Available at: https://digitalcommons.calpoly.edu/theses/3072
Date of Award
6-2025
Degree Name
MS in Electrical Engineering
Department/Program
Electrical Engineering
College
College of Engineering
Advisor
Jason Poon
Advisor Department
Electrical Engineering
Advisor College
College of Engineering
Abstract
Advancements in power electronics require high switching frequencies which the digital control routines of the systems cannot keep up with. Analog computing is added to digital control systems to prevent bottlenecks and allow real-time implementation. This paper proposes a hybrid computing model predictive control system for a buck converter. The control system implements the gradient dynamics of the optimization function using digital computing, and the gradient dynamics for the penalty function and integrator using analog computing. For the full simulations of the entire system, the PLECS RT Box is used to simulate the controller in real time with a 1 kHz buck converter. It also performs the digital computation when implementing the hybrid system with hardware. This project focuses on the hardware development for the analog implementation of the controller. The hardware was tested on a 1 kHz buck with hardware in the loop testing and a 50 kHz physical buck converter. The results of the testing proved that model predictive control can be successfully implemented in real time using hybrid computing. The system showed low overshoot and ripple in the output voltage of the buck converter. Future improvements remain with PWM generation and increasing the switching frequency of the buck converter.