pwm ...

[Magnus Danielson]

> 
> > Why use D/A-converters for CV-control ? A 100 kHz variable pulse-width 
> signal
> > (or even better a tri-state signal = double resolution) could be easily
> > generated with software.
> 
> Do you know what precision (jitter!) you'd need top get a 7 or 8 bit
> accurarcy ?? And what amount of HF this would produce ?
> 8-bit DACs are cheap, so why not use them?

Agree, and you can expand each 8-bit DAC with some simple CMOS/Cap/Buffers
circuitry. Useign this technique you could even spend money on more precision
if you want, cause the price per output tap will not increese that much anyway.

Also, maintaining say 32 outputs with 8 bit precision will require quite some
CPU time that could be used better. The bittwiddeling needed may not seem that
hard, but when doing it efficiently for many outputs you suddenly find that the
generic CPU doesn't apply well to the problem. You will end up burning power
and money for a bad solution....

Besides, the tri-state output will NOT double the precision, you will get 3 
steps instead of the 4 steps needed for _double_ precision. You will need to
use a tri-state gate for each output. You can get true double precision by
replaceing the tri-state gate with two resistors. Then you effectively get a
2-bit DAC.... why not go all the way to a 8 bit DAC rigth away?

Magnus