cbbuntz
Newbie level 1
So say we have a 1st order lowpass filter at 44100khz using a pole zero method and I subtract the lowpass from the input to acheive the reciprocal highpass and want to blend it to correct for proper gain in the analog domain by using:
(1/(1 + (22050/fc)^2))^(1/2) or
1/(1 + (22050/fc)^2) for power
so then the highpass and lowpass will be in quadrature from each other at all frequencies and the sum of the lowpass would be ((gain of lowpass)^2 + (gain of highpass)^2)^(1/2). When you get it right, it seems to match the prototype analog curve perfectly.
Question is, how to solve from where to move the new cutoff to retain 3db cutoff. I've been toying with atan and atan2 and have gotten close but never perfect. I've used a few different ways to compute coefficients. Maybe there's an easier method to use to to solve for new compensated cutoff?
(1/(1 + (22050/fc)^2))^(1/2) or
1/(1 + (22050/fc)^2) for power
so then the highpass and lowpass will be in quadrature from each other at all frequencies and the sum of the lowpass would be ((gain of lowpass)^2 + (gain of highpass)^2)^(1/2). When you get it right, it seems to match the prototype analog curve perfectly.
Question is, how to solve from where to move the new cutoff to retain 3db cutoff. I've been toying with atan and atan2 and have gotten close but never perfect. I've used a few different ways to compute coefficients. Maybe there's an easier method to use to to solve for new compensated cutoff?