This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
if they use the timing numbers for 20Msps and they still have margin in their design, then that would be the most conservative and safe approach.
Note that as the data sheet timng sweeps from 30Msps to 20Msps to 10Msps that the extra time for the bit period at slower sample rates go into providing more setup adn hold time, meaning that the actual skews between clock and data from the device are relatively constant.
At 30Msps the bit period for each bit is 2.77ns and the setup+hold time is 2ns. So the device skews are about .77ns.
At 20Msps the bit period for each bit is 4.16ns and the setup+hold time is 3.4ns. So the device skews are about .76ns.
At 10Msps the bit period for each bit is 8.33ns and the setup+hold time is 7.6ns. So the device skews are about .73ns. See - relatively constant. All the extra bit period goes into more setup time and hold time.
If we consider that the bit period at 12.288Msps is 6.78ns and we subtract 0.78ns of device skew then the setup+hold time must be about 6ns. So I would estimate the setup time to be about 2.9ns and the hold time to be about 3.1ns. Since these are extrapolated numbers, I would derate them slightly to maybe 2.7ns setup time and 2.9ns hold time just to be safe. And then see if there is design margin in your design for these timings.