The impact of thermal energy on cross spectrum PM noise measurements

Virtually all commercial instruments for the measurement of the oscillator PM noise make use of the Cross Spectrum method (arXiv:1004.5539 [physics.ins-det], 2010). High sensitivity is achieved by correlation and averaging on two equal channels which measure the same input, and reject the background noise. We show that a systematic error always present if the thermal energy of the input power splitter is not accounted for. Such error can result in noise under estimation up to a few dB in the lowest-noise quartz oscillators, and in a complete nonsense in the case of cryogenic oscillators.