Nonlinear AlGaN/GaN HEMT model using multiple artificial neural networks

In this work, a complete nonlinear-transistor-model extraction-method is described. As a case study, the AlGaN/GaN High Electron Mobility Transistor manufactured on SiC substrate is modeled. The parasitic components model is proposed, and its extraction results are presented. Low- and high-frequency large-signal measurement data are involved in order to produce multiple artificial neural networks. The network topologies of multilayer perceptron networks are established automatically. A complete learning procedure using back propagation algorithm is described. A good agreement between the measurement data and the model has been observed.

[1]  P. Colantonio,et al.  Advanced Neural Network Techniques for GaN-HEMT Dynamic Behavior Characterization , 2006, 2006 European Microwave Integrated Circuits Conference.

[2]  Christopher M. Bishop,et al.  Neural networks for pattern recognition , 1995 .

[3]  David E. Root,et al.  Technology Independent Large Signal Non Quasi-Static FET Models by Direct Construction from Automatically Characterized Device Data , 1991, 1991 21st European Microwave Conference.

[4]  A. Caddemi,et al.  A New Millimeter-Wave Small-Signal Modeling Approach for pHEMTs Accounting for the Output Conductance Time Delay , 2008, IEEE Transactions on Microwave Theory and Techniques.

[5]  J. Zolper,et al.  Wide bandgap semiconductor microwave technologies: from promise to practice , 1999, International Electron Devices Meeting 1999. Technical Digest (Cat. No.99CH36318).

[6]  S. A. Maas,et al.  Modeling the gate I/V characteristic of a GaAs MESFET for Volterra-series analysis , 1989 .

[7]  A. Raffo,et al.  Temperature dependent vector large-signal measurements , 2011, 2011 Workshop on Integrated Nonlinear Microwave and Millimetre-Wave Circuits.

[8]  I. Angelov,et al.  Extensions of the Chalmers nonlinear HEMT and MESFET model , 1996 .

[9]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[10]  D. Schreurs,et al.  Nonlinear Dispersive Modeling of Electron Devices Oriented to GaN Power Amplifier Design , 2010, IEEE Transactions on Microwave Theory and Techniques.

[11]  Babak Hassibi,et al.  Second Order Derivatives for Network Pruning: Optimal Brain Surgeon , 1992, NIPS.