Thermistors are resistors in which the resistance changes with temperature. They are very sensitive and inexpensive, but they require some effort to convert the highly nonlinear resistance value to a temperature value. In addition to the nonlinearity, there can be significant variation between individual thermistors of the same part number. Finally, thermistors are subject to drift over time, especially as a result of exposure to high temperatures and humidity.
This may not be true of all thermistors, but the thermistors studied have an essentially linear relationship between the logarithm of the resistance and the temperature. It would be possible to plug in the parameters for this relationship and do the antilog calculation every time the thermistor is read. A computationally more efficient technique is to generate a lookup table and interpolate to get temperature from resistance.
As part of this project, a spreadsheet was developed that takes two known temperature / resistance pairs (typically from the manufacturer’s data sheet) and generates a lookup table from -40 to +145° C in 5° C increments. This spreadsheet generates a data file that is used by a data acquisition program written for the TS-9700 analog I/O board.
The spreadsheet mentioned above generates a nominal lookup table for a specific thermistor part number. However, individual thermistors will vary in terms of both resistance and sensitivity. The spreadsheet has the capability to calculate resistance gain and offset corrections for individual thermistors. These values are then manually edited into a calibration file used by the data acquisition program mentioned above.
Thermistors can change their resistance over time, usually from exposure to humidity, especially at high temperatures. The least expensive disk-style thermistors are most vulnerable to this effect. Epoxy coated bead thermistors are less sensitive.
In an early attempt at minimizing this effect, epoxy coated bead thermistors were placed in a tube filled with mineral oil that was in turn immersed in a hot water bath. Over the course of a half-dozen tests involving temperatures cycling from near-boiling to room temperature, the room temperature (20°) resistance of the thermistors changed from an initial value of about 54kΩ to about 84kΩ. This is a huge change, and renders much of the data meaningless.
Glass encapsulated thermistors are much more stable. Future tests will be performed with glass encapsulated thermistors. U. S. Sensor makes a line of very reasonably priced glass encapsulated thermistors that are available from Digi-Key. The 50kΩ 5% thermistors (part number 503JG1J, Digi-Key Part Number 615-1019-ND) are being ordered for this project.