Very nice! I'm filing this away just in case I need it some day.
A while ago, I came across an extraordinarily simple circuit that converts triangles to sines rather well, without the shark kinks or pointy waveform tips you get with piecewise-linear or tanh shaping. Apparently this circuit has been widely used in analogue synthesizers, but I have never seen it used in an actual audio function generator.
The circuit was apparently described by Middlebrook & Richer in 1965, and uses one single JFET, two diodes, and four resistors. More here: Tim Stinchcombe - Tri to sine conversion
I haven't built this one, but in LTSpice simulations, I was able to get better results with it than with the other more common methods I tried (piece-wise linear diode-resistor networks with cusp cancellation, differential amp with cusp cancellation). And the circuit is almost ludicrously simple by comparison.
The only significant downside I found is that there is quite a lot of signal loss - the sine wave that comes out is considerably smaller than the triangle that went in. But at least in simulation, the results are very good considering how simple the circuit is.
-Gnobuddy
A while ago, I came across an extraordinarily simple circuit that converts triangles to sines rather well, without the shark kinks or pointy waveform tips you get with piecewise-linear or tanh shaping. Apparently this circuit has been widely used in analogue synthesizers, but I have never seen it used in an actual audio function generator.
The circuit was apparently described by Middlebrook & Richer in 1965, and uses one single JFET, two diodes, and four resistors. More here: Tim Stinchcombe - Tri to sine conversion
I haven't built this one, but in LTSpice simulations, I was able to get better results with it than with the other more common methods I tried (piece-wise linear diode-resistor networks with cusp cancellation, differential amp with cusp cancellation). And the circuit is almost ludicrously simple by comparison.
The only significant downside I found is that there is quite a lot of signal loss - the sine wave that comes out is considerably smaller than the triangle that went in. But at least in simulation, the results are very good considering how simple the circuit is.
-Gnobuddy
