https://www.allaboutcircuits.com/te...-examples-of-fir-filters-using-window-method/

The second example for the high pass filter mentions that the Hanning window must have an Odd length should it be used as a high pass filter. Can anyone shed any light on why this is so and if possible mention where i can read more about the same?

 

https://www.allaboutcircuits.com/te...-examples-of-fir-filters-using-window-method/

The second example for the high pass filter mentions that the Hanning window must have an Odd length should it be used as a high pass filter. Can anyone shed any light on why this is so and if possible mention where i can read more about the same?

Hi,

For a real quick short suggestion, you can find a lot more info on this in a book on Digital Filter Design. Probably stuff on the web about this.

 

Hi,

For a real quick short suggestion, you can find a lot more info on this in a book on Digital Filter Design. Probably stuff on the web about this.

Any books that you can recommend? I have already tried searching for this in Oppenheim and Proakis & Manolakis

 

Any books that you can recommend? I have already tried searching for this in Oppenheim and Proakis & Manolakis

Oh you already checked Oppenheim? I would have suggested that.

In such a book it looks like the idea is to get a linear phase response, but that would be with any window type or at least most of the common ones. That might also correlate to the window being centered at (N-1)/2, which means there are equal samples to the left and to the right of center and there actually is a center sample. So it's probably that the odd number of samples is not as important as the centering, but centering depends on an odd number of samples.
Also interesting is in many of the examples they use an odd number of samples too.
Keep in mind though I haven't worked with this in detail in at least some years now. I was more heavily into this back in the 1990's.

To look at this in more detail, try using one window that has an odd number of samples and another window that has an even number of samples, then try to filter an example input signal with both of those and see what the difference is on the outputs of both.