Fault diagnosis is a crucial task to guarantee reliability, and reduce losses and production cost in smart machining. In order to do so, it is necessary to implement a fault diagnoser that does not use a large amount of memory and that is capable of detecting the occurrence of a fault in a fast manner. In this paper, we propose timed automaton models, which represent the drilling process on a CNC machine, that are suitable for fault diagnosis. The fault diagnoser uses only the spindle power and Z axis displacement read directly from the system controller, and does not need any additional sensors. The proposed diagnoser is capable of identifying the occurrence of a wrong sequence of events executed by the system and the use of a material different from the expected.