Hysteresis of 74HC14

Read the data sheet -- I think you'll find that proper engineering term is "untrustworthy".

TI lists the high-going switch point as ranging between 1.7 and 3.15 volts, while the low-going switch point ranges between 0.9 and 2.2 volts

-- hopefully without the 3.15 and 2.2 happening on the same part. That makes sense for noise immunity, but it doesn't quite fly for any but the coarsest of analog uses to which you may wish to put it.

Yes, it's not very useful. As well as the postive- and negative-going threshold values, the Fairchild datasheet gives typical values for the hysteresis at various Vdd values, as well as min/max.

At Vdd = 4.5V & 25C typical is 900mV, min/max is 400/1400mV at 25C, and the guaranteed values over -40~85C are the same.

From interpolating, you'd expect the values at Vdd=5.0V to be about

NXP's data sheet shows hysteresis value with the same limits, but substantially different typical values (980mV at Vdd=4.5V):

So, if I did the back-of-envelope calculation right, an 'HC14 RC oscillator could have a period anywhere from 0.356 tau to 1.287 tau at

Best regards, Spehro Pefhany

Not so good for an oscillator, it's true.

Not so horrible for charge dispensing, though, if you're using good quality current sources--all you need is for one of the thresholds to stay reasonably still.

Cheers

Phil Hobbs

Its nice for filtering switch inputs and so. I also used it to generate tones for which the frequency didn't matter as long as it was audible.

I'm just looking to debounce an edge that has slow recurrence, so a few ms time constant should do for my purposes.

(I'd use a 555, but I don't have the space for another IC :-) ...Jim Thompson

My usual rule of thumb is that monostables are fine if you don't care if their delay varies by a factor of 3 either way. If you're using the positive AC feedback method, the timing will be okay on the leading edge anyway. Just an RC lag plus a Schmitt won't be nearly as good.

Cheers

Phil Hobbs

Something I used 10 Gezillion years ago...

If you need ms of delay, you can do it nicely (and symmetrically) with a 555 _provided_ that the pulse width is substantially longer than the desired delay. ...Jim Thompson

As others have said, probably not reliable. Good trip point repeatability is designed in to NE555 and its kin, though.

=A0 =A0 ...Jim Thompson

=A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0|

=A0 =A0 =A0 |

The only purpose for the Schmitt is to sharpen edge rates on slow input signals. Anything else is at your own risk.

Makes a decent oscillator for lots of things. Like charge pumps, as Phil noted. Or time delays that aren't critical.

John

But, do they at least guarantee that the high threshold is more positive than the low threshold in a given part? ;-)

Thanks, Rich

|\ | \ | \ ---------| ?------------ | / | / |/

John

Sometimes you end up with this:

|\ | \ | \ ---------| #%^&@!!------------ | / | / |/

:-)

or

which is an inverted WTF gate

John

Please. "Whiskey-Tango-Foxtrot" gate. Think of the children.

Best regards, Spehro Pefhany

That's what the hysteresis IS. They also guarantee the high threshold won't be too close to the Vdd and the low threshold won't be too close to Vss.

Best regards, Spehro Pefhany

Whiskey? Children? In Canada, maybe.

John

(Actually, bourbon and sugar was a teething pain remedy, when I was a kid)

T (period) = (LOGe((VDD-VL)/(VDD-VL-VH))+LOGe((VL+VH)/VL))*TAU ;-)

(Neglecting propagation delay of the inverter) ...Jim Thompson

I know.

Given the triple constraints on the thresholds and the power supply votlage variable it could be an interesting exercise for someone with some time on their hands to figure out the actual worst-case minimum frequency vs. power supply voltage (I think the worst-case maximum frequency is as I stated).

Best regards, Spehro Pefhany