Speaker
Description
Monolithic Active Pixel Sensors (MAPS) integrate sensor and readout electronics into a compact silicon tile and are an attractive option to build low mass tracker modules for high energy physics experiments. However, tuning the analog circuits in the front-end is heavily influenced by digital-to-analog converter (DAC) parameters and presents a complex challenge also on module level. Traditional approaches, such as expert-based settings or simulation-driven methods, often lack precision or are computationally expensive.
This work provides a comprehensive analysis of the influence of DAC parameters on the front-end performance of MAPS, detailing how these parameters interact to impact overall sensor behavior. The study offers insights into reducing the parameter search space, allowing for faster and more efficient tuning to achieve optimal sensor performance.
Furthermore, we introduce the application of Shapley values, a concept from cooperative game theory, to automatically assess the importance of each DAC parameter using experimentally collected data. Our results show that Shapley values can be approximated with enhanced noise robustness and the ability to handle missing data. Validation on hardware demonstrates how the significance of specific parameters varies under different circuit conditions, providing valuable feedback for designers and users seeking to optimize sensor performance.
