Multi-frequency Nonlinear Network Model Of High Power Semiconductor Devices

With the advent of third-generation semiconductor Ga N and the development of high-frequency,high-power environments in the communications electronics industry,the nonlinearity of devices has become increasingly unnegligible.Establishing accurate high-power semiconductor device models has a good predictive function for designing modern power amplifiers and is also very important for the design of new devices.Traditional S-parameter modeling is based on linear networks,powerless in modeling of large-signal devices.Therefore,it is very urgent and important to find a network model that can accurately characterize the characteristics of devices under large signals,which is also the research goal of this thesis.To compensate for the shortcomings of S-parameters that cannot characterize large-signal devices,X-parameter theory was proposed.X-parameter is a superset of S-parameter and is a good description of the nonlinear characteristics of the device.However,domestic research on X-parameter modeling is relatively late,most of which is based on the study of X-parameter behavioral model theory and the improvement of its algorithm.Thus,it is meaningful that X-parameter behavior model is established based on the actual working characteristics of the device and applied to the actual Circuit design.In this thesis,the principle of X-parameter model and the formula derivation are given in detail.According to the actual experimental conditions,the independent X-parameter and load-related X-parameters are extracted from the Ga N HEMT device of model CGH40010 F by the circuit design software ADS.Based on the load-related X-parameters,an X-parameter model is established.After that,a power amplifier with a center frequency of 3.5 GHz and a single-tube output power of 10 W is designed.By comparing with the power amplifier with the one designed by the circuit model,it can be seen that the power amplifier designed by X-parameter model has better suppression of harmonic power.Combined with artificial neural network,the input signal power,frequency and bias voltage are used as the network input and the extracted independent X-parameter are used as the network output.X-parameter model based on RBF neural network is established by MATLAB and the accuracy of the model is verified.By comparing the error values of the model output data with the actual test data,the accuracy of the model built is demonstrated.The error value does not exceed ±0.1at most.The error value of more than 97% does not exceed ±0.05,and the error value of more than 92.6% does not exceed ±0.03.X-parameter model based on RBF neural network predicts a set of X-parameter data.A power amplifier with a center frequency of 3.7 GHz is designed by ADS and the established model is applied to the actual circuit design.