[sdiy] 2164 Expo correction

[David G. Dixon]

> Without seeing your numbers and doing some calculations I 
> can't be sure, but I think the effect you see might be 
> consistent with a loading of the input resistors by the base 
> currents.  Only for unity gain this effect is compensated by 
> the replication of the input network to tie the inner bases 
> of the input pairs to reference ground.  From your two 
> datapoints the base current should change by ~0.2µA (that's 
> for two transistors) over the range you measured, which is at 
> least a plausible number (depending on what your input 
> current is, of course).  I've not done any further checks...

Very interesting.  I think you're probably right.  If those internal
resistors are fairly consistent across different chips, then my correction
scheme should work well.  It does seem to rely on one particular JFET,
though (4117).