tri2sin

[Martin Czech]

> So far as sine shaping from a triangle wave goes, the method I think 
> is the neatest and partially built an oscillator this way is that of 
> a nice analog computer.  In this case some 13700's were used to 
> generate 10 terms of a Taylor series from the input (triangle) and 
> all of the coefficients had knobs for magnitude and polarity control. 
> This way you get sines and all sorts of other nice stuff.  You have a 
> lack of normal organ-type timbre control in that this is a polynomial 
> series rather than a harmonic (Fourier) series and to complicate 
> matters, it is somewhat challenging to determine the coefficients you 
> need for a given waveform desired.  Also, only having triangle input 
> is a limitation and I suspect that feedback could have very 
> interesting implications.
> 
A simple halv-wave-rectifier array can do the same job, I think.  But
for only sine-wave apps, the normal diode feedback network should be ok
(e.g. Elektor LFO tri2sin or Tietze/Schenk temp.  compensated design).
The only thing is that you have to select pairs of matching diodes,
which can be done right in the evaluation circuit, if sockets are
provided. THD < 1% should be possible.

m.c.