debounce Switches

This individual has the last word on debouncing switches, we don't need you reinventing the wheel.

How about this?

And some oldies...

What's the matter Freddy ol' boy, not getting any lately? Have you tried serving her some Kibbles 'n Bits?

That's okay although I personally would use 1KR's in series with Q's before the wiring junctions...

Projecting again?

ed

d a

vised for modern CMOS.

PG

PG

Maxim published an app note a few years back showing the impedance bounce o n these commodity switches, they're not clean, sometimes 10s thousands of o hms. That's why it's not wise to get too clever with the digital stuff as y ou risk sending a metastable impulse downstream into your logic. With Ganss le's RC techniques, simply making the RC transition through the hysteresis zone confidently longer than any conceivable total bounce time, anything li ke that becomes impossible.

There's no "her" in Freddy's life... he's a fairy >:-} ...Jim Thompson

LOL- typically delusional Trump supporter... you probably have midget hands too.

Once you add resistors, it becomes trivial and not fun any more. Just bang the preset and clear.

Dare to be goofy.

Oh, the sensible way to debounce is in software, or in an FPGA. If analog, use a pullup, an RC, and a schmitt. I'm playing with other ideas at various levels of silliness. Try it... it's good for your brain, like exercize is good for your body.

e:

osted

and a

ce_2.JPG

advised for modern CMOS.

F.JPG

e on these commodity switches, they're not clean, sometimes 10s thousands o f ohms. That's why it's not wise to get too clever with the digital stuff a s you risk sending a metastable impulse downstream into your logic. With Ga nssle's RC techniques, simply making the RC transition through the hysteres is zone confidently longer than any conceivable total bounce time, anything like that becomes impossible.

Even then you want to keep the switch input away from logic, unless it has Schmitt inputs. The worst switches are those cheap momentaries where the us er applies all the contact pressure, other types with an internal snap acti on mechanism of some kind are okay, and relays are okay, but you still need a schmitt in there in all cases.

ideas at various levels of silliness.

If your switch uses a Hall sensor, Wiegand wire, or photointerruptor, the Schmitt is dispensable (there's other ways to get hysteresis). And moving-magnet/saturating-core keyswitches are a personal favorite of mine.

A 6809 doesn't bounce very high. :)

The non-snap-dome rubber and carbon SPST membrane switches are the worst. They don't exactly bounce, but resistance varies with contact pressure. And the user's finger is a nice noise antenna. Those need software or schmitts.

We're doing an aluminum front panel now with a molded silicone thing sandwiched between the panel and a PC board. The silicone layer will have snap domes that make contact with a finger pattern on the board. Every pushbutton site also has a molded-in light pipe in the center, that pokes into a hole on the PCB that has a down-shooting surface-mount LED. Should be interesting.

All the switch states get loaded into a shift register periodically, maybe 10 Hz, so we don't need debounce.

The advantage of the Schmitt is tolerance of analog input voltages, the hysteresis is gravy.

Maybe he could use one if we sent him a glue gun.

Ed

Not debouncing the 6809 processor, but a button. I need to wire the button to the processor. But then, maybe we need to debounce registers in a processor... :)