The puzzling spur of the HMC497 output

I am designing a IQ modulator circuit. The main chip is HMC497. When I chose the LO frequency as 700MHz, and the baseband I,Q frequency as 10kHz, there would be a spur in 618.08MHz with a level of -83.44dBm and in 536.16 MHz with a level of -82.77dBm. The two band of the output was about -13dBm. The LO power is 0dBm. The I,Q baseband power was supplied by E8367D from Aglient and could be adjusted in a wide range. What was worse, the spurs got little change when I adjust the power of baseband and even of the LO.
However, when I change the baseband frequency from 10kHz to 100kHz, keeping the same LO frequency, the two spurs would locate in 853.60MHz and in 546.60MHz
. When I change the LO freuency from 700 MHz to 701MHz, the two spurs (with 100 kHz baseband IQ) would locate in 854.60MHz and in 547.40MHz.
Would sombody tell me some suggestions about such spurs? I couldn't even explain the cause of such spurs which can't be filtered by low pass filter.

- Check the dynamic range specification of the signal generator
- Apply low passes to the I and Q signals

There are LO amplifiers in that chip, they may be oscillating. Check to see if you have enough power supply bypass caps, and that the source impedance is not too bad (note that data sheet specifies a 50 ohm source). Maybe put two 3 dB attenuators at the LO in to see if it goes away. Also make sure it is not a ripple on your power supply.

If those are ok, I would suspect it is noise pick up in your lab, a poor ground somewhere, or defective signal sources feeding it.

< -80 dBm doesn't sound like internal oscillations, picked up noise may be. My basic doubt is, that it may be normal operation of the baseband signal generator, or irregular operation, if the specification is actually better.

That signal generator is only good for 30/55 dB down harmonics/spurs. Probably is that.

The spurs might come from the baseband signals. When I turned off the baseband signals and put the 50 Ohms load in the I, Q input ports, the spurs disappeared. If I just turned off the baseband output key, the spurs were still in there.

Did you notice the comments about baseband generator dynamic range? Besides reviewing the specification, checking the output signal with an analyzer is just plugging a cable and few keyclicks. It's also necessary to decide about a suitable cut-off frequency of a low-pass filter.