ABSTRACT
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 of the instrument. We show that a systematic error is always present if the thermal energy of the input power splitter is not accounted for. Such error can result in noise underestimation up to a few decibels in the lowest-noise quartz oscillators, and in an invalid measurement in the case of cryogenic oscillators. As another alarming fact, the presence of metamaterial components in the oscillator results in unpredictable behavior and large errors, even in well controlled experimental conditions. We observed a spread of 40 dB in the phase noise spectra of an oscillator, just replacing the output filter.
ABSTRACT
The flicker noise of the ferrite circulator is a critical element in ultra-stable microwave oscillators, in which the signal reflected from the input of the reference cavity is exploited to stabilize the frequency. This paper explains why the circulator noise must be measured in isolation mode, proposes a measurement scheme, and provides experimental results. The observed flicker spans from -162 to -170 dB[rad2]/Hz at 1 Hz off the 9.2 GHz carrier, and at +19 dBm of input power. In the same conditions, the instrument limit is below -180 dB[rad2]/Hz. Experiments also give information on the mechanical stability of the microwave assembly, which is in the range of 10(-11) m. The measurement method can be used as the phase detector of a corrected oscillator; and, in the field of solid-state physics, it can be used for the measurement of random fluctuations in magnetic materials.
