Implementation of FPGA-based Phase Detection and Improving the Accuracy Using Machine Learning

Abstract

Abstract In cyclotron accelerators, Digital Low-Level RF or DLLRF system plays a major role in the detection and control of the RF amplitude and phase which means it is necessary to utilize accurate, reliable, stable, and high-speed digital hardware. In this project, Field- Programmable Gate Array or FPGA was selected to design and implement a special phase detection algorithm that is not only independent of the RF amplitude changes, but inde- pendent of the sampling speed of analog-to-digital converter or ADC and the RF frequency. Thus, it will be proved that using the ADCs is possible, without considering the RF amplitude changes, which enables us to detect and control the RF phase in a broad range of frequencies. Finally, the accuracy of this project is maximized by using a linear regression model in machine learning to predict phase differences precisely.