VCDO - asking again + new idea

[jbv]

Don Tillman :

> Yeah, this was discussed not too long ago.  Here's my contribution:
>

(SNIP)

> But Magnus' suggestion got me thinking... you could also but together
> something with a regular audio-range sawtooth VCO, a counter connected

> to a DAC, and a comparator circuit.  The sawtooth reset pulse zeros
> the counter, and as the sawtooth ramps up you compare the sawtooth
> voltage with the DAC output, when the sawtooth voltage is delta-V
> above the DAC you clock the counter.  Pretty clean, and you get a
> simultaneous analog sawtooth.

I didn't get any response, so I'm asking again : did anybody try to
implement that ?

In the meantime, I gave that thing a second thought and came to the
conclusion that it would involve
fast comparators with precision calibration that would be hard to
achieve. So, after some more
brainstorming, came the following idea.

Let's say we have an audio range VCO producing a positive sawtooth, with
a freq range from 0 to
20 KHz. As it ramps up periodically, the sawtooth is sampled through a
fast ADC, the 6 MSB
from the ADC are used to adress a ROM where various waveforms are stored
(64 bytes for each).
Of course, the most significant adress bits of the ROM are used for
waveform selection. Then the data
output of the ROM are sent to a fast DAC.
In order to still have 64 samples for the highest freq (20 KHz), the
sawtooth must be sampled at
1.28 MHz min. In the Farnell catalog I found a 6 bits ADC (CA3306CE by
Harris) with a typical
conversion time of 0.1 us. Then we need a fast 8 bits DAC, and there are
plenty of those.

Yes, I know it's not the cheapest solution for a VCDO, but my estimation
is that it should come for
less than $30.00. The other interesting thing is that a regular VCO can
be used.

But now, let's try to think further : instead of a fixed periodic
sawtooth, what it we feed that thing
with another kind of signal (tri, sine, PWM square...) or even natural
sounds (strings, guitar,
woodwind or human voice) ? What will happen to the signal ? IMHO, the
result should be somewhere
between distortion, waveshaping, quantizer and/or freq
multiplier/divider. Of course, we leave the
strict VCO domain and we enter the "effect" domain.

And what if we increase complexity : instead of sending the data outputs
of the ROM to a DAC, what
if we use them to adress another ROM with sampled waveforms, and then
convert the outputs of that
2nd ROM into audio signal ? Would it just increase the noise and lead to
smthing dull, or would some
interesting cahotic effects appear ?

Any thought on the above ? Has anyone already played with smthing
similar ? Am I re-inventing the
wheel, or what ?

jbv