Practical performance of different decoupling methods for digital sensors

Abstract

Decoupling is a technique to provide a noise-free power supply to digital ICs including sensors and is generally recommended for every such IC. Recommendations often contradict each other, and they are usually only given for an assumed clean, stable, power supply. This is contrasted by factors like cost concerns, especially when using i.e. ferrite beads, inrush current concerns when using very large capacitors, or space constraints in small devices.

In this thesis first the recommendations by manufacturers for decoupling were examined, together with an analysis of the theoretical performance using SPICE simulations of different components for decoupling. The accuracy of the used simulation models by manufacturers in combination with practical effects like trace, power plane, and via impedance were tested with an impedance analyzer, and manufacturer data was shown to accurately model the impedance compared to the actual in-circuit impedance within component tolerance. Multiple common decoupling circuits were then examined practically, in combination with the HDC2080 and ADS1115 digital sensors, and the sensors' performance was examined with different decoupling and noise conditions. Degraded performance was found if decoupling is insufficient.