LM393: LM393DGKR differential filter value

Hello Red,

Do not place a capacitor on the positive summing node (pin 3) if you are using hysteresis. Adding the capacitor will delay the hysteresis, causing brief "chatter" bursts at the transition.

You have asymmetrical hysteresis due to the diode in the feedback path, which means the falling-edge will have hysteresis, but the rising edge will not - so it still can have some "chatter" on one edge. You can add a second diode (opposite direction) and a second feedback resistor to provide hysteresis for the other transition. However, you have large hysteresis (2k/2k), so you may be able to parallel a higher value resistor (~100k or more) from Pin3 to the output to provide basic hysteresis until the R900/R900 path conducts.

The hysteresis feedback works against R901 - so any filtering should be before R901. 2.5V looks like a reference line, so the reference line should be cleaned-up at the source.

It is okay to add a capacitor from Pin2 to ground for filtering, since this is not in the hysteresis feedback path. So I would connect C900 from Pin 2 to GND, assuming you can accept the slight delay caused by the time constant created by C900, R902 and R903.

Hi Paul

I got your points thanks a lot  

Back to the question, How to calculate the cut of frequency if C900 already here,

Is the equation equal to 1/(R902*C900) ?