[sdiy] Cheap Midi->CV

[Simon Brouwer]

Hi Andre, all,

At 00:01 8-9-2005, you wrote:
>On 2005-09-07 00:29 +0200, Simon Brouwer wrote:
> > Hi Michael,
> >
> > At 21:12 6-9-2005, you wrote:
> > > (...)
> > > I like the "hit the bench" mentality - it's so DIY! If you
> > > can live with 12 bits check out Mr. Maddox and his now GPL
> > > PolyDAC.
> >
> > I had thought that 12 bits would be plenty.
> >
> > The AD8300 that Paul uses has a maximum integral nonlinearity error of
> > 2 LSB. Add to this a quantization error of 0.5 LSB. With a full scale
> > of 10 V (10 octaves) that would amount to an error voltage of ca.
> > 0.0061V, or 0.073 of a semitone interval. With a full scale of 5 or 6
> > octaves the error voltage would be correspondingly smaller.
> >
> > Do you think having more pitch accuracy than that would be worthwile?
>
>I don't need a frequency counter to know when a guitar string is
>7.3 cents off. So yes, probably.

Sure you can hear it while tuning your instrument (if my calculation is 
correct, at  e.g. 440Hz, a difference of 0.073 of a semitone would give a 
beat frequency of about 2 Hz). I wonder if you would notice the difference 
while playing, though, especially between different notes?

A couple of years ago I built a MIDI2CV using an 8 bit DAC, using 4LSB 
steps per semitone, for a range of 64 semitones (more than 5 octaves). I 
found this quite usable, not really out of tune.
If I remember correctly, this DAC (a ZN426) had accuracy specified at 
better than 0.5 LSB which would be a factor 2 worse than a 12 bit DAC with 
2 LSB accuracy if used for a range of 10 octaves.

So I was feeling quite comfortable choosing a 12 bit DAC for my next 
project. I am not so sure about that now...

Would the accuracy of VCO's not be the limiting factor?


Vriendelijke groet,
Simon Brouwer.

 >>> nl.openoffice.org <<<