Impossible ? ... a dangerous term to use ..
I have no specific knowledge of the PT6 and associated systems but have seen a situation on another, somewhat similar, engine where the prop, due to a suspected system fault, went into feather while the engine continued to operate. As you might envisage, this caused some confusion in the cockpit.
Subsequent failure mode analysis technical investigation concluded that the likelihood of this happening was quite low and that the original certification analyses remained valid.
It is important to keep in mind that all of the systems with which we play have been designed on the basis of reasonable engineering assumptions .. at the time ... and with reasonable assumptions of expected in-service system deterioration.
On both counts if
(a) the assumptions were ill-conceived (ie the design analysis was wrong - and this can, and does, happen especially if the matter under study is innovative or the system operation not really well understood and inadequate research and/or testing is applied)
(b) the in-service use of the system does not approximate that assumed at the design and certification stages (ie either the operator uses the system in a manner different to that presumed by the design team or the aircraft is operated in environmental conditions outside the presumed design case parameters)
(c) one of the known possible failures (which was assigned a reasonable very low probability of occurring) does happen (ie Lady Luck is against you)
... then the "impossible" (read for "impossible", an occurrence which has been determined to be highly unlikely) might well jump up and surprise the crew.
There are plenty of examples in any number of engineering disciplines ....