Digital line "blanking" effect (debounce?)

On Tue, 15 May 2012 11:32:19 -0500, Tim Wescott wrote:

Kind of like the logic on a refrigeration thermostat to prevent cycling too frequently.

Maybe a couple of 4011's connected something like this?

In __ o---+---------| \ __ | | )o--------------| \ | +--|__/ | )o----+----o Out | | +-|__/ | | | | | | +----------+ +--| ----------+------------+ | | | | | | | | | | + ----------+ | | | | | __ | | | | __ | +----| \ | | | __ +--| \ | | )o----+-------+ | +-| \ | )o--|-----------|__/ | | | | )o--|__/ | | | +-|__/ | | | | --- --- | C --- --- C | __ __ | | | / |- + / |-+ | +-----------o( | +--o( | | | \__|- + \__|------+ | | .-. .-. | | | | R | | | | R '-' '-' | | Vdd o-----+----+

(replace the RCs with something that treats the input diodes more gently if you prefer).

I have to give in to some compulsive behaviour: NE555

Positive _and_ negative edge triggered one-shot, but configured as non-retriggerable during your 50usec interval?

At that speed why not use a 555 timer ?:-) ...Jim Thompson

Moto used to make a switch debounce part that was basically a shift register and a multi-input AND/NAND gate to a latch. Believe it was 6 deep. It wouldn't do the instanteous switch you want, it always delayed the output by 6 clocks, but it removed any glitch smaller than those six clocks...

Charlie

Sounds like a PLL to me.

Jamie

What is supposed to happen if the input is continuously switching faster than 200kHz? ...Jim Thompson

Yup. I've shown several such circuits here, using a 555. But it lacks Tim's "From rest, if the input changes the output immediately changes to match". It has a delay similar to the Moto debounce.

And Tim's timing specification is confusing... 200kHz and 250us... which is it? ...Jim Thompson

You're kind. My specification was confused, not just confusing. I _said_ the caffeine hadn't kicked in yet.

Here, let me grab a CALCULATOR, which keeps track of the @#$% decimal point for me: 2.5us, which, since it's a half-period, works out to 200kHz max frequency.

In that case, having the output switch at, or slightly below, 200kHz would be perfectly acceptable.

So 1MHz input => 200kHz output ?? No synchronism? ...Jim Thompson

You need 2 int's in software, one external on an IO pin, your external reference signal and one internally to maintain a square wave time constant with a high speed internal counter ported to a output pin.

Basically the external INT would query the internal hardware clock register(ISR) for a value lets say on the low side of the input pulse and then query it again on the high side of the input pulse. At the last query, this count value that has elapsed will then be used to program the internal interval counter that fires off an INT to toggle the output or if you want to directly program the hardware regs to have the counter directly toggle the output, etc..

Sounds like you're making a 180 degree tracking pulse?

Jamie

The desire is to roughly track the duty cycle of the output of a comparator, without switching some IGBTs so fast that they turn to smoke immediately (yes, IGBTs, 200kHz -- I know, I know).

Things are in closed loop, so in the event that the comparator output was chattering, slowing down the transitions to the IGBT gate would cause extra hysteresis in the controlled variable, but would still retain the correct duty cycle.

Yes, it could be done in software. If the customer trusted the software not to tank at some crucial moment and slag some expensive electronics.

I agree. Spehro's approach, R-S with "qualifier" looks like the way to go. ...Jim Thompson

Something like this...

Requires "presence" of one state for >= 2.5us before it can be triggered to the other state.

Logic reduction is left for the digital guys. A mixed NAND/NOR R-S would likely simplify the steering. ...Jim Thompson

Poking around, timing from the output side, a major issue jumps out... the R/S flop is essentially level triggered... so more logic, or front-end it with something edge-triggered. ...Jim Thompson

=20

match

Immediate thought, state machine. Now i will read on and see what others thought.

?-)