Ideal VCO for mass production

[Mikko Helin]

>Which brings me to the this question: what is the first module that i should
>release?
>I chose the VCO, because it is a fundamental sound source which people can
>start off with. Filters can be used to filter other signals, but i though,
>from my own
>point of view, a good VCO should be the first thing i should produce. If you
>think
>wrong, please let me know.... so i ask this question: what is the best/ideal
>VCO
>for mass production?

VCO is OK, but you need at least a PSU before it can run.

>
>Who here has a fantastic design that fits the needs and requirements of an
>e-musician? I am happy to pay royalties, and fully acknowledge the person
>responsible for the design. ASM-1, or Tom's VCO4b? Tell me... i want to
>do something about it.... this time without the slanging match...
>

I haven't followed this thread too long. The VCO in ASM-1 is basically
the same as in Electronotes (ENS-76 option 1, the option 2 had
CA3046 type exponential converter vs. AD818 pair in option 1), except
that it lacks the saw2tri2sine waveshapers. That circuit has propably
also used in some commercial synth. Don't know anything about
Tom's VCO4b.   Problem with analog oscillators is their temperature 
dependency, for which there aren't too many solutions (temperature
 compensation and heater circuits). 

So it would be easier to design a digital oscillator. That hadn't 
be expensive, as the prices of A/D and D/A convertors are pretty low now. 
The VCO could be built around SX-28 (Scenix 50 MHz PIC compatible
microcontroller) with 16 bit (is that enough?) ADC to read the control 
voltage and >= 16 bit DAC to generate the sound. Some people would
like to drop the DAC and use pulse width modulation, but DAC's
are now about $1.50 a piece (I mean 24-bit DAC's like CS4334  from 
Crystal Semiconductors). Of course there's the programming part,
with SX-28 you must use 8-bit bytes to calculate the phase
of wave format (frequency), and if you use 32 bit phase accumulation
register it means you have to use four bytes and sum these one by one 
and add carry flag etc. then take the 16 -24 MSB's and output
these to DAC one by one, generate clock signals, and everything
must happen just at right time, so the machine code has to
timed pretty precisely. Other than that (the time the programming takes)
I can't see any particular problems. Adding new code (waveforms
and like) is also possible as SX-28 (and it's 18-pin little brother
SX-18) support ISP (in-system programming) with two osc & data
pins & GND.

Just some ideas,

-Mikko

 

>
>L8r,
>
>bud / FTS
>www.ozemail.com.au/~budweiser
>
>