I tested a comparator MAX40025 using the universal board and discrete components.
The blue curve is one input which is a short pulse, the pulse width is less than 50ns.
The green curve is another input which I set to be a constant voltage(the threshold). This constant voltage came from a voltage divider connected to the power supply.
As can be seen from the oscilloscope image(please ignore the yellow curve and the histogram), the threshold was distorted when the input pulse arrived, and the comparator output was distorted due to this.
Also when I adjusted the threshold level, the blue curve would drift vertically too.


The comparator is fast with a propagation delay of 280ps, so I suppose it could handle this input pulse.
Does the distortion come from the crosstalk between the two input ports? How can I eliminate it? Any hit would be appreciated.

Thank you.

 

We're not psychic. Please post your schematic, also some well-focussed pics of the scope connections.

 

Quite likely there is the potential for crosstalk between comparator inputs. For that you need to read and study the specifications sheets. and the published "internal equivalent circuit", which is intended to reflect the functions of the actual internal circuit but not give away any secrets.