This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM2903: Short pulse(noise) input case

Genius 13940 points
Part Number: LM2903



My customer has a question for LM2903.


The datasheet specifies its “Response time” as 1.3us with 5mV overdrive case.

Is this response time just a delay or does it work as a kind of noise filter?


If a short pulse or noise is input like the below figure(VIN-), is it ignored(Case 2)? Or just delayed(Case 1)?




  • Hi Oba,

    why not running a simulation?



  • Hello Oba,

    It is a delay. However, if the pulse is narrower than the rise and fall time, then you end up with a "blip".

    The output will start to rise, and if the pulse ends before the output fully rises, then you are left with a lower amplitude pulse.

    So if you had a 1ns rise and fall time, and your pulse width was 1ns, then you would end up with a triangular, half amplitude output pulse.

    The models must accurately model the output behavior to see this effect. Most of the models have "near-perfect" outputs that may show the full pulse, just delayed.

    So, yes, there is a low-pass effect caused by the slewing (rise/fall times) of the output. So you do not get a full amplitude narrow pulse, but a rounded, lower amplitude pulse that keeps decreasing as the input frequency increases, to the point where it just quits.

    This is why we say the maximum toggle frequency is f=1/(TPLH+TPHL+Tr+Tf), to give enough margin so that the output is not too distorted.

    So depending on the pulse width, you may see a little glitch on the output delayed by the prop delay. So it is not completely "filtered" out. It would be better to add an actual RC filter on the input to filter the pulses. An external filter would be better controlled than relying on the device to device variations (or if you change devices/manufacturers).