More ramblings about VCDO...

[jbv]

Harry Bissell :

> These are some "generic" thoughts on the VCDO in general...
>
> 1) Point made before... A lot of the sampled waveforms have similar harmonic
> content, they just don't sound different from each other.

Disagreed. When played in steady & continuous form, yes, most waveforms sound
the same, simply because :
1) their are electronically generated signals with features not often found in
natural
sounds (fixed phase & amplitude relationships between partials, etc.)

2) the human ear  - at least until the invention of techno    ;-^    -  is not
designed / trained
for that kind of sounds...

But put in different contextx, and with filter & amplitude enveloppes and
various FX
applied to them, different waveforms might lead to rich & various results...

> 2) You will get some "jitter" in your clock which may or may not be
> objectionable.

Agreed. But what is the typical jitter of a Xtal clock ? How much will it
disturb the counting ?

>
>
> 3) You probably need two counters so that one can be reset while the other
> counts. I don't think you can make do with one. This is like the "finite"
> reset time of an analog sawtooth. Two counters fixes this problem.

Unless you don't reset the counter. If you let it free run and just latch the
current
value, you just need a few more bytes of firmware to compute a simple
substraction
between successive values...
Another way of simplifying the previously described VCDO is to skip the
comparator :
instead of a bipolar output, the VCO should deliver a simple TTL compatible
square
which can be used as a latch pulse. But in that case, you MUST go the hard way,

you can't rely on software and uC onchip comp anymore...

>
>
> 4) Ther solution does not track "realtime". It is always one cycle behind.
> My guess is that without very clever code, you will need to divide the
> incoming wave by 2 to assure symmetry (which I know you say you don't
> need... ) and a consistant edge to work with.
> you will have the "tracking delays" common to a Guitar Synth. Which I can
> assure you are VERY substantial and annoying. The lower the input frequency,
> the worse off you are. Count up the milliseconds and see how far off you'll
> be by being one cycle behind.
>
> 5) The high speed VCO, or frequency multiplication techniques work in the
> very same cycle.

Yes, both points are true. But this would be annoying only when using that
thing
as a pitch tracker. Since its main purpose remains VCDO, the output of the
audio VCO
doesn't actually need to be used as an audio output, only to drive the digital
section.

Among other advantages of this technique (not previously listed), I can see
also :

- the possibility to output various simultaneous waveforms that are perfectly
in
sync,

- the possibility to perform sophisticated interpolation (or less sophisticated
rounding
and truncation). Of course, these options could be switched off for a more
grungy sound...

Note : there was a big mistake in my first post : in case several waveforms are
generated
simultaneously, interpolation HAS to be performed for each WF (only increment
needs
to be calculated only once if functions have the same length).

>
>
> 6) Of course none of this may matter if you are fronting this with MIDI and
> you can compensate for the time problems.

Yep, but then you lose the CV control...

>
>
> It is an interesting thread, ain't it.... :^) Harry

Yes, it made me write two postings on the same day !    ;-)

jbv