I decided to try active low-pass filtering on this circuit. The first amp operates as a comparator to synthesize the pwm signal. The LM393 would be a good choice. The other one is a low pass filter built around a power amp IC such as a LM3886. I chose a triangle for my input signal to introduce harmonics. In earlier versions of this cct I thought it made the most sense to use a linear sawtooth carrier, vut for reasons completely beyond me, it always resulted in a slanted looking output wave. For some reason, using the triangular exponential waveform seems to correct the problem. As-is this cct still has a lot of problems. It can only "reproduce" waveforms around 1 kHz. As always, any constructive input is sought and appreciated.
-j
