maybe this is the solution for constant beat rate detuning ...

[Haible Juergen]

Hi Don, and DIY list,

Maybe you remember the recent threat about linear detuning
in expo VCOs.

One possibillity was injecting a small amount of constant current 
directly into the VCO's capacitor.
Now I really had immense problems to find a cheap way to implement
a current source that goes from 0 ... 10nA with more or less linear
voltage control. Hey, I tried *everything* I could think of ! A weighted
current mirror with a 10meg resistor in one emitter path, an OTA with
10Meg resistor setting the bias current, and an input voltage in the
range of it's offset voltage (needs trimming) ... all not very pleasant.
It's always the same: 10nA cry for a low reference voltage somewhere,
if you want practical resistor values, and with low voltages you always
run into pn junctions that vary to a great degree, compared with your
low reference voltage.

Ok, now there's the idea:

Split the capacitor of the VCO in two: a 1n and a 100n in series.
feed your expo control current int the low end (i.e. to both caps),
and the offset current to the connection between the caps. So
you have a capacitive current divider.
Maybe (?) I also need a second discharge transistor over the 100n,
but this dosn't have to be discharged that fast, just prevent it from
accumulating charge from the unipolar offset current.

What do you think?
Will this work?
Will it affect the VCO's precision and tracking in any way, when the
offset current is set to 0 ?

Any other (better) method I have missed?

JH.