Graphite Based Flexible Thermocouple Array for PCB Thermal Monitoring

This paper describes the development and validation of an experimental, cost-effective, and flexible thermocouple crossbar array enabled with the help of deep learning. The thermocouple is constructed using graphite traces on cellulose paper, with individual graphite thermocouples exhibiting a thermopower of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$4\mu \mathrm{V}/^{\circ}$</tex> C. Integrating a deep learning-based regression model enables the crossbar array to achieve promising performance in localized temperature measurements, with an RMS temperature error close to 0.17ºC Additionally, the potential of this novel sensor array to streamline sensor circuitry is discussed alongside its application in PCB temperature monitoring and mapping. Results demonstrate its capability to accurately predict temperature variations among different PCB components. The focus on affordability, flexibility, and disposability positions these scalable crossbar arrays as potential solutions for real-time thermal mapping across various industries.