possible tempco with KTY81

[Martin Czech]

Is it was previously discussed, all boils down to:

R(T0)+C1*(T-T0)
---------------
       T

ie.: one can never eliminate T, no matter what type of simple circuit
is used, no matter how expensive the tempcos are.

One has to do a fit for a certain temperature T0 such that the error is
zero at T0 and small in the environment. The error will get larger if
the actual temp T is far away from T0.

I still believe that the considerable nonlinaerity of
semiconductor tempcos (KT8x) can be used in voltage dividers
to have a better fit then that of a perfectly linear tempco.


m.c.