Digital Filters Design Automatization for DS-ADCs Applications

Computational and signal processing tasks are largely executed using digital methods. This is because digital circuits, known for their robustness, can be implemented using remarkably small and simple structures. This advantage bolsters the predominance of digital techniques in nearly all areas of communication, supporting a broad range of applications with reliable, straightforward designs. Nevertheless, the physical world remains analog to data converters (ADCs) are needed as interface with the digital processing (DSP) core. Within the context of this thesis, the application focus is on Delta-Sigma Analog-to-Digital Converters (DS-ADCs). The primary goal is to automate the design process of the digital filters required for such structures. The existing workflow for designing filtering structures is highly time-consuming, as a unique filter must be created for each specific application. This inefficiency significantly affects the time-to-market (TTM), creating a competitive disadvantage and making it challenging to meet stringent deadlines. Therefore, an automation process is essential to stay competitive and adhere to tight project timelines. To address this, an in-depth analysis of the current workflow has been conducted to identify potential improvements. This examination particularly focused on the interfaces between the various stages leading up to the creation of a new filter design. The aim is to streamline the process, thus enhancing efficiency and reducing the overall time required for a new filter design. The current workflow has been restructured and an ad-hoc framework has been developed in order to build a robust link between the filtering structure designed in MATLAB at a concept level, and the generation of an equivalent HDL. With this approach, the user, can produce any type of filtering structure needed for DS-ADCs with minimal effort and without writing a single line of code. This significantly simplifies the process, making it more straightforward and efficient, while retaining the flexibility to create an array of filter structures as needed. To test its efficiency, the framework has been benchmarked against already existing tools that share the same goal in automating filters design. Specifically, approaches based on the use of HDL Coder have been thoroughly evaluated and comparisons with results of an HLS based framework have been made. By analyzing the different MATLAB and HLS tools, the advantages and disadvantages of the developed framework are highlighted, particularly in terms of workflow efficiency and area and power consumption of the structures generated. This comprehensive analysis further sheds light on the effectiveness of the new framework, providing crucial insights that can guide future refinements and enhancements.