I have always used and recommended the simple silicon diode sensor for measuring temperature. Its worked extremely well for me over the years when used in a range of -65C to +135C. The upper end of your temperature range may kill the diode idea but I'll describe anyway.

I've used the cheap 1N4148 diode. You forward bias it at 1mA via a fairly decent constant current source. Then put the diode at the remote sense point connected up through a shielded twisted pair and brought to your circuit board and inputted into an amplifier. The response of the diode, if I can recall correctly, is about 2.2mV per degree C.

My favorite story about diode temp sensors dates back to about 1975 or so when I was building one of my first commercial devices with a diode temp sensor. We had prepared a characterization study of some diodes to check on linearity over temperature. After careful measurements were tabulated we plotted the data points onto some vellum grid sheets that were the size of a drafting table (huge sheets something like ANSI E size (34" × 44")). It was neat to find the diode forward voltage drop at constant 1mA current to be "linear" within a couple of grid marks from 125C down to -55C using a six foot long steel scale as a reference.

Side Note: I did some thinking about the grid resolution we must have used so long ago. These large vellum sheets had light blue grid marks printed on them that were spaced at 0.1 inches. Across the whole 44" width this would correspond to some 440 grid spaces. When plotting a range from -55C to +125C we needed a total of some 125+55+1 = 181 grids to represent degrees. So it would seem likely that we had plotted the temperature axis as 0.5C per grid on the vellum sheet and this using up 36.2 inches of width for the temperature axis. The diode exhibits about 650 mV drop at 25C room temperature so the measured voltage range we were working with was on the order of 400 mV to 850 mV. The vertical size of our paper had 340 grid lines and I would guess that we had plotted the 450 mV range on the vellum at some 2mV per grid line thus using up about 22.5 inches of height for the vertical voltage axis. If we assume we were actually using a fairly sharp pencil at the time then "a couple grid widths" was equivalent to linearity of a degree C or so over the measurement range.

It is amazing to recall the steel rule across that drafting table almost as if it was yesterday despite it being some 33 years ago!

mkaras