A Non-Contact Testing Method for a Switch-mode Power Transformer in an SMPS using Radial Basis Function Neural Network to Analyze Magnetic Flux Leakage Density

Switch Mode Power Supplies are known to be significant in electronic devices mainly in desktop computers which uses an ATX power supply. Such power supplies are difficult to troubleshoot since it is composed of multiple sets of components that purposely supply the right amount of regulated voltage. Switch mode power supply units became more complex having topologies that allowed different components to play specific roles and thus are executed together with proper assembly and quality inspection. This study offers a non-contact testing method to classify defective and non-defective ATX power supply units with the use of its switch mode power transformer. The magnetic flux leakage density emitted by the transformer is analyzed and compared to the frequency and voltage peak from the transformer which is obtained through contact testing. Its peak magnetic flux leakage behavior is then analyzed through the voltage emitted by the sensor used in this study serving it as an input to the Radial Basis Function Neural Network that is responsible for defectivity classification. This study has then obtained an accuracy of 66.67%.