[sdiy] square to sine

[David G Dixon]

> If you're going to use 2164 VCAs to get the expo conformity 
> then I'd definitely go with a sinewave oscillator from the 
> outset.  It's crazy messing about with shaping squarewaves or 
> tracking filters if you don't have to.

Yeah, I've abandoned the square-to-sine idea now.  I'm going to shape a
tricore.

> The quality of the sinewave output is going to be much better 
> if you can start with a half-decent sinewave oscillator, 
> instead of trying to strip away the harmonics from a 
> squarewave.  Those lightning fast edges of the squarewave 
> will do their best to couple onto all signals on your PCB and 
> find their way to polute the sinewave output with nasty 
> high-order harmonics!
> 
> If you design a state-variable filter such that it is just 
> slightly unstable to start with, and then add some 
> level-dependent damping (feedback from the BP output,) you 
> should get a very pure sinewave output from the LP output.  
> This is because any tiny amount of clipping that happens to 
> control the amplitude is done at the front end, and this is 
> double integrated before it arrives at the LP output.  The 
> two lots of integration do a good job of attenuating what 
> little distortion is introduced during any soft clipping.
> 
> That's the route I would take now, knowing a little more 
> about your intentions.