RE: Thermistors and Their Resistance Ratings

shabaz — Fri, 29 Jul 2022 13:32:30 GMT

Hi Nico,

Nice work with the charts, it's the best way to learn in this case, getting hands dirty with real measurements and observations : )

The multimeter reading discrepancy might be because of the small amount of current causing the thermistor to heat itself.

Regarding your approach, it sounds good, and exactly as [mention:050549436bd64f50a71d3db7750d1220:e9ed411860ed4f2ba0265705b8793d05] says. Once you have your potential divider, if sized right, then your curve is a lot less curvier, and that reduces the ADC requirements massively. It doesn't matter that it's not precisely linear, because if you have the coefficients, and since you know the constants in your potential divider (i.e. R value and the applied voltage), then you can always calculate back from that to get the temperature.

To calculate the best R, there is perhaps a formula, but I don't know it, and simply trial-and-error can be used, i.e. plot the curve for different values of R, and visually pick the best one. There is an example of different R value curves about halfway down the page of this blog.

If your microcontroller doesn't have the performance to do the calculations, they can be pre-calculated to build up a lookup table, that can be stored in Flash (and any measurement in-between can be interpolated, there are basic and more accurate ways of doing that, but if Flash is plentiful then you can store a larger lookup table).

RE: Thermistors and Their Resistance Ratings

neuromodulator — Thu, 28 Jul 2022 22:42:38 GMT

What you want to do to get the T from the R is to find the Steinhart-hart coefficients. Try to do that empirically, but if you can't just stick to the values that you got from the datasheet. Then to use a divider you may want to test mathematically with the equations and coefficients at hand, with what R values you get the most linear response within the range of T that you will use. Finally since the transformation from the divider V to T is not linear, you may want to derive the equation to perform that conversion. Also, avoid dissipating too much power into the thermistor, because that energy gets converted into heat and might affect the reading.