The paper deals with the diagnostics of industrial automation devices and in particular with the real-time fault isolation. A diagnostic model-based method is proposed with the aim to isolate the occurrences of faults in automatic machinery simplifying the synthesis of the diagnostic models and the application of the developed technique to real cases. The theoretical foundation lays on the diagnoser automaton, which is suitably extended with time and with special modeling techniques that make it suitable for real-time execution rather than off-line test of the diagnosability. The model here adopted is called TiDiaM (Timed Diagnostic Model) for which automatic synthesis algorithms and a Java real-time executor (TiDE, Timed Diagnosis Executor) are proposed. The approach has been applied to a test case (the subsystem of an automatic machining centre called Pallet changer) in order to demonstrate its real-time applicability to an industrial example.