expo mix up

[Tom May]

Magnus Danielson <cfmd at swipnet.se> writes:

> From: "jhaible" <jhaible at debitel.net>
> Subject: Re: expo mix up
> Date: Wed, 6 Oct 1999 12:08:05 +0200
> 
> Hi!
> 
> Another favorite topic just poped up:
> 
> > > >I sink the VCO/VCFs current in the transistor that has the control
> > > voltage applied. This is the same as the Formant and the ASM-1 I think.
> > > But the minimoog VCO uses the second one, with the op-amp's feedback
> > > going to the first.
> > >
> > > Sorry, this is the wrong way round. But the question remains the same.
> > > What advantages are there in doing it either way?
> > 
> > The first difference is the CV polarity, of course. If you have a certain
> > CV polarity (after a certain number of inverting adder stages, for instance)
> > and a given polarity of the expo converter (npn or pnp), you can choose
> > the right input of the converter to avoid an extra inverter stage.
> > 
> > The two inputs have different requirements for the driving impedance,
> > however. The "reference" transistor (in the feedback loop) can work
> > with a medium driving resistance, like the 1k (tempco) of a passive
> > resistor divider. The same resistor divider applied to the other, the "expo"
> > transistor will cause a remarkable HFT error. Why ? The external
> > impedance adds to the internal rB, which is transformed to an increased
> > rE. This will linearize the expo function at higher currents.
> 
> It is noteworthy to point out that the rBE is being closely matched in many
> cases and that the rBE is typical 300 mOhm in a MAT-02. The dual transistor
> configuration will compensate out the rBE of the expo-transistor by using
> its match in the ref-transitor.

Magnus, could you go into a little more detail here?  It seems that
the ref transistor draws come current that loads whatever is driving
its base.  Since the collector current of the ref transistor is fixed,
I'd expect the base current to be fixed except for beta-dependent
temperature variations, whereas the expo transistor base current will
vary with the CV, but it's not clear how rBE fits into this or where
the matching of rBE comes in.

And what exactly do you mean by rBE?  The lowercase "r" suggests a
dynamic (small-signal) resistance, which according to the Ebers-Moll
equations (and the moog ladder) should be a function of the collector
current and therefore fixed for the ref transistor and variable for
the expo transistor.

fTom.