NN-Based Encoding Applied to a Time-to-Digital Converter for Time Interval Measurements

This paper presents a novel proposal for encoding TDL status into a timestamp using an artificial neural network (NN). The proposed method utilizes software-generated signals that imitate real signals observed in FPGA, which can exhibit capture errors (referred to as bubbles) due to process voltage, and temperature variations besides metastability. With this method, it is possible to achieve a high level of precision, up to 10.08 ps, making it a promising solution for deployment in measurement systems with LiDAR or PET applications