Real-Time Adaptive NFC Charging System with Neural Network-Based Impedance Matching

This paper presents an adaptive near-field communication (NFC) wireless charging system designed for consumer electronics, utilizing printed ISO/IEC 14443 Class 1 coils on a cost-effective FR4 substrate. The proposed system incorporates a neural network-based adaptive impedance matching network operating at 13.56 MHz. Performance analysis reveals that in practical setups, coil and component losses significantly reduce maximum achievable power transfer efficiency, limiting it to around 41 %, compared to previously reported ideal efficiencies of 63-92 %. The study also highlights that in many scenarios, losses from adaptive impedance matching hardware exceed the potential benefits. To address this, a bypass functionality is introduced to enhance system performance under these conditions. Future developments should focus on integrating the adaptive matching network on-chip, exploring solutions such as capacitor banks, or transmission lines for impedance matching to further improve NFC charging efficiency.