I am outputting a 50 MHz single tone with the AD9954 using a 100 MHz crystal oscillator as the reference clock. On a spectrum analyzer I see the 50 MHz signal fine, but I also see a 100 MHz spur. This spur is about -35 dBm below the 50 MHz tone. No matter what frequency I output with the AD9954, I still see the 100 MHz spur. As a result, this spur mixes with the orginal signal to produce other unwanted spurs. It is mostly likely the crystal oscillator leaking through.
What are some common practices that I can look into to reduce this oscillator leaking?