The sensors in the Atmocube have automatic self-calibration/cleaning and data normalization algorithms/protocols which ensure long-term stability without the need for any manual action from users. The manufacturer datasheets for all the sensors in Atmocube and detailed specifications can be found at: Sensor Accuracy and Technical Specifications
The CO2 sensor is a patented photoacoustic technology based sensor which has an automatic self-calibration algorithm that assumes that the sensor is exposed to the atmospheric CO2 concentration of 400 ppm at least once per week.
The PM sensor does not require factory calibration because the measurement principle is based on laser scattering. Verification is performed at the stage of manufacturing with a 3% atomized KCI solution (Potassium Chloride). Deviation to a reference instrument is verified in end-tests for every sensor after calibration. The PM sensor has an automatic fan-cleaning procedure which is triggered periodically following a defined cleaning interval. For more details on sensor calibration and long term drift please refer to: PM Sensor Specification Statement
The VOC sensor is a metal oxide (MOx) or chemiresistive sensor that works on the principle of measuring the resistance of VOC gasses present in indoor environments in proportion to the logarithm of the resistance of the MOx material. It has an automatic calibration procedure where it resets the baseline to the cleanest air measured over a period of time. For more details on how the sensor works, please refer to the following resources:
The CH20 sensor is an amperometric electrochemical sensor that works on the principle that formaldehyde molecules pass through a membrane to react with water molecules in the electrolyte to release carbon dioxide and generate an electric current. The concentration of formaldehyde is determined by measuring the magnitude of the electric current. As the sensor is based on a chemical cell, it doesn't require calibration. The electrolyte is very selective to HCHO molecules. A baseline isn’t required as the drift can be neglected. The annual drift is specified in the sensor datasheet - long term drift is <5 ppb or <5% of measured value, whichever is larger - per year.
In addition to built-in sensor algorithms, external normalization and compensation algorithms are used to properly interpret/compensate for readings, depending on the changes in the environment (temperature, humidity, pressure, etc.).