Analysis of linearization methods for low noise amplifiers

Loading...
Thumbnail Image

URL

Journal Title

Journal ISSN

Volume Title

School of Electrical Engineering | Master's thesis

Department

Mcode

Language

en

Pages

54

Series

Abstract

Recently, ultra-linear radio frequency (RF) receivers have attracted increased attention due to the rising popularity of full-duplex communication systems and the dense occupancy of the RF spectrum by multiple carriers. In full-duplex systems, both the transmitter and the receiver are integrated into the same silicon die, causing the leakage from the transmitter to appear as an interfering signal to the receiver. As the RF spectrum becomes saturated, interfering signals with closely separated frequencies are prevalent, especially in the sub-6 GHz frequency band. These interfering signals generate intermodulation distortion (IMD) components, significantly degrading the performance of the receiver. To address this issue, the low noise amplifier (LNA), usually the first stage of the RF receiver chain, must be sufficiently linear to suppress these IMD components and enhance the dynamic range. This thesis introduces a design methodology for highly linear LNAs, specifically improving the third-order input intercept point (IIP3) by suppressing third-order intermodulation distortion through implementing different linearization methods. Enhancing LNA linearity necessitates trade-offs among key metrics such as linear gain, power consumption, noise figure, bandwidth, and design complexity to optimize overall performance. Initially, the thesis reviews ten methods for LNA linearization and evaluates their impact on the metrics mentioned above. Using the analysis, several methods are implemented using a 22 nm CMOS process, demonstrating the practical application of theoretical findings. Finally, this thesis presents two highly linear LNAs based on common gate (CG) topology operating at 1 GHz as the research output. The first design comprises a common gate-common source (CG-CS) LNA with derivative superposition, achieving a differential gain of 14.4 dB, a noise figure of 4.7 dB, and an IIP3 of 12.7 dB. The second design incorporates both derivative superposition and harmonic termination to the CG LNA, resulting in a gain of 11.7 dB, a noise figure of 5.3 dB, and an IIP3 of 12.9 dB. The designed LNAs achieve more than a 6.5 dB improvement in IIP3 over the reference standalone CG LNA.

Description

Supervisor

Ryynänen, Jussi

Thesis advisor

Stadius, Kari
Jokiniemi, Kimi

Other note

Citation