I working on a design that uses an altera APF10K10ATC144-3 device. I have discovered some unusual long delays in a configuration where the input clock (50ns period ) signal at a global clock pin:
a) directly feeds an output A b) passes through an inverter and feeds and output B
According to the quartus II v4.2 sp1 simulator, the delay between the rising flank of the clock at the input and the rising edge of the output A is 15.3 ns (!) and, surprise, between rising clock and the falling edge of output B is 14.9 ns, i.e. output B is even faster (!). Since 15ns is an unusual delay I took the scope and measured, approximately 5 ns in both cases, the delay being obviously slightly less to output A than to output B.
The question is, why is the simulation wrong? (Yes, I have selected the correct device in the assignments). I can understand that reality is always worse than simulation but have almost never the opposite! In any case, the simulator is almost useless in this case.
Any suggestion is appreciated
Pere