[sdiy] VCS3 oscillators

[harrybissell]

The schematic of the 3080 shows Darlington transistors for the PNP
mirrors... a sure sign that the PNP gain is low (ie lateral PNP structure)

Who knows for the 3280... which does not show the details of the mirrors.
Anyone have that info ???

I'm plating with a spice model (potential) for the 3280... assuming that it
is
similar to the 3080 in mirror construction. So far it does look like it is
behaving
something like the real thing

H^) harry

Scott Bernardi wrote:

> I'm pretty sure all the OTA chips are made with lateral pnps, although
> they may be large ones optimized for current handling capability.  I'm
> sure the 3080 uses a standard analog (i.e. opamp) manufacturing process.
> The Beta of lateral pnp transistors drops the more current you put
> through them. The Wilson current mirror configuration used in the OTAs
> has a built in mismatch of approximately 1/Beta^2.  So the mismatch is
> going to get worse when you are running higher control currents - one of
> the causes of control voltage feedthrough. And Beta  is _very_ process
> dependent. And that's why you would have to go through and "cherry pick"
> 3080's for minimum control feedthrough.
> My guess is that they learned from the experience and really put some
> effort into optimizing the matching and current handling capability of
> the pnps in the CA3280, which is why it has superior control feedthrough
> specs.
> The best conceivable OTA you could make would be with dielectrically
> isolated pnps like in the THAT transistor arrays - where the pnp
> transistors have specs comparable to the npns.
>
> James Patchell wrote:
>
> > You will need to be careful with the CA3096.  The PNP transistors are
> > lateral PNP's and have very poor characteristics (very low Beta, for
> > example).  You may not find that the circuit functions the same as it
> > would with discrete parts.  One way to solve this would be to use the
> > THAT340 transistor array which has two PNP and two NPN transistors,
> > and they are all vertical BJTs...very good performance.
> >
> > But, the THAT340 is pricey.  (~$4).
> >
> > At 11:25 AM 9/30/2004 +0200, Yves Usson wrote:
> >
> >> Hello (2nd post)
> >>
> >> My questions are aimed at those among you who built a VCS3 or SynthiA
> >> clone
> >> or who are lucky enough to own one of these beasts.
> >>
> >> Does the shape control of osc2 and osc3 acts on the shape of the ramp
> >> signal or does it act only on the pulse shape ?
> >>
> >> What I understood from the schematic of the oscillator 1 (that can
> >> be found on internet) and from the reading of Jurgen Haible's messages
> >> posted some years ago on this list, the shape control of osc1 only
> >> changes the shape of the sine signal and not of the ramp signal.
> >>
> >> I want to build a clone of the VCS3 and I intend to use CA3046 and
> >> CA3096
> >> transistor arrays in order to reduce the number of parts and simplify
> >> the
> >> design of the PCB.
> >> Do you consider this would be a complete heresy or an acceptable
> >> compromise
> >> that would preserve the original sound ?
> >>
> >> Thanks for your comments.
> >>
> >> Yves
> >>
> >> Yves Usson
> >> email : yves.usson at imag.fr
> >> http://www-timc.imag.fr/Yves.Usson/personnel/SDIY
> >
> >
> >         -Jim
> > ***************************************************************
> > http://www.oldcrows.net/~patchell
> >
> > ***************************************************************
> >
> >
> >