Started 15th April 2018
One of my Oregon Scientific TH122N wireless sensors began to show temperatures several degrees high. Past experience pointed to the thermistor having begun to fail.
Negative Temperator Coefficient (NTC) thermistors can be characterised by their resistance at a given temperature and their β value (see Wikipedia). I had an idea the Oregon thermistor had a resistance of 10KΩ at 25 °C and so bought a bag of thermistors off ebay with this specification; the β value of these was 3950. Fitting one it became apparent that whilst the resistance was correct, the β value was not (temperatures were not reported correctly away from 25 °C). Thermistors with lower β are not easy to find on ebay. I set out to see if adding a series and a parallel resistor could let me reproduce a lower β thermistor.
It turns out this can be done, adding a resistor in series of 720 Ω and a resistor in parallel (to both the thermistor and the series resistor) of 102 KΩ allows the β 3950 thermistor to be used to replace the Oregon one.
The first graph shows experimental data for resistance against temperature for the TH122N. I got this by replacing the thermistor with a resistance box and reading the temperature returned for each resistance value. The second graph shows sensor data and the result of fiddling with the values in the thermistor equation, from this I determine the correct value of β to be 3100 with a resistance of 9700 Ω at 25 °C. The final graph shows the sensor data and the thermistor equation for a β 3950 thermistor, both with and without the additional resistors.
How I fitted the resistors (slight postscript, I had to go back and solder the push fit connectors).
Similar material from someone else:
Wireless Sensor Thermistor Calibration