[sdiy] Patchell's Boggling 13700 Expo Converter Thing???

[Ian Fritz]

Magnus --

At 04:05 AM 6/22/2003, Magnus Danielson wrote:

>Ehum! May I point out that non-linearities on _linearized_ OTA inputs is
>greater than they possibly could be since the common approximative curcuit
>leaves much to desire. The linearized OTA must be driven by a diffrential
>current source with a stable common mode current in order to operate 
>correctly.
>The normal fashion of doing things is by tossing in a few resistors, which is
>not the best way to acheive the diffrential current source.

Yes, this is well known, and we understand it.  But how do you 
actually  implement the current-source driver?  Do you have a working 
circuit? It seems to me you would need to use something fairly complicated, 
like a Howland circuit.  And you would still be relying on cancelation 
between the linearizing diodes and the signal transistors.

>Jim's design does take another interesting approach, but it may not be the 
>only
>approach to get there. I think again habbit works over analysis mode.

Well, then, you will need to show us a better working circuit with 
analysis.  Have you analyzed Jim's approach?  It may have an advantage in 
that it's the signal transistor response in the two units that 
cancel.  There is a minor approximation in the analysis, but the linearity 
is very good and the circuit is quite simple.

   Ian