# possible tempco with KTY81

prophet5 prophet5 at box.tin.it
Wed Nov 17 01:11:09 CET 1999

```I have not jet tried myself but this is my opinion about temperature
compensation and KTY81 :

The standard input stage of the exponential converter is

R2       Rt
__/\/\/\___/\/\/\__             Rt = KTY81 temp.sensor
|                   |
R1       | |\                |
Vc____/\/\/\_____|_|-\_______________|  Vc'
__|+/               |
| |/  Op. Amp.      /
|                   \ R4
---                  /
///                  |___ Vi
|
/
Vi = -Vc R2 + Rt     R3              \ R3
------- * -------           /
R1     R3 + R4           |
---
///

For KTY81 :         Rt = 800  ohm    @ 0  C = 273 K
Rt = 1200 ohm    @ 50 C = 323 K

If we assume a linear relationship beetween 273 and 323 degrees Kelwin :

Rt = Rt0 + A*T = 8*T - 1384 ohm   (T in deg. Kelvin)

We want for 1V / octave :

R2 + Rt     R3        q
------- * ------- * ----- = ln 2
R1      R3 + R4   K * T

The R2 + Rt0 term causes an error but if we chose R2 = - Rt0 = 1384 ohm :

R2 + Rt0 + A*T      R3        q         A      R3       q
--------------- * ------- * -----  =  ---- * ------- * --- = ln 2
R1          R3 + R4   K * T      R1    R3 + R4    K

were q / k = 11604 and A = 8.

The dependence from T is eliminated !

Now we can assume :

R1 = 2K2 (set the Op.Amp gain to 1 circa)
R3 = 1K

This jelds R4 = 59.87 R3 ==> R4 = 56K + 10Ktrimmer

Is all correct or i'we misunderstanding something ?

```