Computer simulation has significant potential as a supplementary tool for effective conceptual-change learning based on the integration of technology and appropriate instructional strategies. This study elucidates misconceptions in learning on diodes and constructs a conceptual-change learning system that incorporates prediction-observation-explanation (POE) and simulation-based learning strategies to explore the effects on correcting misconceptions and improving learning performance. Thirty-four sophomore students majoring in engineering participated in the experiments. The empirical results indicate that the system significantly corrects participants' misconceptions on diodes and improves learning performance. This study shows that conceptual change instructions could correct misconceptions effectively by constructing scenarios that conflict with existing knowledge structures. The results also show that misconceptions on diode models and semiconductor characteristics could be corrected in more than 80% of cases. Conversely, difficulty in correcting misconceptions correlates with the fundamental definition of voltage, circuit analysis, or the interaction between different diode concepts.