An Efficient Neural Network Macro-model for Electromagnetic Radiation Spurious Emission from Voltage-Variable Capacitors

Voltage-variable capacitor is an essential part of the matching element in RF circuits such as transmitting and receiving channel in the mobile devices. However, the nonlinearity of the voltage-variable capacitor produces unwanted harmonics with a large exciting signal, which generates the electromagnetic radiation spurious emission (RSE) problem. This paper presents an efficient nonlinear capacitance power series model method coupling with harmonics test systems to predict the nonlinear behavior. In addition, a simple neural network model is trained to reduce manual operation of optimizing the nonlinear parameters. As a consequence, the nonlinear parameters are obtained quickly, and provide the intuitive guideline in designing phase.