[sdiy] Temperature stable lin-exp converter

[René Schmitz]

Hi Jay et al. 

At 16:17 30.05.02 -0700, Jay Schwichtenberg wrote:
>So to me it looks like you need to multiply Vin by (k * T) / q  instead of
>adding.

Right.

>What I don't know is (k * T) / q inverted from what you'd get from the sense
>transistor?

If you keep the collector current fixed you get a b-e voltage proportional
to kT/q.

>I have a couple of thoughts on how to do this divider thing. One is to look
>into an OTA in the feedback path of the expo summming amp being controlled
>by q / (k * T). The other is to use an OTA as the summing amp taking the
>compensation and run it into Id (diode bias in LM13X00, CA3280). There has
>been talk about using Id to make a divider but it isn't speced well for the
>LM13X00 and CA3280.

The OTA will have a temperature dependance on its own, which either needs
to be 
corrected, or can be used directly instead of the sense transistor. There
is a circuit 
at my website which does that. (see VCO4, and VCO2 for explanations) 

A seperate sense transistor and a linearized OTA are the better solution
linearity wise. 
I think a linearized OTA (LM13x00 input diodes) looses its temperature
dependance.
The problem will probably be to fine trim such a circuit.

Cheers,
 René
-- 
uzs159 at uni-bonn.de
http://www.uni-bonn.de/~uzs159