MidiDAC acurracy measurement

[Tony Allgood]

>I just measured the voltage for all Cs, to get an impression of how
well the 12 bit DAC tracks.... Maximum error seems to be 0,24 semitones.

Hi Jorgen and all,

0.24 semitones, ie +/-12 cents is a lot and not what I have at all.
Worst case error was found to be +/-2cents. Spec of the DAC puts it at
closer to +/-6cents. However, I have been measuring it on the freq of a
VCO and not on actual voltage. I do not posess a 4.5 digit DVM to try
this out. The expected DC error of 0.5LSB ie. 1.2mV cannot register
accurately on my 3200 count meter. I therefore chose frequency and
measured that instead using a Korg DT-1 chromatic tuner. Although this
will undoubtably have errors in the chromatic scaling, by using the same
note at different octaves should have no effect on accuracy. Drift of
the VCO would also have an effect.

What you may be getting is false readings due to the 31.25KHz clock
pulses that are present on the CV lines. Typical pulses are about
+/-10mV in size when seen on my 100MHz scope. But these are very narrow
and are nicely integrated out by a VCO's expo and associating circuitry.
Measurements on the Kenton Pro-2 and Design Labs Prologue show a similar
degree, or worse in the case of the Prologue, of digital noise on their
CV lines.

Regards,

Tony Allgood  Penrith, Cumbria, UK

Modular synth circuits, TB303 clone and Filter Rack

http://www.techrepairs.freeserve.co.uk/projects.htm