07-03-2013, 10:23 PM
The way I've sort of got around this is by defining my devices to operate within a range of timings. From the best case to the worst case, it will always function within that range, even if the odds of the worst case happening don't really justify including it in the timing calculation. Even so, if I'm going to be making it part of a larger device, I'll base the timing on the worst case - that way I can be sure that it has enough time to output, regardless of the probable output timing.