19 Nov
2017
19 Nov
'17
2:15 p.m.
Ralf Mardorf <ralf.mardorf-ZCLZIpdjs0kJGwgDXS7ZQA(a)public.gmane.org>
writes:
On Sun, 19 Nov 2017 13:10:24 +0100, David Kastrup
wrote:
Well, the art is actually distilling useful parameters. For a grade 2
IIR I get something like parameters a1, a0, b1, b0, but manipulating
those would be a lesson in frustration. For the LaPlace transform of
the transfer function of the high EQ (don't ask me which Onyx I was
looking at, probably the 1620 because I have it) I get something like
Pot ranges from b = -1 .. 1
Out = In*((3.47kOhm+1/(0.0033uF s)+10kOhm-b*10kOhm)|510kOhm)
/((3.47kOhm+1/(0.0033uF s)+10kOhm+b*10kOhm)|510kOhm)
Where x|y is (xy)/(x+y) and s is the transform domain variable (for a
Fourier transform, it would correspond to 2 pi j f with j being sqrt(-1)
and f being the frequency).
Obviously, with b = 0 output and input correspond.
If we convert this to a general IIR form, we'll likely get more
polynomial coefficients than we are comfortable choosing all on our own.
--
David Kastrup
Correct. The problem is _exactly_ that it can do
whatever any other
can and that you have far too many free parameters to get under
control. A few distinctly different good choices in practice lead to
better results, particularly given time constraints, than a full
continuum of every available choice.
The continuum is what tool builders can work best with. But the
actual use cases want ready-made tools.
You could use presets for some parameters of the EQ and only change
the desired parameters of the EQ.