phase noise measurement

I was trying to measure VCO phase noise with Agilent N9320B spectrum analyzer. My problem is that whenever I connect my VCO output to the RF input of the analyzer, the signal on the oscilloscope is gone as my signal swing is reduced from 0-3.3V to about 1.5V-1.7V, and the result shown by the analyzer is not correct. I suspect that it is due to the low input impedance of 50ohm which reduces my signal swing. So I wonder how I can do the correct measurement. Thanks.

Use a buffer emitter or source follower.
Check the linearity and noise of the buffer to do not affect the signal generated by the VCO.
Ideally during a VCO phase noise measurement the power supply is a battery.