A fast small signal modeling method for GaN HEMTs. (January 2021)

• Record Type: Journal Article
• Authors: Zhao, Ziyue; Lu, Yang; Yi, Chupeng; Chen, Yilin; Cai, Xiaolong; Zhang, Yu; Duan, Xiangyang; Ma, Xiaohua; Hao, Yue
• Is Part Of: Solid-state electronics. Volume 175(2021)
• Publication Date: 2021-01
• Subjects: GaN HEMT -- Small-signal model -- Parasitic inductances -- Extraction method; Semiconductors -- Periodicals; Semiconducteurs -- Périodiques; 621.38152
• DOI: 10.1016/j.sse.2020.107946
• Languages: English
• ISSNs: 0038-1101

Abstract

Highlights: A new method is proposed for the 19-element small signal model. The value of parasitic capacitance and inductance are obtained simultaneously. The influence of Cpga and Cpda on parameter calculation is verified. The optimization algorithm is used to optimize the parameter values. The accurate initial value improves the efficiency of optimization. Abstract: An accurate and efficient parasitic parameter extraction method is proposed for gallium nitride (GaN) high electron-mobility transistors (HEMTs). In this letter, a 19-element small signal equivalent circuit model is established to describe the characteristics of the device precisely. The simulation of the high frequency behavior is improved by introducing the series parasitic inductances into the cold pinch-off model. Through matrix transformation and equation derivation, the influence of Cpda and Cpga on parameter extraction is evaluated. It is proved that the error caused by Cpga and Cpda to the parameter extraction is less than 3%. Ignoring these influences can not only simplify parameter calculation, but also ensure its accuracy. The parasitic capacitances and parasitic inductances can be obtained simultaneously, improving the efficiency and the accuracy of parameter extraction. Meanwhile, accurate and ideal initial values are provided for the subsequent parameter optimization. Because of the accuracy of the initial values, parameter optimization eliminates the error caused by previous steps, and does not need multiple iterations, which improves the efficiency of the model parameter extraction. Compared with the traditional extraction method, the method in this letter can fast and accurately obtain the values of parameters. The simulated results show good agreement of scattering parameters up to 40 GHz.