FPGA Central

The literature on biquad coefficient calculations such as RBJ and
Orfanidis assume that minimum overshoot/undershoot is desired. In the
case of the cookbook shelving filters, overshoot/undershoot is tightly
coupled to steepness and only occurs with high values of Q.

I want to decouple these in order to create a biquad shelving filter
with shallow slope but still incorporating over/undershoot. Is this
possible? Can anyone give me a push in the right direction?

Bob Fredericks
[email protected]

Bob wrote:
> The literature on biquad coefficient calculations such as RBJ and
> Orfanidis assume that minimum overshoot/undershoot is desired. In the
> case of the cookbook shelving filters, overshoot/undershoot is tightly
> coupled to steepness and only occurs with high values of Q.
>
> I want to decouple these in order to create a biquad shelving filter
> with shallow slope but still incorporating over/undershoot. Is this
> possible? Can anyone give me a push in the right direction?

You need more degrees of freedom than a simple biquad can give you for
this.

There are several options. The simplest one is that you use one biquad
for the shelving filter (with adjustable slope), and use two more
biquads (of peaking type) to generate the over/undershoot. You should
easily be able write a routine that generates the three biquads given
parameters "shelving frequency", "shelving slope", "shelving boost"
and "over/undershoot" (in dB).

You can set the Q of the over/undershoot peaking filters depeding on
the slope of the shelving filters to generate a smooth-looking
magnitude curve. The center frequencies depend on the "shelveing
boost" and the "shelve frequency".

> Bob Fredericks

Regards,
Andor