The method I am trying will make their response linear over the entire range,

What is the basis for that statement?
I pointed out that the quasi-log curve of a thermistor does not accurately follow the log curve of a transdiode.

 

Have it your way. Go find your old fart. We are supplying you with modern old farts that do it in code in a minute instead of breadboards.

 

There is some old fart in his eighties out there that could breadboard this up in a minute.

Well, I'm an old fart-- in my 70's, not 80's, but still old-- and am quite familiar with log/anti-log amplifiers, having designed and used them many times in the past. And yes, I could breadboard that circuit in a jiffy. But I won't. Why? BECAUSE IT WON'T DO WHAT YOU WANT IT TO DO.

For precisely the reasons @crutschow cited in post #8, using a transdiode or log amp will not-- repeat, WILL NOT-- give you a linearized output from a thermistor.

Deal with it.

 

I agree that it is important to share our concerns about the approach he is considering, but I feel that once those concerns are understood its up to studiousscholar to decide the most appropriate design direction. After all, he or she is the one that will have to implement the design and live with the consequences.

I just want to throw the old diode breakpoint approximation method into the pot. Remember how hot the ICL8038 function generator chip used to run? A good deal of that was the diode breakpoint approximation circuit that turned the triangle wave into a low distortion sine wave.

The circuit below is only to illustrate the concept. It was used for gamma correction in the days when TV cameras used vacuum image pickup tubes and it worked beautifully on video signals of about 1 volt when made with a monlithic transistor array. I have also done this with small signal diodes like the 1N916 instead of transistors and saw beautiful results.

Just a thought.