THERMISTOR TEMPERATURE MEASUREMENT (C, F)
APPLICATIONS > TEMPERATURE MEASUREMENT >

Measurement Accuracy Specifications

Hardware and Software Setup

Fig 3.16 - Temperature Measurement (Thermistor)  Voltage Divider Circuit

Thermistors are two wire devices whose resistance varies with temperature in a known fashion, often accurate to +/- 0.2°C.  The instruNet thermistor measurement feature supports ysi-Omega 100ohm to 1Mohm thermistors between the temperatures of -80°C to 250°C.  When Thermistor is selected in the Sensor popup, instruNet assumes a thermistor is connected in the prescribed fashion, and subsequently returns the thermistor temperature in degrees C units after applying a Steinhart & Hart resistance-to-temperature conversion.  Measuring a thermistor temperature involves a voltage divider circuit with a shunt resistor of known value, applying a voltage across the pair and measuring the voltage across the thermistor, as illustrated above.  The voltage across the thermistor is measured between a pair of instruNet Vin+ and Vin- input terminals while the excitation voltage for the divider is supplied by the instruNet Vout terminal.  instruNet calculates the thermistor resistance using the following equation.

RTherm(Ohms) = Rshunt * (Vin+ - Vin-) /  (Vout - (Vin+ - Vin-))

TemperatureTherm(°C) = -273.15 + 1.0 / (a + b (Ln( RTherm )+c (Ln( RTherm ))³)

a, b, and c are a function of 3 points in the resistance-to-temperature table, and are calculated by instruNet after the user completes a short dialog box interview.  To minimize "self heating", it is recommended that thermistors operate at less than 100uW (50uW is better).  An example would be a 2252 ohm thermistor (which will vary from 394.5 to 7355 ohms in the 0-70°C temperature range).  With 0.37V excitation voltage (i.e. Vout) and a 1000 ohm shunt resistor, the current and power dissipation at 70°C would be 100uW:

Current (Amps) = Volts / Resistance = 0.37V / [1000 + 394.5] = .265mA

Power (watts) = Current 2 * Resistance  = .0002652 * 1394.5 = 100uW

The voltage across the thermistor would vary from 104mV to 320mV as the resistance across the thermistor varied from 394 to 7355 ohms; therefore, an input Voltage Range of ±.6V would work nicely in this case.  Shunt resistors with an initial accuracy of .025%, and a temperature drift of 20ppm/°C, such as the Caddock part #TN130-resistance-0.025%-20, are recommended 1, 3, 6,

To do temperature measurement using a Thermistor, the user must:

1.   Set the Sensor field in the Hardware settings area to Thermistor.  This will cause a series of dialog boxes to appear, asking the user several questions about the thermistor type (i.e. resistance at 25°C), shunt resistor value, and excitation voltage.  It also recommends an excitation voltage and shunt resistor value based on the thermistor in use, and the temperature range of interest -- in many cases, using the recommended values are the best options.  Based on the responses to these questions, instruNet loads the following fields in the Constants setting group, with the following information:

• Ro - thermistor resistance at 25°C, in units of ohms
• Rshunt - actual shunt resistance, in units of ohms
• Vout - excitation voltage output the Vout screw terminal
• alpha- the 'a' Steinhart & Hart coefficient
• delta,Rlead - the 'b' Steinhart & Hart coefficient
• GF - the 'c' Steinhart & Hart coefficient
• V_poisson -  the maximum expected temperature, in degrees C units
• Vinit - the minimum expected temperature, in degrees C units

instruNet also sets the Voltage Range field in the Hardware settings area to a value appropriate to the specified temperature range.  The smaller the temperature range, the better the accuracy; therefore one should not make the specified range unnecessarily large.

If you want to run through the dialog box interview again, select Voltage in the Sensor field, and then select Thermistor, to invoke the interview again.

If you want to manually set any of the fields in the Constants settings group, do the interview, and then set them to your liking, after selecting Constants in the Settings popup.

2.  Wire your thermistor per figure 3.16, click here if you need more guidance setting up the software, and click here if the measured value is not correct.  To reduce noise, 0.001 seconds of integration is often helpful (i.e. set the Integrate field in the Hardware setting area to 0.001) 5, 10.

Theory of Operation

Thermistor Measurement Errors
The maximum thermistor measurement error is the sum of the following component errors:

1. Errors within the thermistor itself (typically +/-.1C to +/-.2C). For more details, please consult your thermistor supplier.
2. Thermistor linearization error: +/- .05C.
3. Shunt Resistor Accuracy Error: This is dependent on the type of shunt resistor in use, the specific thermister in use, and the temperature one operates at. For example, if you use a Caddock #TN130-1K-0.025%-20 shunt resistor, then it's initial accuracy is 0.025%, which causes a .025% error in the thermister resistance measured by instruNet. For example, a 2252ohm thermister at 0C sports a resistance of 7355, and this value multiply by 1.00025 is 7356.8, which corresponds to a .005C error. Please consult your thermister resistance-to-temperature tables when making these calculations. In most cases, a .025% shunt resistor with a 20ppm temperature coefficient provides less than a .01C error. For a list of resistor suppliers, click here.
4. Thermistor voltage measurement errors: These are due to the voltage measurement accuracy of instruNet 100 itself. With .001 sec of integration, one can expect instruNet 100 measurment errors of +/- 700uV, 75uV, 15uV, 10uV on the +/- 5V, .6V, 80mV, 10mV ranges, respectively (for the instruNet 100 Rev 3). instruNet measures two voltages, the voltage divider excitiation voltage, and the voltage across the thermister itself (both introduce a small error). The thermistor resistance is calculated as follows:

   Rthermistor = Rshunt * Vthermistor / (Vexcitation-Vthermistor)

   where:

Rthermistor = calculated thermistor resistance in ohms
Rshunt = shunt resistance resistance in ohms
Vthermistor = voltage measured across the thermistor
Vexcitation = excitation voltage across the voltage divider

5. Shunt resistor self-heating.

   The power disipated by a resistor (shunt resistor or thermistor) is:

   PowerDisipated (Watts) = (VoltsAcrossRes * VoltsAcrossRes / Resistance).

   This power causes the resistor to heat up:

   TempChange (C) = ThermalResistance (C/Watt) * PowerDisipated (Watts)

   And this causes a change in resistance. For a Resistor:

ChangeInRes (ohms) = TempChange (C) * TempCoeff (ppm/C) * ResValue (ohms)

6. Thermistor thermal self-heating.

   This is calculated in the same manner as the Shunt resistor self-heating above in (5); however, it is not necessary to calculate the ChangeInResistance in the end, since the 2nd equation already provides the change in temperature, which is the error of the thermistor heating up on it's own due to current passing through it. In most cases, one should put <100uW (75uW is better) through a thermister to keep it's thermal heating <.07C.

Example Maximum Error Calculation
Suppose we use a 2252 thermistor with a 1000ohm shunt resistor on a 0-70C range with .37V of excitation. instruNet would measure the excitation voltage on the +/-.6V range accurate to +/-75uV, resulting in a 75e-6V/.37V = .02% error, which corresponds to approximately .01C. At 25C, the thermistor resistance would be 2252ohms, and the voltage across the thermistor would be .37V * 2252ohms / (2252ohms + 1000ohms) = .25V. This would also be measured on the .6V range, which would induce a similar error of around .02C. At 70C, the thermistor resistance would be 394.5ohms, and the voltage across the thermistor would be .37V * 394.5ohms / (394.5ohms + 1000ohms) = .104V. This would also be measured on the .6V range accurate to +/- 75uV, resulting in a 75e-6V/.104 = .07% error, and .07% of 394.5 is 0.2ohms, which corresponds to approximately .02C. The thermal heating of the shunt resistor would be the worst when the thermistor was at it's maximum resistance of 7300ohms at 0C (.35V across the shunt). At this point, the power disipated across the resistor would be .1mWatt, and at 116C/Watt thermal heating, this would result in a .01C rise in shunt temperature. And with a 20ppm/C temperature coefficient, this would correspond to a .2e-3ohm change in resistance, which would result in a .2e-3/1000=.00002% error (not much). The error from the self-heating of the thermistor itself would be be tiny as well at 1.5e-5 degrees C. Notice that the excitation voltage, the shunt resistor resistance, and the measured temperature range all effect the accuracy calculation.