As you know, a nonlinear phase filter (in passband), distorts the frequency contents of passband region of signal. In general, Why do we use such filters in the case they cause distortion in the frequency contents of signals?

  • $\begingroup$ dsp.stackexchange.com/questions/18956/… $\endgroup$ Jul 10, 2018 at 17:20
  • $\begingroup$ dsp.stackexchange.com/questions/32129/… $\endgroup$ Jul 10, 2018 at 17:20
  • $\begingroup$ also, in cases where you don't care about linear phase / constant group delay, you obviously don't benefit from linear phase / constant group delay. ("If a bike can't carry an elephant, unlike a truck, why do we still use bikes?" because we don't always have to transport elephants) $\endgroup$ Jul 10, 2018 at 17:22
  • $\begingroup$ Thanks for your nice comments. Suppose we have a rectangular pulse (simply 0 for low state and 1 for high state) , which we added different noises and we want to extract its transient times. What type of filtering you propose? I just know that we should use lowpass filter to cut out the noise in the higher frequencies. But IIR or FIR? I want to just simulate that in PC with different filters. $\endgroup$ Jul 10, 2018 at 17:37
  • $\begingroup$ That depends. A lot in that problem says "I don't care about constant group delay" (and you're able to infer that in that case, you don't need constant group delay, so an IIR would be sufficient). But maybe your system has other constraints (input ranges, stability, latency constraints, complexity constraints, implementation on specific hardware)… So, it depends. $\endgroup$ Jul 10, 2018 at 17:43


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.