how to pick thermistor based on data points?

From the numbers above it looks like a thermistor specified as R25= 1.9k and Beta = 4152 (average).

R25 is the resistance at 25C. The nearest standard R25 value is 2k.

--
Tony Williams.

acannell asked,

Thermistors are usually specified for their resistance at 25C, which is 77F, so you need to know your part's resistance at that temperature. Common thermistors drop about 4% per degree C (and this agrees with your 180F measurement), so your thermistor's resistance should be about 1870 ohms at 25C. Hmm, that's not near a common part value. For example, 2252 ohms at 25C is a common value, which would show about 9F too low on your gauge. 2.0k is another common value, this would only be 3F low at 80F. It's possible your gauge is no more accurate than that anyway.

Converting those to Celsius, that corresponds to something like: R25=1900 ohm Beta=4233

The (simplified) formula for the resistance of the NTC thermistor is then: R=R25*exp(beta*(1/T0-1/T)), where all temperatures are in Kelvin (Celsius + 273.15).

greetings, Tom

You've already got a couple of good answers.

Note that negative temperature coefficinet thermistors can be reasonably accurately modelled by the Steinhart-Hart three-paramter formula

Betatherm makes "interchangeable" thermistors and Farnell stock their

+/-0.2C parts, including one that offers a 2.252k (+/- 0.8%) resistance at 25C. They aren't cheap. Cheaper parts come in at +/-5% and 10% tolerance on resitance, and a temperature guage designed to use one of them would include some means of calibrating out this tolerance.

There are other sources for interchangeable thermistors - Thermometrics and Yellow Springs Instruments (YSI) come to mind. Newark used to stock a +/-0.5C YSI part (at ten times the price of the Betatherm 0.2C units) and may still do.

-- Bill Sloman, Nijmegen