“Sensor Interfaces” – Prof. Boris Murmann (Stanford University)

While the 2015 & 2016 courses covered data conversion, this year’s course shifts the attention toward modern sensor interfaces. It covers the design of the constituent CMOS mixed-signal circuits, spanning a broad range of topics from “electrons to transistors, bits and algorithms.” The overall objective is to deepen the attendees’ understanding of sensor interfaces and their efficient translation into transistor-level circuits.

The course begins with an overview that identifies modern driver applications, along with typical interface design approaches and associated circuit challenges. Throughout this course, the considered applications include MEMS/inertial sensing and a variety of examples from the biomedical space. This introduction is followed by a design-centric evaluation of CMOS technology, using the gm/ID ratio as the core variable. The understanding from the gm/ID-based characterization is then used to translate block specifications into transistor sizes in a systematic and re-use friendly manner. The considered blocks are typical interface gain and filter blocks and include switched-capacitor stages. This material on gm/ID-based design is a more detailed treatment of the introduction given in 2015.

The remainder of the course focuses on higher-level concepts and begins with a system-centric exploration of common interface solutions. Focus is placed on architectural options that enhance front-end sensitivity and selectivity based on application understanding. This material will also review widely-used techniques such as chopping and correlated double-sampling. The final two lectures zoom out further and look at the interface in terms of the processed information. This leads to an introductory discussion of compressed sensing and machine learning (inference) techniques, which are currently being considered in advanced research.