[PolyModular]

[Martin Czech]

> How about using a Time-Division multiplexed system? That way only one conductor
> required via front panel and so no DIN socket system. Each module input would
> then have an overhead of a demultiplexer (4051) , n sampling caps and n buffers
> (where n is the number of channels). And each output would have an overhead one
> multiplexer (4051). Also, a master counter and high frequency clock (20KHz x
> n)would be required to go to each module. That way also, a pot could attenuate
> the whole TDM signal which would effect every channel and even VCAs.  
> 
> Ed.

This sounds good in theory.  

But what could happen as well?

I fear that this would be the most effective way to swamp the system
with hf dirt all over.  This perhaps doesn't matter if you use analog
recording (?), or in the case of bad speakers (like mine), but in case
of ADC conversion this would give a hard time for the input guard filter.

Could well be that you have squeeking noise all over.
This is especially "nice" if you pitch such a sample down...

Of course, the higher the frequency, the better you can filter it out.
OTOH switching speed and resistance and parasitics set a limit for that.
And this was all linear thinking. It could turn out that nonlinear effects
create sidebands and difference bands that creep back into the audio
range. Would be very hard to get rid of them.

Such remaining noise is very good audible, even if it is -70dB down,
it's simply there. A sharp spectrum line should not be confused
with wite noise. The first is much more objectionable.



m.c.