SV: Re: [sdiy] Digital VCO

[karl dalen]

--- Luís Marka <luis.marka at terra.com.br> skrev:

> How did the early DCO-based synths work? The Juno 106 has six DCOs that, 
> by using external integrators, outputs 2 waveforms + sub. All with a 
> 12MHz clock. Don't know about sine, probably an external (analog?) 
> waveshaper Saw -> Tri -> Sin should be used.

Wery simple, most of the so called DCO based units used 8253 counters
(each unit contains 3 16 bit counters) the output square are shaped into a 
short pulse that *pings* a normal VCO integrator and there you have your
ramp! (sawtooth). Then the rest is just a matter of waveshaping into whatever
wave you want. Roland did some special ASIC waveshapers+counter combination 
chip for the Juno series, the 106 might have it to.

KD

> 
> Paul Maddox escreveu:
> > Eric,
> >
> >> There are 20-pin PSoCs available for only few cents more than the 
> >> 8-pin version. If there are enough features to use the extra I/O then 
> >> it would be simple to upgrade to the larger devices. I just set the 
> >> 8-pin chip as a goal to see how much I could squeeze the design down. 
> >> Plus, using an inexpensive 8-pin chip allow you to get a whole lot of 
> >> 'em on one board for that 'cloud of oscillators' effect.
> >
> > hehe, fair comments and an admirable goal!
> >
> >> Yeah - AVRs run a lot faster than PSoCs. The PSoC has a max CPU clock 
> >> of 24MHz (12MHz recommended) and the average instruction takes 6 
> >> cycles - 2MIPs.
> >
> > ouch, that's a big shame..
> > Have you considered over clocking?
> > I've heard of 16Mhz AVRs running at 48Mhz...
> >
> >> Cool idea. I was really more interested in how much sh*t I could 
> >> stuff into one little box. The engineering trade-offs are part of the 
> >> fun. Now if the result ends up being useful too then that's even better.
> >
> > Assuming that if you're going for a 'synth' that a ramp generater 
> > ought to be easy enough to do without causing too much problem. Also 
> > if you're carefull with the code 'switching waves' when you get to a 
> > high frequency could, let you go even higher.
> > For example if, say upto 5Khz you use 256byte samples of waves, you 
> > could switch to a 128byte wave at 5.1Khz, and also halve your control 
> > value (half the number of samples = half the speed required).
> >
> > It'd require a bit of software jiggery pokery, but it's a common 
> > enough trick in DSP (I forget what it's called), though usually you'd 
> > do some cross fading of waveforms, use one with less harmonics and 
> > keep the frequency the same.
> > Then just post filter the wave at your sample rate (62Khz or whatever 
> > you go for) and you should get a pretty reasonable VC-DDS :-)
> >
> > Paul
> >
> >
> >
> 
>