The spec on that relay that KAI pointed out suggested that 200 microseconds was a pretty significant time window, as that's the stabile "break" time. It's only eight samples at 40 KHz, which is what that 25 microseconds implies. I think you have to sample at quite a bit, a couple of orders of magnitude, perhaps, faster rate than that if you want to go without an integrator.

Using the integrator, 25 microseconds might work, but ... if you want precise information, information that you might use over a broad sample base in order to get to where you have sufficient confidence in the numbers to publish the result as a "typical" spec, I think you probably want an integrator to smooth the bounce-interval current and you want to sample it more like 50-100 times, perhaps more.

RE