dav_mt
Newbie level 5
I just designed an FIR filter in Matlab, and saw the impulse response. I obtained a low pass filter. Up to now there were no problems. I read the cut-off frequency(0.144cycles/sample) and had 2 null frequencies at 0.325cycles/sample and 0.5cycles/sample.
Then I inputted a discrete sine wave block(with sampling frequency = 12kHz) instead of the impulse and after trying several frequencies, gain and phase were observed from matlab scope and the frequency and phase spectrum were plotted. Please note that in order to read the gain and phase blocks i used a filter to reconstruct my signal and have readable gain and phase values (input vs output)
I did thesame procedure for a sampling frequency of 20khz, yet in the fs = 12khz case i found the response to be very similar to that of the impulse response with obviously a different cut off frequency (12k * 0.144cycles/sample) and heavy attenuation at the null frequencies just like the impulse input case. However I did not observe any attenuation at fs = 20khz at the null frequncies. Am i doing a Matlab mistake or is there a theoretical explanation for this difference ?
Obviously in both cases two low pass filters were observed whose cut-off depended on fs, using f0/fs = k/N = f(cycles/sample)
Then I inputted a discrete sine wave block(with sampling frequency = 12kHz) instead of the impulse and after trying several frequencies, gain and phase were observed from matlab scope and the frequency and phase spectrum were plotted. Please note that in order to read the gain and phase blocks i used a filter to reconstruct my signal and have readable gain and phase values (input vs output)
I did thesame procedure for a sampling frequency of 20khz, yet in the fs = 12khz case i found the response to be very similar to that of the impulse response with obviously a different cut off frequency (12k * 0.144cycles/sample) and heavy attenuation at the null frequencies just like the impulse input case. However I did not observe any attenuation at fs = 20khz at the null frequncies. Am i doing a Matlab mistake or is there a theoretical explanation for this difference ?
Obviously in both cases two low pass filters were observed whose cut-off depended on fs, using f0/fs = k/N = f(cycles/sample)