3

E.g. the datasheet of NCU18XH103F6SRB mentions different B values for pairs of temperatures (reference temperature to a higher one). For reference, here the values of this chip:

[T]=degC [R]=kOhm [B]=K
25 10
50 3380
80 3428
85 3434
100 3455

To that I have the following two questions:

  • Are both temperatures defining a range where it is used or is the second/ higher temperature the least error prone value at this higher temperature?
  • For getting maximum accuracy across a big range, I would like to merge all these into a single equation. Right now I am doing this by calculating the R at the higher temperature using the respective B values and then using these temperature-resistance pairs, together with the reference pair, to parametrize the Steinhart-Hart equation with five coefficients. Is there a way giving better accuracy?
toolic
  • 5,637
  • 5
  • 20
  • 33
Torsten Knodt
  • 273
  • 2
  • 8
  • 1
    See this https://www.vishay.com/en/thermistors/ntc-rt-calculator/ – Antonio51 Jun 30 '22 at 21:23
  • @Antonio51 Thanks, but I know how to handle this for a single B value, but I was missing the information how to use multiple B values, as given in the datasheet above, in the best way to optimize accuracy. – Torsten Knodt Jul 01 '22 at 09:05
  • May be of some interest https://www.seventransistorlabs.com/Calc/Thermistor.html – Antonio51 Jul 01 '22 at 09:25
  • And this http://www.patarnott.com/atms360/Arduino/pdf/negativeTemperatureCoefThermistors.pdf – Antonio51 Jul 01 '22 at 09:36

1 Answers1

2

The beta is defined by the thermistor resistance at those two temperature points. If you are using the beta equation you would pick the value from the temperature pair that best fits your temperature range.

There are three coefficients in the Steinhart-Hart equation and you have four pairs so you can solve for least-squares error, for example.

These coefficients are probably close enough given the 1% tolerance of this part:

enter image description here

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • Thanks, but this is one of the reasons for asking. Is this pair then something like a range where the B value is good or is the second value the average in which it is good? And following up to this my idea was to use multiple values of them to further improve my accuracy. – Torsten Knodt Jul 01 '22 at 09:08
  • It's a pair of temperatures from which the beta value is determined. *Nothing* is said about the average over that range, though you might assume that a narrower range has less deviation. – Spehro Pefhany Jul 01 '22 at 10:50