Sat Nov 6 03:05:22 CET 1999

```Martin, Rene, et al. --

I'm not following this argument. The term being compensated is
exp(-V/kT). Since V is proportional to the tempco R, we need R
proportional to T. This is a linear T dependence of R. It is not correct
to say that the tempco should be the same as the tempco of 1/T. The two
temperature effects are not being subtracted, they are being ratioed.

Ian

Rene Schmitz wrote:
>
> Hi Martin and DIYers!
>
> >We could adapt the additonal series R so that it fits the tempco
> >at eg. 30C.  This means 1321Ohm series resistor and gives the needed
> >3361ppm @30C.  The original tempco is scaled with 0.4308 by this, so we
> >get a new tempco table for KTY81 + series R 1321Ohm, and compare that
> >to the theoretical ideal needed tempco:
> >
> >Tj C  tempco   Ohm (typ) needed tempco
> >------------------------------------------------
> >
> >  0   3662    2136       3723.7
> > 10   3576    2189       3594.3
> > 20   3446    2282       3473.7
> > 25   3403    2321
> > 30   3360    2361       3361.1
> > 40   3231    2443       3255.7
> > 50   3145    2530       3156.8
> > 60   3059    2620       3063.9
> > 70   2973    2713       2976.4
> > 80   2886    2811       2893.8
> > 90   2800    2912       2815.8
> >100   2714    3017       2742.0
> >
> >We can see that for an ideal Transistor pair, the nonlinear tempco of
> >the KTY series is not harmfull at all, it helps. It gives better
> >compensation then a wirewound linear tempco.
>
> I'm speechless! I didn't analyze this so deep. Many thanks Martin, for
> doing the math for me. It matches to my experimental findings, my VCO3
> which uses these, is quite stable. (Although I have to admit that the
> scheme I use has an opamp output offset drift problem. But this has nothing
> to do with the KTY in general.)
> Seems as if I have to change the description on my website.
> Until now I assumed that the required tempco was linear, so I thouhgt that
> the KTY would be inferior to a wirewound tempco. Your figures tell me that
> I'm wrong.
>
> The KTYs come in different resistor values the KTY 81-1xx is 1k, the KTY
> 81-2xx is 2k.
> And they're available with different tolerances: -110 means 1k, +/-1%; -120
> means 1k, +/-2%. One could try if a 2k KTY with a parallel resistor gives
> (even) better results.
>
> >Of course there are the typical problems: Ib is not zero, Ics variation,
> >matching, self heating, thermal gradients. It is not clear to me in the
> >moment if these effects are not stronger then anything else considered
> >here.
>
> Well here my argument is that it is the same regardless of the type of
> tempco used.
> There is not much that you can do against these errors, exept using good
> pairs, and thermal coupling of the resistor to the transistors. The KTY
> comes in a TO-92, and can hence be easily glued to something like a 2SC1583
> or a minidip.
> Btw, I've put up a page which compares Rbe for various transistor pairs,
> I've measured these for a bunch of devices. The link is on the startpage of
> my site at uni-bonn.
>
> Bye
>  René
>
> error:asciiartisnotfully  \ uzs159 at uni-bonn.de
> supportedwithyourcurrent   \ http://www.uni-bonn.de/~uzs159
> displaydriversettings...    \ http://members.xoom.com/Rene_Schmitz
>

```