More ramblings about VCDO...

[mbartkow at ET.PUT.Poznan.PL]

Tim Ressel <Tim_R1 at verifone.com> wrote:

> I have considered taking a Xilinx FPGA and creating a basic NCO
> (Numerically-Controlled Oscillator). It would clock at about 2-5 MHz and put out
> phase information i.e. a sawtooth wave. Put that phase info into a D/A and you
> have a ramp. Add some XOR gates inside the FPGA and it puts out triangle. A PWM
> is easily done with a digital magnitude comparator inside the chip. All we need
> is a sine wave, which can be accomplished several ways. You would need a small
> processor to take up MIDI and analog signals and set the NCO accordingly. 
> 
> This approach offers several advantages. Locking oscillators is a snap (all run
> from the same clock), phase offsets and phase modulation can be done with
> precision. NO drift. Synch is pretty much a thing of the past, or to look at it
> another way, synch is handled by the processor which handles the NCO.

Tim,

The above "advantages" are in fact true disadvantages if you cannot switch them
off. What you described is a typical DCO featured in many second generation
hybrid synths (KORG Poly 61, Roland Alpha Junos etc) which sound tiny and weak.
Digital perfect stability is a no no. What everybody wants now are the little
instabilities which true VCOs offer.

I think there is a very nice way to introduce some imprefectness without affecting 
the global stability, especially when a bunch of DCOs is considered (as in a poly-
phonic synth).

Make each synth voice out of at least two (three ? four ?) DCOs. Each DCO group
(I mean the oscilators A in all voices form one group, oscilators B is a next one
and so on) is driven from a different master clock. Now, make the clock voltage 
controlled in some narrow range (varicaps come to mind, since we operate in the 
MHz range).
Now the trick I learned from Analogue Heaven (there is some interesting stuff
posted from time to time, BTW): get a white noise source and bandpass filter it
about 30Hz. Use this filtered noise to slightly detune each DCO group. Mind you,
each grup has to be detuned differently, so you need few different noise 
sources. A pair of evenly tuned DCOs sounds extremely fat and analoguish
when slightly modulated by two noise signals.

regards,

mb


--

Maciej Bartkowiak, PhD
========================================================================
Institute of Electronics and Telecommunication     fax: (+48 61) 8782572
Poznan University of Technology          phone: (+48 61) 8791016 int.171
Piotrowo 3A                             email: mbartkow at et.put.poznan.pl
60-965 Poznan POLAND               http://www.et.put.poznan.pl/~mbartkow
========================================================================