Schmitt trigger pulse stretcher, How big an R?

I like! No more parts and it only starts to run out of time when the input pulse is less than 200 ns... that's a big stretch :^) (I'll leave the spec at 10 us.)

Thanks everyone, George H.

Yes. The R as shown discharges the timing cap all the way to GND, which helps increase the charging time.

A 555 or an 'HC123A would be the kosher way if you wanted better accuracy.

Cheers, James Arthur

I tied R to the gate output to increase the timing cap's swing so if you have enough time, it'll get arbitrarily close to ground.

A mosfet's nice too, and makes the timing cycle less dependent on the trigger width. I was trying not to invert the signal but I guess it doesn't really matter--George can fix that.

He only needs 10,000:1, so unless I've slipped here, a 10mA discharge current and a 1uA charge current get the job done, amperes not required.

Cheers, James Arthur

Yeah but it's my 10 meg resistor... hardly any current will flow through it. (I switched it around and got the same pulse width.)

George H.

The R won't do squat for short input pulses. John's MOSFET is better, really. Plus it saves the first inverter.

Your timing ratio's 10,000:1, so even a 50 ohm 'HC14 output (discharging) should allow a 500k timing resistor (charging). A 2 ohm FET makes that even prettier, gets you to 1,000,000:1 pretty easily.

Cheers, James Arthur

The first inverter is connected to an external input.

George H.

You mentioned there was a NAND gate too. If you do this:

then the output is Trc for short inputs and the input pulse width when the input is longer than RC - see how this differs from the stretchers discussed so far which all extend the pulse by Trc, i.e. even for an input longer than RC the output is longer.

I drew in the 220R resistor to protect the schmitt input esd diodes from C overshoot.

High value resistors are something I try to avoid. Up to 470k I am content but I occasionally use up to 10M. I have never used more than 22Meg!

piglet

Huh, Yeah that might be useful. Thanks.

Oh is that enough. I always put diode from ground to the input when capacitivly coupling like that.

With FET opamps big R's can be fun... you can measure pA's. I'll put a few (though hole) 100 M and 1 Gig R's in the mail for you if you want to play. (we ordered a bit many of the 100 M ohms, so I've got surplus.)

George H.

That circuit is misadapted from a higher voltage 4000 series application. I t just squeaks by with HC logic with its VT- min of 0.9V, but doesn't make it with HCT and its VT- min of 0.5V. If there's any accuracy requirement at all on your 100ms timing, you'll end up screen testing any logic schmitt w ith their relatively large hysteretic uncertainty.

It just squeaks by with HC logic with its VT- min of 0.9V, but doesn't mak e it with HCT and its VT- min of 0.5V. If there's any accuracy requirement at all on your 100ms timing, you'll end up screen testing any logic schmitt with their relatively large hysteretic uncertainty.

OK thanks Fred... would it be any better if I did it with the other polarity? (Fast charge up to ~4.4V and slow decay towards gnd.)

Timing accuracy is not at all important. It just needs to be longer than a clock period to make it synchronous. (Clock period is still a bit undetermined.)

George H.

The NAND feedback version I posted will also trigger on much shorter input pulses, just 2*Tpd vs the time to discharge a capacitor. Of course the converse is that there is then a recovery time issue if input pulses come in with very short low time between them.

Thanks for the kind offer of 100Ms - actually I have a few up to 1Gohm reserved for lab experiments which I have not needed yet and anyhow would never dare to design such high value resistors into products.

piglet

te.

R.

n. It just squeaks by with HC logic with its VT- min of 0.9V, but doesn't m ake it with HCT and its VT- min of 0.5V. If there's any accuracy requiremen t at all on your 100ms timing, you'll end up screen testing any logic schmi tt with their relatively large hysteretic uncertainty.

Maybe if you sketch the timing your trying to achieve we could suggest an a lternative.