Detect Data Deviation for Temperature and Ambient Light Sensors, and Create a Simple Calibration Method

Abstract

Listen

Translate article

We noticed that sensors output can have incorrect data output. This document describes how we can calibrate the temperature and ambient light sensors connected to a Raspberry Pi using data from other microcontrollers or sensors. For more accurate readings, we can apply some techniques to calibrate the sensors data output using simple algorithms. An important note is that this technique can be used only if we do not have any professional tools. We will determine how temperature sensors can be influenced by heat.To measure temperature, we used sensors from Bosch Sensortech, Aosong Electronics, Sensirion and Winsen Electronics. For ambient light we used sensors from Maxim Integrated, Rohm, Sensortek, Texas instruments, Taos and Vishay.The sensors are connected to multiple Raspberry Pi modules and use the power from the main board. The Winsen sensors use a serial connection, and they are the only ones that uses power from a USB port. We read data from each sensor every 120 seconds and store it in a JSON file. Our main concern is data accuracy. We have multiple sensors from same manufacturer, and we noticed that same model can record different values in similar situations. We will see if the ambient light conditions are influencing the readings or not even in case of temperature sensors.We can read values close to real ones by creating some baselines, but we need to study how the temperature is increasing and decreasing based on amount of heat generated by the module. In the end, we will be able to find out minimum requirements for a basic algorithm that can predict or calculate the temperature or ambient light intensity.