I am having a problem, present in three different DDS eval boards, where the sync_clk output shows blips (either in phase or frequency) at roughly 300us intervals. Of course this does affect the RF output in a similar manner, but this is not really relevant since the problem must be introduced somewhere before the sync_clk. To illustrate the problem, I sent the sync_clk output through an RF splitter and then put both of those outputs into a mixer with a low pass filter. If the sync_clk was a pure single frequency, the result would have been a flat line on the scope. Instead I saw spikes in regular intervals: evidence that there must be some periodic phase or frequency blip in the sync_clk (and master clock?).
Please let me know if you are able to reproduce this in your lab.