Calibration and Validation of INA219 as Sensor Power Monitoring System using Linear Regression

Abstract

This paper introduces an innovative approach utilizing the INA219 sensor and ESP8266 for an efficient power monitoring system, complemented by straightforward calibration and validation techniques. Real-time data is seamlessly stored and displayed in Google Sheets through Blynk apps. The system undergoes calibration using fixed DC lamps and resistors as voltage loads, with digital multimeters, oscilloscopes, and power data loggers employed for comparative analysis. Calibration employs the linear regression technique, and accuracy, precision, and uncertainty analyses are determined through Mean Absolute Percent Error (MAPE), Relative Standard Deviation (RSD), and Gaussian distribution. Notably, the load voltage and shunt voltage sensor coefficients of determination (R2) stand at 0.999 and 0.997, with corresponding accuracy rates of 99.27% and 93.71%, precision levels of 99.82% and 99.55%, and uncertainties of 0.37 V and 0.89 mV. The research reveals a noteworthy finding: for achieving accurate current measurements when employing an external shunt resistor smaller than the INA219's internal shunt resistor, calculating the current using Ohm's law, proves more accurate than direct measurement.