We have many datasets with thermocouple readings. The thermocouple is type T, and the temperatures to which they are exposed are in the range of 15 to 90°C, so well within the temperature range for this type of thermocouple (this should limit calibration drift).
We always specify class I, so the TC should be within 0.5 °C deviation from the IEC tables.
However, if I have a measurement with a typical setup (one particular TC connected with a fixed data-acquisition system), and I measure a temperature increase of let's say 25 °C over a year, can I state that the error on this temperature increase is less than 0.5 °C? In other words, can I remove the systematic error (which I assume is constant) from my readings, and rather use the repeatability? And does anybody have knowledge of such repeatability errors?
The thermocouples are installed in the field and are unreachable - therefore cannot be retrieved for recalibration, nor can another sensor be installed next to it for installation reasons.
Dear Jan Verstricht , I would say that if you know the exact systematic error for a TC the you can subtract, but what you have is a statement that the "static error" is within 0.5 deg from "ideal" curve.
If a TC does not have static error (lucky TC), by "subtracting" that 0.5 deg it will shift to the edge of acceptability.
In addition, you do not have distinction between +0.5 deg and -0.5 deg: it is just a measure of maximum deviation, whatever the sign.
So, in my humble opinion, unless they are individually calibrated, there is no benefit in compensating numerically for this worst case deviation taken as the real systematic error.
What is by the way the target uncertainty and performance of your readings?
Dear Jan Verstricht, systematic error can always be considered to be subtracted from the measured values. Actually the calibration data which you have is for the thermocouple but not valid with custom added data acquisition system. As output of thermocouple is low level the measurement is affected by DAQ accuracy. So the calibration of system must have been included full measurement system. Once it's accuracy is ascertained it may be considered as the final value throughout the experimentation.
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity