04-27-2016 12:09 PM
04-27-2016 01:03 PM
If you use the CLK0 output, and the CLKDV ouput with divide=2, the pjase relationship ia guaranteed to +/- 100 ps. The rising edge of the 70 MHz, will happen at every other 140 MHz clock rising edge.
So, yes, use of the DCM's multiple outputs is a good thing to insure the timing is known, and does not vary.
A caution though: which rising edge of the 140 MHz designates the 70 MHz rising edge is not selectable: after a reset of two like configured DCM's their 70 MHz may be in phase, or exactly not in phase, as the divide by two may get initialized differently between the two DCM's. That is why all such outputs need to be from one DCM.
04-27-2016 01:58 PM
04-27-2016 02:02 PM
Two separate DCM, one with 140, the other with 70 MHz are not expected to start up the same every time. If driven by the same CLKIN, outputs from both will be matched to each other, but the divisor may be 0 or 1 initially, so the phase of the 70 MHz may change from reset to reset with respect to the 140: instead keep all such clocks devired from the same DCM.
04-27-2016 02:55 PM