SV: Re: SV: Re: [sdiy] Roland DCOs

karl dalen dalenkarl at
Wed Feb 27 00:28:30 CET 2008

--- Tom Wiltshire <tom at> skrev:

> On 26 Feb 2008, at 15:44, karl dalen wrote:
> >
> >> The other thing to watch out for on these DCOs is that the pulse  
> >> width
> >> of the pulse output is fixed. Not fixed ratio, like say 20%, but a  
> >> fixed
> >> time. Thus the harmonic structure changes as you go down the  
> >> keyboard.
> >> The JX10 does not a powerful bass monster make. :-)
> >
> > ?
> > The PW are fixed yes but its also fixed over the entire frequency  
> > range.
> > So there are no change in the spectrum as if it whould have been if it
> > its PW was changed over freq! However the PW can be changed over time
> > and or freq but the higher the freq the less bits of PW are available
> > making the chip completely useless as DCO **WITH** variable PW.
> This doesn't have to be a problem. It looks like the JX-3P and 8P do  
> use the counter to generate the pulse waves directly (at least, I  
> can't see any comparators) and would therefore suffer as you describe  
> from decreasing resolution.

The 82c54 are a tricky beast, many strange pulse modes within.

> However, the Juno 106 uses a comparator  
> fed with the Ramp waveform and a PWM CV in the traditional manner,  
> which avoids the problem completely. It depends whether you want PWM  
> as a feature of the chip or a part of the waveshaper stage following it.

I never said it was a problem, i merely described some of the behaviour
of the 82C54 IC. Yes it would suffer terrible.

However if you read the JX3 manual you will find out that PWM are not
generated by the method i described above, nor by a comparator as in JUNO106
its generated by changing phase ie fine tune of DCO2 by comparing phase
positions of the oscillators trough the nand gate.Its one of those old
dirty trix to get PWM, similarly dirty trix are two saw waves one inverted. 

Btw, im a little bit unsure what Tony actually meant by fixed PW/time!?

> > In other words current errors in ordinary VCO designs give you
> > freq errors will give you spectral errors in DCO, all you do is
> > to translate one error source into another error domain.
> I agree with Rainer about this. Moving errors into domains where they  
> are less perceptible to humans is one of the techniques for dealing  
> with them. The human ear is more sensitive to frequency than to  
> amplitude or harmonic spectrum, so the errors in the DCO are in the  
> 'right' places.

Yes off course but i only described the translation of the errors
betwen regular closed loop VCO design and open loop design as in DCOs.
However suprices are awaiting you around the corner. ;-)

> If you ensure that the compensation CV tends to be on the low side  
> rather then the high side, then the Ramp wave will always be a  
> perfect ramp, but it's amplitude might vary a bit. Still, at least  
> you've eliminated any harmonic effects.

Im was talking about triangle and other low content harmonic waves,
when it comes to the saw you can actually chop off 50% of the saw, make it
completely flat and it will still sound like a saw.Tinyiest spike or
flatness in the wrong place in a triangle or sine will be very audible.
> That said, you're right that getting a decent amplitude compensation  
> CV over a full MIDI range does present some challenges.

Peanuts, just make sure your using an accurate 12bitter DAC
and 2 scale resistors. Check out CEM3xxx data papers you will
find all info you ever need in there.


Sök efter kärleken!
Hitta din tvillingsjäl på Yahoo! Dejting:;185753627;24584539;x?

More information about the Synth-diy mailing list