[sdiy] sequential multitrack schematics.

[Tim Parkhurst]

On 9/28/06, Rainer Buchty <rainer at buchty.net> wrote:
> >You can't use a saw or triangle... it would discrete switch them
> >anyway, internally.  If the DC points of each wave were different, the
> >result could be nasty.
>
> If I read the CEM3396 datasheet correctly, you need to keep track of
> frequency, pulse width, and selected waveforms. Hence, modulating Vwf
> would give somewhat unexpected results anyway.
>
> If you change the frequency, you will alter the waveform w/o also
> adjusting Vwf. Similarly, you need to adjust the pulse width when
> changing waveforms.
>
> I wonder if there have been any non-uC-based synths using the CEM3396?
>
> Rainer
>

Hey Rainer,

Remember that we're talking about the 3394 (with built-in triangle
VCO) and NOT the 3396 (no built-in VCO, takes a square wave and
reshapes it). The 3394 has a hard switch between tri / saw / tri-saw
mix waves, determined by a voltage level. The 3396 will allow you to
morph (under voltage control) between a tri and saw waves. This is the
chip used in the Oberheim M6 and M1000 synths. I've done the saw - tri
mod with an LFO on my M6 and it sounds okay. Nothing Earth-shattering,
but it sounds okay. The square to saw/tri is a neat trick, and is
bascially a voltage-controlled integrator (a CV compensation is
required to keep the waveshapes correct across a wide frequency
range).

And yes, I guess you could run a 3396 without a micro, although I
would think a micro could map any non-linearities in the waveshape or
VCF frequency across the audio range (probably a big part of the
auto-tune routine). Oh, and don't forget that the 3396 needs a square
wave input for it's audio signal (no built-in VCO). I guess you would
have to try one out to see if there are any big "gotchas" in the 3396
response curves. Could be a fun project.


Tim (a small "gotcha") Servo
-- 
"Imagination is more important than knowledge." - Albert Einstein