Extraction of multifrequency multibias currents and charges of a HEMT using the finite-difference method

Novel tools and techniques for extracting nonlinear large-signal characteristics from small-signal characteristics of a 300mum PHEMT are presented. The large-signal characteristics---multifrequency multibias currents and charges---were generated from two-port S-parameters measured using a Vector Network Analyzer. The Poisson equation was used to make the currents and charges compatible with nonconservative small-signal data. The Poisson equation was solved for currents and charges using the conventional finite-difference method. A suite of MathematicaRTM functions and definitions has been created for importing measurements, smoothing data arrayed in one and two dimensions, extracting the extrinsic circuit, and applying the finite-difference method. The quasidynamic concept has been proposed as a framework for describing the device's frequency dispersive phenomena and time delay effects and that these effects can be explained by frequency dependent currents and charges. Also, the proposed concept was used to explain how frequency dependent currents and charges can be implemented in commercial harmonic balance simulators. The extrinsic elements of the equivalent small-signal model were optimized so that the model best match the measurements at an extended frequency and bias range. The transcapacitance element was used to realize the time delay effects in the small-signal model.;Good agreement has been observed between the linearized drain currents and charges and the small-signal Y-parameters. The techniques and tools show promise for integration in a table-based and neural-network modeling process.