Mathematical Formulation
A multi-dimensional array of coefficients acts as a correction model to adjust sensor output based on multiple environmental inputs. Employing polynomial offset compensation matrices allows micro-controllers to correct for complex, non-linear sensor drift caused by temperature or pressure changes. In silicon pressure transmitters, this method corrects the raw output signal in real-time by calculating the correction term from current operating conditions.
Calibration Execution
Matrix coefficients are calculated by fitting calibration data collected at several temperature and pressure points to a multi-variable polynomial equation. In production facilities, automated calibration systems run the sensors through thermal cycles while logging the exact reference values. This logged data is then processed by a least-squares algorithm to generate the unique matrix coefficients for each individual sensor.
Accuracy Improvement
Sensor sensitivity and offset vary from lot to lot, which requires that each device receives its own unique array of coefficients. This custom tuning eliminates systematic errors and improves the total accuracy of the sensor across its entire rated operating envelope.
Computational Constraint
Limited non-volatile memory in small embedded systems constrains the maximum degree of the polynomial used for compensation. A third-order or fourth-order polynomial is selected to strike a balance between processing load and correction performance. System developers benchmark the execution time of the matrix calculations to ensure the sensor can maintain its required update rate.