[sdiy] CEM3328 seems to only have buffers at the end

[Antti Huovilainen]

On Mon, 20 Feb 2006, Harry Bissell Jr wrote:

> I mean the VCF input differential pair, which imho makes the
>  rest of the ladder a slave to it (current flow in the series circuit).

Well, No :)

While the input differential pair has distortion, the feedback from next 
stage is distorted with the same shape, so the distortions cancel each 
other out for f << cutoff. The same applies for all stages (feedforward 
distortion of stage N is cancelled by feedback distortion of stage N+1).

You really should read my paper about it here :)
http://dafx04.na.infn.it/WebProc/Proc/P_061.pdf

>  Its almost impossible to really 'clip' the Moog ladder... it really does
>  more of a compression than a true clipping.

The distortion is very soft (tanh), but it is there albeit being 
partially cancelled below cutoff.

>  I don't think there is 'inherent feedback' in the Moog Ladder unless

Any lowpass filter has feedback. Take the trivial RC 1pole lowpass for 
example. The differential equation is

dVc/dt = (Vi-Vc)/RC

That "-Vc" part is the feedback.

In Moog case the cutoff is controlled by control current and the 
differential pairs (the pairs inside the ladder work the same as the input 
pair) produce distortion so the differential equation is

dVc/dt = Ictrl/C*( tanh(Vin/2Vt) - tanh(Vc/2Vt) )

There's still feedback, only now it is distorted.

SSM2044 also uses this fact to cancel the distortion from the input pair. 
The feedback part of last stage is also a differential pair, so the 
distortions cancel out. The middle stages produce less distortion and 
their effect is ignored.

Antti

"No boom today. Boom tomorrow. There's always a boom tomorrow"
   -- Lt. Cmdr. Ivanova