A Low-Power Fully-Differential Chopper-Stabilized LNIA in 22nm FD-SOI for MEMS-based Thermopile Sensors

This paper presents an analog front-end interface circuit composed of a fully-differential, chopper-stabilized, lownoise instrumentation amplifier (LNIA) for MEMS-based thermopile sensors in 22 nm FD-SOI technology operating at a 0.8 V supply. A fully-differential architecture was chosen to provide a broader full-scale range for the succeeding processing stages with the given supply voltage. This configuration utilizes the inherent thermopile output resistance as the LNIA input resistor, allowing for precise gain control of the amplifier gain to accommodate thermopile devices with varying output voltages. In this specific study, the LNIA was designed to operate as a readout circuit for a differential thermopile sensor modeled in COMSOL Multiphysics. The modeled thermopile sensor generates an output voltage in the range of $\mathbf{6 1 2} \mu \mathrm{V}$ to 4.9 mV and has an equivalent resistance of $51.6 \mathrm{k} \Omega$. The proposed chopper LNIA demonstrates an input-referred noise density of $53.9 \mathrm{nV} / \sqrt{H z}$, representing a $\mathbf{9 9. 9 7 \%}$ reduction in noise of the base amplifier without chopping. This noise performance was achieved while maintaining a low power consumption of $31.2 \mu \mathrm{~W}$ when operating at a chopping frequency of 67 kHz, with an NEF of 12.98. This study shows promising implementation of a chopper-stabilized LNIA for low-energy applications, making it suitable for MEMS-based thermopile sensor systems.