We failed conducted emissions and the LISN plot shows a broad hump around 500 kHz rather than clean harmonics. Before we start swapping filter components blindly, what's your systematic approach? Near-field probe first, or straight to filter simulation? How do you tell differential from common-mode quickly?