An 11.1--27.5-GHz SiGe LNA With 1.62-dB Noise Figure and 32.2-dB Gain Using Noise Matching and Gain Flatness Optimization Techniques

Abstract

Achieving wideband simultaneous noise and input matching (SNIM) is critical for low-noise amplifiers (LNAs). However, the intrinsic noise regulation mechanism of transformer feedback remains unrevealed in existing studies, which mostly focus on bandwidth extension. This article systematically reveals this mechanism, demonstrating that, compared with conventional inductor feedback, transformer feedback reduces the equivalent noise resistance ( $R_{\mathrm {n}}$ ) and flattens the optimal noise impedance ( $Z_{\mathrm {opt}}$ ) without degrading the minimum noise figure (NF) ( $NF_{\min }$ ), establishing a theoretical basis for wideband SNIM. Based on this, an 11.1–27.5-GHz three-stage LNA is implemented in a 0.18- $\mu $ m SiGe process. A transformer-feedback network in the cascode input stage achieves wideband SNIM, while high-coupling interstage transformers extend bandwidth but introduce a midband gain dip and high-frequency gain roll-off. To compensate, an improved current-reuse topology with parallel L-R pole allocation is proposed to flatten the gain response. Measured results show that the proposed LNA achieves a peak gain of 32.2 dB over a 3-dB bandwidth of 11.1–27.5 GHz and a minimum NF of 1.62 dB. It consumes 11 mW from a 1.5 V supply.