When designing a filter with a prescribed magnitude response by the Bode ``stick diagram'' method, we think in terms of poles and zeros ``breaking'' at certain frequencies. For example, in the term

which is scaled to have unity gain at , the pole is said to ``break'' at frequency . This is easily seen to be the dB point of the term, since

which has magnitude dB. Thus, the gain of the term is approximately constant out to , where it reaches its dB, or ``half power'' frequency, followed by its asymptotic roll-off of dB per octave. A zero term similarly starts out flat, reaches magnitude-gain dB at its break frequency , and asymptotically approaches dB per octave for .

The Bode design procedure is then to start at dc (
) and map
out break-frequencies for poles and zeros so as to follow the desired
response as closely as desired. Since this tool is commonly applied
in control-system design, there is also usually consideration for the
phase plot as well, which has similarly simple
behavior.^{17}

The basic Bode ``stick diagram'' consists only of straight line segments, each having slope given by some integer number of nepers per neper (or integer multiple of dB/octave, etc.), with the knowledge that the actual response will be a smoothed version of the stick diagram, traversing the dB points at line-segment intersections corresponding to isolated breaking zeros and poles, respectively.

Download spectilt.pdf

[Comment on this page via email]

Copyright ©

Center for Computer Research in Music and Acoustics (CCRMA), Stanford University