SEM VCO buffer (was: RE: Japanese duals (was: Re: CA3046,SSM2210&2220 Replacements)))))

[Martin Czech]

:::Beware the CMOS schmitt trigger...
:::The threshold voltages vary widely from manufacturer to manufacturer. Also not
:::stable with temperature...
:::Standard CMOS is pretty good with respect to both trigger level and temperature.
:::
:::Use Schmitt trigger when you want "some" hysteresis but don't care how much...as
:::long as its "some"
:::

Your're right. These designs are crap from an analog point of view.
They must be, you can't have precision analog components on digital
logic process.  I know all this, CMOS IC design is my day job.

But: as long as we build only a few devices we don't have to worry at all.

Indeed every single wafer will vary in threshold (hysteresis) voltage,
so you don't even have to change the manufacturer in order to get
other hysteresis. These guys come cheap, if some trigger has marginal
hysteresis, take another one.  As long as the threshold voltages don't
change during operation this is no problem. You also have to select for
leakage, but this has to be done also for opamps, or big fun is comming
your way. The hysteresis is also dependend on supply voltage, which means
that you can change it or not, anyway, supply variation is no problem
with a proper regulator. The temperature problem is wellknown to us,
other components of other designs will also drift, it is not only the
exponential pair that matters.

The interesting thing about a relaxation or triangle osc. with such a
CMOS trigger is the minimal parts count, and (I hope) fast switching,
this means better hf tracking. This could be interesting for hf vco,
especially if only pulse output is needed.

In both cases you will not get away without hysteresis.

Another application could be the "cheap" vco thing, six comparators in
one package.

m.c.