Digital Dynamics

Fred this thread has got me thinking.  The whole DSP & DAC thing exists as a general purpose solution.  You run the DSP sampling a low F as you can to get the most out of multiplexing the ALU hardware (but this complicates the design of digital filters running at higher frequencies).  Then you stick a (what used to be complicated & expensive) DAC on the output and call it a day.  This works fine for music reproduction and such, but isn't all that necessary for synth waveforms which are constructed on the fly, likely with separate envelopes.  Coupling them together externally via low resolution PWM multiplying constructs seems really interesting and could be what I've been looking for for years.  I'm going to have to try this out!  I've tried straight PWM and it's just too difficult to do modulators higher than order 2.  But two simpler modulators in an analog multiplying arrangement just might work.  Bravo!

 Dewster - Well have a CMI theremin yet! ;-) Good luck!

One other little tip about using the 4053..

There is some crosstalk between the switching signal and the signal path, and this can be a bother - it can cause the envelope to not quite go to 0V (it depends on the manufacturer of the 4053 and on the speed of the switching signals - I found that the +ve pulses (or rather, edges - the mechanism is, I think, primarily capacitive) coupled to a greater extent, so when integrated tend to give a small unwanted +ve output)

 - I put a pair of parrallel signal diodes (A-K, K-A) to force a vdrop, somewhere between the analogue envelope and the waveform switch to overcome this problem.. the waveform will go silent when the envelope drops somewhere below the diode VF - If I remember correctly, usually when the envelope drops to about 200mV.. The envelope buffer needs to be R-R if one wants simplicity, and one can of course bias this so it eliminates residual voltage - probably a better solution.. I didnt have R-R Opamps when I played with this.. we didnt  have any of the stuff we can buy for peanuts today... LM358's were the standard LV Opamp.. 

The crosstalk (or more correctly, the control signal injection) is quite constant, and proportional to the PWM, so doesnt affect the waveform to any noticable extent, as its integrated away - mostly - it may add a slight DC bias to the waveform output but this is no bother.

Thanks for those hints Fred!  It's got me looking at 2nd order modulators again (for the PWM) - this time with 3 level outputs.