Wide bandgap-HEMT device (GaN) modelling for high power amplifier design

Abstract

Material properties of wide-bandgap semiconductors are excellent candidates to build highly-efficient and highly-linear power amplifiers required to support cellular communication. AlGaN/GaN HEMTs are considered the most capable of all available wide-bandgap devices, as they combine material properties of GaN with the principle of HEMTs. This dissertation presents a large-signal modelling strategy for small-sized (4*100 μm) AlGaN/GaN HEMT that is capable of being ported to computer-aided design of power amplifiers. A large signal model capable of simulating the output power and non-linear behavior of the device is very crucial for the design of a Power Amplifier. This is the main problem addressed by this thesis. Large signal modelling begins with the development of a linear model. A bottom-up empirical modelling approach is followed in this work. It required selecting an appropriate electrical equivalent circuit which accounted for the complex parasitic and maintained a clear physical interpretation of the model parameters. Further, an efficient algorithm for extracting and optimizing parasitic is adopted. This has set the foundation for non-linear modelling. The large-signal model includes a non-quasi static formulation of the gate-charge and a dispersive-drain current (Ids) model. Non-quasi static parameters are calculated from small-signal intrinsic parameters by using path-integrals. Various mathematical interpolation techniques are applied while doing integration taking care that the parameters do not lose their physical significance. The Ids-model parameter extraction is based on DC-IV and pulsed-IV measurements. Large-signal models are developed and directly implemented in CAD-software to perform model simulations. The results of the simulations performed are compared with the measured data. The results show high correlation between them making it a competent model for High Power Amplifier design for small-sized AlGaN/GaN devices.