Low-Power, High-Linearity Mixer Architectures: A Comparative Study

Authors: Lalita Chouhan, Assistant Professor Mr. Divyanshu Wagh

Abstract: This paper presents a comparative analysis of major RF mixer topologies. The Gilbert cell remains the mainstream core thanks to its high conversion gain (CG), strong port-to-port isolation, and suppression of even-order distortion. Multi-tanh linearization implemented with multiple differential transconductance branches delivers excellent linearity but suffers from low CG. Current-bleeding improves both linearity and CG by injecting additional bias current through a dedicated source, at the expense of higher power. Switched-biasing can yield a very low noise figure (NF) by replacing the tail current source with parallel nMOS devices; however, using devices in place of a current source degrades linearity and raises power. A folded-cascode arrangement lowers supply voltage by folding the LO switching pair (often with pMOS switches), but typically worsens NF. Bulk-driven operation reduces power by using the LO(RF) gate for switching and the bulk for amplification, yet it also reduces linearity. Cross-Coupled Post-Distortion (CCPD) cancels third-order terms to achieve high linearity, but it lowers CG and increases power due to auxiliary devices. Similarly, MGTR linearization raises linearity via an auxiliary transistor while penalizing CG and NF. Overall, mixer design is governed by inherent trade-offs among conversion gain, noise figure, linearity, and power consumption.