Performance Comparison of Harmonically Tuned Power Amplifiers at 28 GHz in SiGe BiCMOS

Abstract

As the demand for wireless electronics is increasing, more and more gadgets are connected wirelessly and devices are being improved constantly. The need of the new research and development for advance electronics with high performances is the priority. The data transfer rates are improved for faster communication and better efficiency is to reduce the battery consumption in handheld devices. This thesis presents three single-stage power amplifiers (PAs): class-AB, class-F and inverse class-F (class-F-1) at 28 GHz. The PAs have identical input networks: input matching, base DC feed, and base stabilizing networks. At the load side, there is a different load network for each PA. Class-AB PA load network has a single inductor with a parasitic capacitor to create a resonance at 28GHz. Class-F PA load network is composed of a parallel network (one LC tank in series with an inductor) and a series network (one 3f0-resonance LC tank in series with a capacitor) to create a multi-resonance load network. Class-F-1 load network is composed of a parallel network (two LC tank in series with an inductor) and a series network (one 2f0-resonance LC tank in series with a capacitor) to have a multi-resonance network. The main purpose of using multi-resonance load networks in class-F and class-F-1 is to shape the collector currents and voltages in order to achieve the highest efficiency possible. The chosen bias point is VCE=2.3V and ICE~12mA. As the results, class-AB PA achieves the peak PAE of 44%, 15 dBm OP-1dB, >19 dBm Psat , and 10 dB Gp. Class-F PA achieves the peak PAE of 46%, 14.5 dBm OP-1dB, ~18 dBm Psat , and 10 dB Gp. Class-F-1 PA achieves the peak PAE of 45%, 15.1 dBm OP-1dB, >18 dBm Psat , and 10 dB Gp.. In order to compare the linearity performances among three PA classes, a two-tone signal and a modulated signal with different modulation schemes (QPSK, 16QAM, 64QAM, and 256QAM) are applied to the PAs to produce IM3, ACPR, and EVM. After the analysis and comparison on efficiency and linearity, class-F PA gives the highest efficiency but has the worst linearity while class-AB has the best linearity but has the worst efficiency among three. Class-F-1 PA results lies in the middle of two other classes in term of efficiency and linearity.