AW: expo-converter

[jhaible]

>     Sounds something like I might be doing.  I am trying to use the Vbe of
the
> reference transistor as the thermometor (since the collector current is
> constant, Vbe should be linear with temperature) and using that voltage to
> control an OTA that is in a feedback loop of opamp to cancel out the scale
> drift.
>
>     -Jim

This is more or less what I was proposing some time ago. Only that the
differential
pair at the input of the OTA (in opamp feedback loop) would do the
compensation
all by itself. Rene then went on one step further and used a differential
pair (no more
OTA) to to the same trick - the advantage was that the transistors could be
all on
one array.
Disadvantage of both methods is the nonlinearity of the OTA (or noise, if
you're
in the linear region).

A linearized OTA would do the trick. No more opamp tempco now, so back to
temperature sensing the expo pair and multiplying. OTA still in the feedback
loop
of an opamp.

Now, why not use the OTA straight forward (no feedback). Multiplication done
with I_diode rather than I_abc. The circuit I'm proposing is at
http://home.debitel.net/user/jhaible/jh_temp_expo.html ,
but mind you, it's *still untested*, so don't blame me if it doesn't work.
If someone is faster than me on going to the breadboard (which is likely),
feedback is higly welcome !

JH.