Source impedance influence on cross-correlation phase noise measurements

The phase noise floor of an oscillator has been shown both theoretically and experimentally to be the ratio of the source noise power divided by the delivered power. In a 50Ω system this is determined by -177dBm - P_OUT in dBc/Hz. Recent measurements have shown what appears to be better than theoretical noise floors in some oscillators, assuming a 50Ω environment. However, oscillators rarely exhibit an output impedance of exactly 50Ω and vary significantly farther from the carrier. While the input impedance of a modern cross correlation analyzer may be 50Ω, the assumption the analyzer introduces 50Ω common mode noise can be erroneous. This paper presents theory and measurements that demonstrate the extremely low phase noise measured on some oscillators is real and not in violation of theoretical limits by isolating the noise sources in a series of additive phase noise measurements.