Fast convergence of white noise cross-correlation measurement archived by vector average in both frequency and time domain

Abstract

White noise measurement can provide very useful information in addition to normal transport measurements. For example thermal noise measurement can be used at sub Kelvin temperature to determine the absolute electron temperature without applying any heating current. And shot noise measurements helped to understand the properties of nano and mesoscopic normal metal/superconductor structures. But at low temperature and for relatively small resistance it is difficult to measure the sample's noise magnitude because the background thermal noise can be much larger and usually there are other pick-up noises. Cross correlation technique is one way to solve this problem. This article describes an improved cross correlation algorithm that averages in both frequency and time domain, and the realization of a simple instrument set-up with PC and sound card. With this set-up it is shown even with much larger background noise and pickup noises, 100pV/$\sqrt{Hz}$ white noise level can be easily measured in seconds. Compared to the normally used cross-correlation methods, it is several orders of magnitude faster.