Calibration Methodology
Reference measurements taken at multiple distinct values across the operating span of an instrument allow for the determination of its characteristic response curve. Implementing multi point calibration compensates for non-linear sensor outputs by establishing several calibration coefficients. This approach provides higher accuracy than simple zero-and-span adjustments.
Mathematical Modeling
Polynomial regression or multi-segment linear interpolation is used to calculate the correction curves from the recorded data points. When applying multi point calibration, the choice of polynomial degree must avoid overfitting, which can introduce artificial oscillations between the calibration points. Generally, a third-order polynomial is sufficient for most piezoresistive pressure sensors.
Metrological Rigour
Thermal chambers and deadweight testers automate the generation of reference conditions for multi point calibration. Each calibration point must reach thermal and pressure equilibrium before the data is recorded. This systematic approach ensures that the measurement uncertainty is minimized and traceable to international standards.
Error Mitigation
Random noise and environmental fluctuations can distort the calibration coefficients if the reference system is unstable. Performing repeated measurements at each point and averaging the results filters out high-frequency noise. This averaging ensures that the calculated correction factors represent the true behavior of the instrument, enhancing the long-term stability of the device calibration under field conditions.
It also allows for the calculation of the standard deviation at each calibration point, providing a clear metric of measurement repeatability.