Calibration

Verifying the accuracy of Thermocouples and RTDs is a difficult but exact science. It requires a system that has a stable temperature source, an accurate reference thermometer, repeatable measurement and control and finally a data processor. Each component of the system must be in concert with the other components in order to minimize system uncertainty. The components must have corresponding supportive characteristics for resolution, accuracy, linearity, traceability, stability and repeatability. Examples of how these specifications can affect system uncertainty are:

Resolution and Accuracy
If desired accuracy is .01 degrees C then the resolution or ability to read this accuracy must be at least .001 degree C.

Linearity
It is tempting to state linear accuracy at one temperature (usually 0 degrees C), while this is helpful (all thermocouples have zero output at this temperature) it is important to know the measurement accuracy over the entire range of the readout. If the readout were perfectly linear, its accuracy specification would be the same across its entire range. However, all readout devices have some non-linearity component and are not perfectly linear

Stability
Readout stability is important, since most measurements are made in a wide variety of ambient conditions and over varying lengths of time. Consequently the temperature coefficient and long-term stability specifications are extremely important.

Calibration
Beware of “no calibration” claims. Latest ISO specifications require calibration at least once annually.

Traceability
A manufacturers’ calibration tolerance is at best at a secondary standard level. It is important for the reference thermometer to have NIST traceable accuracy.

Now lets’ put this all together. It all starts with the ability to maintain the desired calibration temperature. In order to provide the optimum stability two sources are necessary. This is due to the broad range of temperatures involved. In general if a RTD is being tested a temperature bath would be used. If the sensor is a thermocouple with a higher temperature range a furnace is used. Depending upon the source employed the stability ranges from .0001 degrees C to .5 degrees C. Achieving this type of stability requires a highly stable control sensing element, fast maximum stability, and a source design that minimizes and controls heat loss.

Since our calibration procedure employs the comparison method, the need for a highly accurate reference thermometer is essential. Our system uses a Standard Platinum Resistance Thermometer (SPRT) with accuracies of better than + or - .006 degrees C at 0 degrees C. This accuracy is achieved by abiding by the International Temperature Scale – ITS-90. The SPRT has accuracies traceable to NIST.

The final element, the processor allows the information to be formatted into a user defined report and can analytically address the tolerance and accuracy of the sensor. A good example is the Callendar Van Duesen (CVD) equation. The system uses CVD equations and applies associated uncertainties of a Platinum Resistance Detector over any point within its operating temperature range. The result is a report that provides a resistance limit of error function.The practical uses of this report are many, but one of the most useful is determining sensor resistance interchangeability as a function of temperature. Simply stated this allows the user to determine uncertainty within a predetermined range and correct for the error in the instrument.

Our ability to put together the high tech components that have the characteristics necessary to assure uncertainty data is an important element in our success. The system is fully integrated into our quality assurance program and is a testament to our motto…Temperature measurement…the right way!