AW: Jitter in VCOs

[Haible Juergen]

	>This is kind of like the discussion last month about bringing
trim pots out
	>to the front panel. It was pointed out that the downside to
such an
	>approach is that you don't have a stable reference for your
instrument and
	>it makes it harder to get it to behave predictably. 

I don't think it's really the same thing. A synth with trimmers on the
front
might be good for experimental stuff, but a terrible mess if you want to
play "ordinary" stuff.
The imperfect VCO topic is a different beast, really. 

	>I would think that
	>instead of trying to build in imperfect VCO, it might be more
varsatile to
	>build a "perfect" VCO and then do things like modulate the
frequency or
	>trigger refernce voltage with a tiny bit of noise source or
(something
	>similar) when you want an "imperfect but rich" VCO.

Ok, but we should not end up at perfect-as-DCOs oscillators, and
then re-introduce the underdone "random" functions we all know from
digital rompler synths ...

We have better, more subtle possibilities to randomize perfect
VCOs, than you have in digital, so it might be not as bad.

There is one thing which I immediately tried, when I understood that
the noisy zener in the EMS VCOs must have some effect: I went to
my modular and modulated the VCOs with a tiny amount of noise.
There *was* some effect, *but* it was hard to find the right level of
noise, and more important, I didn't have that much noise sources
in my modular system !
Routing the same noise to several VCOs is definetly not the same
as several non-correlated noise sources.

With noisy zener diodes, or  noisy opamps, you have lots of indpendent
random sources.

One thought occured to me during this thread, and this might not be
so different to your approach, and that is making the noise path in an
expo converter an AC instead of a DC path, so it can easily be switched 
on and off.

So it I want to have a VCO like the Synthi A's, I'd use a low-noise
reference
(or a different topology) instead of the reverse biased transistor, but
I'd build
the same part in parallel *with* the original element (i.e. a reverse
biased transistor
with a similar current flowing thru it), then buffer this one, and
couple it to the
"improoved" VCO with a capacitor and a switch.

Or take a normal expo converter with an opamp adder and approx. 50k / 1k

resistor divider to the base of the transistor. The noise of your opamp
will be
divided down as well. Then start experimenting with capacitors across
the
50k resistor, so the noise finds an AC path to the expo converter ...

BTW, both options (noisy opamp and zener in EMS) should have a different
effect: The EMS has linear FM with noise, the opamp noise causes expo
FM.
Don't ask me, how large the difference would be, however.

Coupling caps should be so large, that you get a reasonable amount of
low frequency
1/f noise, mind you.

Hope this starts some wild experimenting ...

JH.
 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 3015 bytes
Desc: not available
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/19970717/d0965633/attachment.bin>