How to plot the internal hysteresis loop of a comparator

[kyttaylor]

I've got a single-ended comparator, and I need to plot the internal hysteresis loop of this comparator which should look like the attached graph below. How do you perform this simulation on Cadence? Supposing the Ref voltage is ground, I believe I need to DC-sweep the input from let's say from -100mV to 100mV and then from 100mV to -100mV, but cadence simulation is not allowing to sweep the input from positive to negative. Any thought?

Also, any papers or books on how to hand-calculate the internal hysteresis voltage range of a comparator? I'm not interested in added external hysteresis because I know about doing this through positive feedback or schmitt trigger. I'm just trying to derive through equations the internal hysteresis of a comparator. Thanks

 

[keith1200rs]

A simple way is to do a transient analysis with a triangular input. Do it slowly so speed doesn't affect the results.

Keith

 

[kyttaylor]

A simple way is to do a transient analysis with a triangular input. Do it slowly so speed doesn't affect the results.

I don't see how transient analysis would work since it will plot a voltage versus time plot. In contrast, the hysteresis loop, I believe, is a voltage (output) versus voltage (input) plot. That's why I tried a DC sweep, but I can't sweep the input from a positive voltage to a negative.

 

[checkmate]

It depends on which wave viewer you are using. Many allow you to plot one wave as a function of another. Also, running ONE transient simulation is much more efficient than running MANY DC simulations, especially for high gain comparators.

 

[dick_freebird]

You would see, if you tried reading the manuals, that the X axis of
waveforms can be changed between the analysis variable
(time) and any of the plotted expressions in the subwindow.

There is alao a "hysteresis sweep" DC option for this very
purpose, that you could select.

 

[keith1200rs]

A simple way is to do a transient analysis with a triangular input. Do it slowly so speed doesn't affect the results.

I don't see how transient analysis would work since it will plot a voltage versus time plot. In contrast, the hysteresis loop, I believe, is a voltage (output) versus voltage (input) plot. That's why I tried a DC sweep, but I can't sweep the input from a positive voltage to a negative.

It does work. You just need to plot input voltage versus output voltage after the transient analysis.

You 'should' be able to do a DC sweep in either direction, but that seems to be problematic with some simulators.

Keith.

 

[kyttaylor]

Thanks guys