[sdiy] MS20 & nonlinear Feedback & tweaking.

René Schmitz uzs159 at uni-bonn.de
Tue Oct 30 21:16:25 CET 2001


Hi Nils and all!

>One question:
>
>Isn't this a good example of non-linear feedback? There's a classic "guitar
>stompbox style" diode clipping in the feedback loop. Or is this one of those
>"tricks" I - as a bluddy beginner - doesn't know about.

Sure this is nonlinear feedback. Although its questionable whether the 
resulting nonlinear differential equation qualifies as one that can be
chaotic. 

Why they did it: The gain in the feedback loop is higher for small signals
than 
for large signals. The oscillations that the filter will produce can only
build 
up to about 2Vpp. Above that there isn't enough loop gain to sustain an 
oscillation. When the diodes conduct the gain is about unity.

To sustain an oscillation the signal must be feedback in the same phase, and 
the loop gain must be unity. The integrators have 90degrees of phaseshift at 
the resonant frequency, that makes 180degrees, the other 180degrees are
made up 
by inverting the signal. The gain of each of the integrators is 1/(sqr 2)
(I think) 
at the resonant frequency. So the feedback amp must have a gain of 2
exactly to 
sustain the oscillation. The trouble is one can't precisely set the gain. 
(Think about tolerances for example.)
So one chooses a gain slightly higher than 2. But that means that the
oscillation 
will build up its amplitude each time it gets amplified again. 

To make a long story short, this is a method for amplitude stabilisation like 
it is sometimes seen in sine oscillators. It is preventing that the
oscillation 
builds up so much that harsh clipping at the rails will happen. So you make
a soft clipping, which only slightly distorts the sine. 

The result here is that the oscillations are centered arround GND and not
seen on 
the corners of the waveform as it is in other filters. There is usually
also a 
form of softclipping involved to stabilize the oscillation. For example the
OTAs 
inputs can do that.

Cheers,
 René


-- 
uzs159 at uni-bonn.de
http://www.uni-bonn.de/~uzs159

 




More information about the Synth-diy mailing list