williB said:
fabbie said:
"You would have to calibrate it against a known accurate device. "
Do u happen to know any accurate device i can use? Any examples?
thats easy ---
you take ice measure its voltage output (0 centegrade)
boil water ( 100 centegrade)
then like Nigel said you have to ' interpolate ' the results ..usually the graph of the in between results come with the part..
I'm afraid you are being fed inaccurate advice!!
A thermistor is nowhere near linear and interpolation is useless over 100 deg span. It is accurate for most purposes with around a 5 deg span between data points.
Nor would you ever need to measure a bunch of data points for a thermistor against a known temp. Their response is a fixed formula (I do not have it handy, look it up). Now the data sheet will only guarantee its constants to 2%, 5%, 10%, whatever. If you want to be more accurate than the tolerance, you can measure a single temp, figure out the exact resistance coefficient for that device, and you know what you need to know to calculate temp quite accurately via interpolation with a lookup chart or a nonlinear calc.
The LM35 and others ARE linear, thus much easier to use. They do have significant inaccuracies (calibration compensated for it). I also noted none of the ones I saw handled cold temps below freezing.
There's the DS18S20 temp sensor, has a wide temp range, very accurate, already digital. If you've got a microcontroller, it's a tempting option.