[sdiy] SVFs with different gains in the integrators

[Guy McCusker]

> True.  But I'm not sure that's an advantage.
>
> One of the big features of the State Variable Filter is that the frequency and Q are independent.  Compared to, say, varying the component values on a Sallen Key filter.  And if you want to make them dependent, it's a simple matter to patch it in.
>

Yes, that's definitely a serious issue with this design. If you want
to control Q independently of F you have somehow to manage the product
of the two integrator frequencies to remain constant. If I remember
correctly the THAT app note that Tom referred to tries to do exactly
this.

Another difficulty is that it is pretty hard to get very high Q. For Q
of 200, say, you'd need a factor of 40,000 variation between the two
integrators. Not to say that someone couldn't put a smart design
together to do all this, but it's a lot harder to tame than the
regular design. I still think it is interesting to know about though.

> Also note that the level of the bandpass output changes.

I think it is the case that the peak level of the bandpass stays
constant in this design, which is not the case if you vary the
bandpass feedback in a regular SVF. My calculations could be wrong.
But I always thought that was why it sounded like "variable slope":
you're able to narrow the bandwidth without causing peaking at the
centre, so around the centre you have a steeper slope from the same
level down.

>  It looks to me like Serge was seriously exploring new filter behaviors specifically for their ability to expand musical expression

I agree 100% with this and did not intend to be disparaging towards
Serge here. The three designs are very different -- Dakota summarised
the architectures very well --  and the naming corresponds to (one of)
the intended musical uses of each. It's just that once I understood
how they work I got a kick out of the cleverness of the naming.

Guy.