<![CDATA[Smoke detection in enclosed spaces is important for early fire detection and air-quality monitoring. This paper characterizes and compares the performance of three MQ-series gas sensors (MQ-2, MQ-7 and MQ-135) under an Arduino UNO R3-based system. Three smoke sources (paper combustion, cigarette smoke, and motorcycle exhaust) were evaluated against a commercial CO detector. Parameters assessed include linearity, sensitivity, response time, error, measurement uncertainty (ISO GUM), and output-voltage correction. MQ-2 and MQ-7 exhibit logarithmic voltage–concentration relationships consistent with the power-law model of SnO₂-based metal-oxide-semiconductor sensors, while MQ-135 exhibits a quadratic pattern; all R² values exceed 0.98. MQ-7 produced the lowest errors among the three sensors for paper combustion and cigarette smoke, indicating suitability for indicative early-warning detection; however, its best-case relative error of 38% confirms it is unsuitable for precise quantitative CO monitoring without additional calibration. The three sensors exhibited errors exceeding 85% for motorcycle exhaust due to cross-sensitivity resulting from the complex chemical composition of engine exhaust. A voltage-correction analysis showed the errors in Arduino ADC readings were less than 6%, confirming the reliability of the electronic system; this does not mean the accuracy of the gas concentration will be the same, since it depends on the selectivity of the sensors, calibration and environmental conditions.]]>
Copyrights © 2026
