I am curious as to why one couldn't simply use a transimpedance amplifier
and put the thermistor between a reference voltage and the input to the
amp? Providing the sensor itself is linear,the output of this should be
linear,too.

Sean


On Thu, 2 Dec 1999, William K. Borsum wrote:

> Beg Borrow or Steal a copy of the YSI catalog--all you ever wanted to know
> is in there--including a way to use two thermistors (e.g. 6K and 30K) to
> get very linear outputs.  Really does work--we have an old signal
> conditioner product designed around this.  Called "Thermlinear" by YSI.
> They even sell matched pairs  with three leads.
>
> I think I've still got a few of the boards floating around--would be
> willing to sell, along with the schematics, for some nominal price each.
> Boards are about 2x3 inches and designed to plug into a passive back-plane
> or ribbon cable via a pair of 10-pin male headers.  I'd have to go look,
> but I recall each card had two circuits on it.  Really a neat system once
> you understand how it works, and figure out how to calculate the resistor
> values needed in the conditioner circuit.  Designed for accuracies of 0.1%
> to .005% depending on the linear range selected.
>
> Only real draw-back is setting up the resistors--you need to be able to
> measure resistance to at least 0.05%--.01% or better is preferred, and have
> a wide enough selection to pull values off the shelf for each
> conditioner--otherwise its calculate, order, wait, measure, order, wait,
> etc. Boards are through-hole, and use standard 1/4 watt metal film
> resistors--usually read illy available.
>
> In our case, we needed a linear output around 39.5 DegC +/- 3 degrees, and
> ended up with a signal of 1 volt per degC, and were effectively measuring
> .001 degree resolution with a target 0.05% accuracy..
>
> The system could be simplified quite a bit if lower accuracies are needed.
> Remember, most off the shelf thermistors are around 1% or worse anyway.
>
> Kelly
>
>
> At 12:55 PM 12/2/99 -0500, you wrote:
> >Not thaaaat simple. Resultant voltage in a resistor divider is not
> >thaaaat linear.
> >I made a computer program to search for the best *pseudo-linear* piece
> >of curve for a particular temperature range you need, based on a
> >resistor division combination. I will dig it and email to you directly.
> >Wagner
> >
> >> Ercan Duran wrote:
> >>
> >> Hello.
> >>
> >> I'm trying to use a NTC thermistor with a 16F877 device.
> >> It's a 10K thermistor B Constant 25/50 C(K) = 4100.
> >> If I read resistant of the NTC, how can I calculate temperature in
> >> Celcius degree.?
> >>
> >> Thanks in advance.
> >>
> >> Ercan DURAN
> >
> >
> William K. Borsum, P.E. -- OEM Dataloggers and Instrumentation Systems
> <borsum@dascor.com> & <www.dascor.com>San Diego, California, USA
>