Characterization, modeling and compensation of long-term memory effects in GAN HEMT based radiofrequency power amplifiers

• Autor(a) principal: Tomé, Pedro Mirassol
• Data de Publicação: 2020
• Idioma: eng
• Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Texto Completo: http://hdl.handle.net/10773/30994
• topic: Analog linearization; Analog signal processing; Charge trapping; Digital predistortion; Digital signal processing; Electron trapping; GaN high-electronmobility transistor; Long-term memory effects; Power amplifier linearization

Gallium nitride (GaN) high-electron-mobility transistors (HEMTs) have emerged as the most compelling technology for the transmission of highpower radio-frequency (RF) signals for cellular mobile communications and radar applications. However, despite their remarkable power capabilities, the deployment of GaN HEMT-based RF power amplifiers (PAs) in the mobile communications infrastructure is often ruled out in favor of alternative siliconbased technologies. One of the main reasons for this is the pervasiveness of nonlinear long-term memory effects in GaN HEMT technology caused by thermal and charge-trapping phenomena. While these effects can be compensated for using sophisticated digital predistortion algorithms, their implementation and model-extraction complexity—as well as the power necessary for their real-time execution—make them unsuitable for modern small cells and large-scale multiple-input multiple-output transceivers, where the power necessary for the linearization of each amplification element is of great concern. In order to address these issues and further the deployment of high-powerdensity high-efficiency GaN HEMT-based RF PAs in next-generation communications and radar applications, in this thesis we propose novel methods for the characterization, modeling, and compensation of long-term memory effects in GaN HEMT-based RF PAs. More specifically, we propose a method for the characterization of the dynamic self-biasing behavior of GaN HEMTbased RF PAs; multiple behavioral models of charge trapping and their implementation as analog electronic circuits for the accurate real-time prediction of the dynamic variation of the threshold voltage of GaN HEMTs; a method for the compensation of the pulse-to-pulse instability of GaN HEMT-based RF PAs for radar applications; and a hybrid analog/digital scheme for the linearization of GaN HEMT-based RF PAs for next-generation communications applications.