ADEV Calculated from Phase Noise Measurements and Its Possible Errors Due to FFT Sampling

Abstract

In this paper, we show that fast Fourier transform (FFT) sampling plays an important role in the calculation of Allan deviation (ADEV) while using the numerical integration as a tool for the time and frequency (T&F) conversion. In order to avoid generation of unreasonable ADEV values, FFT sampling data are re-generated with logarithmic frequency space using an interpolation skill. Therefore, results from both the numerical integration and the power-law processes could match each other quite well. Besides, spurs in spectral density have non-neglectful influences upon ADEV results. For example, when the data of our lab's phase noise measurement system are processed, the ADEV generated from the spectral density with spurs may reach to three times the one while spurs are removed