New non-linear microwave network parameters with application to gallium nitride high electron mobility transistor modeling with x-parameters

Abstract

High power amplifiers have been receiving increasing attention as key components responsible for a sizable portion of base station cost in wireless communication systems. Higher performance and lower cost can be achieved on a device technology and a circuit design. On the device technology, Gallium Nitride (GaN) technology brings unparalleled performance in efficiency and linearity thanks to its higher energy bandgap, higher breakdown electric field, higher electron velocity and higher operating temperature. Large- and small- signal modeling of GaN HEMTs, which is the aim of our work, is an important step for high power amplifier designer. In general, most of the published models are accurate only in linear mode since they rely on DC and multi-bias S-parameters measurements for model parameters extraction. The availability of X-parameters measurements and their benefits compared to S-parameters presents an opportunity for incorporating non-linear data directly in the modeling process. The main objective of this research is how to exploit X-parameters to more accurately and quickly build a nonlinear device model that captures the device behavior at the fundamental and harmonic frequencies. Nonlinear network Z-, Y-, ABCD-, T-, G- and H-parameters are essential to extract and validate large-signal model based on X-parameters. The expression of nonlinear impedances, admittances, ABCD-, T-, G- and H- are derived from X-parameters. Moreover, standard conversion rules between these nonlinear network parameters are established. The nonlinear network parameters can describe any topology of pure linear or nonlinear or a mix of linear and nonlinear components. In this work, a new equivalent circuit modeling technique based on X-parameters measurements was proposed. The new model is subdivided into extrinsic and intrinsic parts. Extrinsic elements are extracted with a technique based on de-embedding open structure fabricated on the same wafer of the device under test and forward measurements. Xparameters of the intrinsic part are determined thanks to X-parameters de-embedding technique. The intrinsic part is modeled as a modified Pi-Network of nonlinear characteristic impedances modeling the drain-to-source, gate-to-source, drain-to-gate and gate-to-drain junctions.

Date	18 Feb 2016
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Ammar B. Kouki (Supervisor) & Fadhel Ghannouchi (Co-supervisor)