In this article, we present an active diagnosis method to enhance diagnosability of a plant modeled by a finite-state automaton. Some properties of active diagnosis and silent blocking are studied. To avoid silent blocking, the notion of stop-free event set is proposed, and a stop-free control policy that does not introduce silent blocking is formulated. We develop a heuristic method based on the verifier of the plant to compute a feasible stop-free event set that guarantees the existence of a valid control policy. With the stop-free event set, the set of disabled edges is computed, and an online control policy that is based on the current diagnostic state is computed, which guarantees that the closed-loop system is diagnosable. The structural complexity of the proposed control structure is polynomial with respect to the number of states of the plant.