hysteresis variation mc sim

[CAMALEAO]

Hi all. Imagine you have a comparator with hysteresis and you want to determine how much trim you need.
You do MC sim.
Now if you simulate using DC simulation sweeping the input of the comparator, how would like to know the following:
We sweep up and determine the value by taking the input when the output reaches VDD/2.
The same when it is sweeping down.
By doing the difference of the two simulations we the sigma that we get from these two subtraction is the one that we have to check to define the trim range?

Or is something else that I need to do? For example, the offset of the comparator? If so, how to do this? I was thinking, select the voltage at which the comparator should trip and using the dc sweep again, the same one as above, going up subtract the reference by the middle point when sweeping up.
Now I have the offset voltage. Then what? Subtract this value from the hysteresis window?


I am a bit confused.

Thanks in advance.

 

[dick_freebird]

Ideally you would defeat the hysteresis for test
and trim. Depending on hysteresis implementation
the hysteresis may be symmetric or not, and may
wander across temperature, so if asymmetric w.r.t.
input null, the pin-observed offset may drift as well.

Is the hysteresis internal, or from an external
feedback loop?

Using a classical op amp / comparator test loop
can get you a good Vio number although you
would have to run transient simulations which
is a real time waster inside a MC simulation loop.
You also would be a little concerned about
time domain effects such as grossly imbalanced
low-overdrive prop delay, which the loop would
average into a skewed integrator voltage (though
enough gain and low enough filter pole might
make it tolerable enough).

 

[CAMALEAO]

Thanks.

The hysteresis is internal. I have ran transient and getting strange results. Would like to run DC sweep saving hysteresis point (option in dc simulation).
Now forgetting about all the second order effects the question is basically the one I did above. Because in transient I am discounting the offset.

--- Updated Jan 23, 2021 ---

I am asking this because the confusing bit is when we sweep up the offset is included there and when we sweep down the same, so I guess we just subtract one sweep from the other to get the sigma right? what's you view?