transfer function order (was Re: filter waveform question)

[Martin Czech]

> > On the other hand it can be shown, that any H(s) can be reproduced
> > by chaining or parallel use of second order filters (state variable),
> > and first order sections (simple RC high or lowpass).
> 
> This is right for (ideal) linear filters. In the real world, with things
> like
> nonlinearity, headroom, and so on, they can sound very different.
> Just think about two 2-pole sections in series. You will typically have
> one high Q and one low Q partial filter. Which one will be the first stage
> ?
> How does the second stage filter the distortion products, when the
> first stage is (ever so slightly) overloaded ?
> 
Serious questions, but I mean this was a theoretical answer.
A transfer function is a (quite succesfull) way to 
describe reality, it assumes ideal behaviour.

But nevertheless, I wonder what will happen if you take
a four stage filter (e.g. cascade) and add the voltages
from every stage together. This should introduce some zeros.

It is not quite the same as using integrators, have to
figure out what will happen in this case.

m.c.