This paper presents a model-based approach for performance analysis and fault detection in non-isolated buck converters, focusing on the impact of soft faults, such as gradual component degradation, and hard faults, including short-circuit or open-circuit failures, on the converter’s dynamic behavior. Pre-fault and post-fault state-space models are developed to extract fault signatures, which serve as key indicators for fault detection and identification. A state observer is designed to generate residuals, which are then utilized in a diagnostic algorithm to accurately identify faults. The research includes detailed modeling of state-space dynamics, analysis of capacitor equivalent series resistance (ESR) effects, and simulation of fault scenarios to diagnose inductor and capacitor soft faults as well as hard faults in the switch.