Hybrid Small-Signal Modeling of GaN HEMT Enhanced by the Integration of SVD and RIME Optimization

Abstract

For the 20-element small-signal model of GaN HEMT, a combination of the singular value decomposition (SVD) algorithm and the frost ice optimization algorithm is proposed in this paper to extract and optimize the intrinsic parameters of the small-signal model. When the traditional algorithm is employed for parameter extraction, issues of low extraction accuracy and efficiency are encountered. By introducing an optimization algorithm for parameter extraction, the accuracy and efficiency of the process are enhanced. However, previous studies have focused on improving the optimization algorithm to optimize the eigenparameters of GaN HEMT without taking into account the correlation among the parameters within the model. In this study, the SVD algorithm is utilized to process the real and imaginary parts of the intrinsic model Y-parameters, thereby strengthening the correlation between the intrinsic parameters. Subsequently, the new intrinsic model Y-parameters and the RIME algorithm are employed to extract the intrinsic parameters. The experimental results demonstrate that the combination of the SVD algorithm and the frost ice optimization algorithm breaks the isolation between the eigenparameters, improves the parameter correlation, and can accurately extract and optimize the eigenparameters of the small-signal model within the frequency range of 0.5 - 20.5 GHz.