AW(2): VC delay module ideas?

[jhaible]

Wow, a lot of mails on this topic, when I came home on sunday and checked
my mailbox !

I think Don's idea with the OTA-based Delay line is very interesting.
My fist thought was "wouldn't that cause tracking problems, too,
with the delay approximated only (and not starting with a precise
delay from a BBD line) ?" - But of course this question is a stupid one.
You can look at the *whole* thing as a *filter* rather than a delay +
filter,
and with all OTAs tracking the whole "filter" will be tracking. Just 
as a BBD + simple z-plane filter would be tracking. It's the mix
of both things that would cause trouble.

So, what's the problem ?
I think for this PM stuff we need a filter inside the delay that is *not*
tracking, with the whole system still producing the properly scaled tones.
Correct me if I'm wrong, but things like how much of HF and LF end is
absorbed or reflected in a waveguide should not depend on the pitch
you want to play (i.e. length of the waveguide), should they ? As far
as I know (and that's not much in the PM case) the whole thing gets
its life from interaction of parameters that are partly fixed (properties
of the wood or metal, shape of the end of a tube, ...), and partly
variable (length of waveguide, ... ? ). Otherwise we could almost as well
use a sampler and simply transpose the sound.

It should be fairly impossible to build an analogue sytem with variable
delay and a fixed filter function, and the whole thing still properly
tracking.
Just imagine a simple thing like a delay and a fixed 6dB/Oct lowpass
ant 5kHz, and a fixed 6dB HP at 200Hz. How would you adjust / correct
the delay time so that feedback over the whole system gives oscillation
that tracks over the whole audio range ?

JH.