Some design environments may prevent Design for Testability techniques from reducing testing to a combinational problem: ATPG for sequential devices remains a challenging field. Random and deterministic structure-oriented techniques are the state-of-the-art, but there is a growing interest in methods where the function implemented by the circuit is known. This paper shows how a test pattern may be generated while trying to disprove the equivalence of a good and a faulty machine. The algorithms are derived from Graph Theory and Model Checking. An example is analyzed to discuss the applicability and the cost of such an approach.