Learning to measure resistance noise demystifies the ubiquitous 1/f excess noise

TL;DR

This paper reveals that resistance noise measurements using spectrum analyzers are inherently affected by the measurement process itself, which introduces non-thermal noise components and alters the observed spectral characteristics.

Contribution

It demonstrates how the measurement method influences resistance noise spectra, especially the 1/f excess noise, clarifying the distinction between thermal and non-thermal noise contributions.

Findings

Resistance noise measurements are affected by the measurement setup.

The conversion process introduces non-thermal resistance noise.

The spectral shape of resistance noise can be significantly altered by measurement methods.

Abstract

To study resistance noise (DeltaR) by a spectrum analyzer we must convert this noise into a noise voltage (DeltaV) at the reach of such generalized voltmeter. Whenever a current Iconv is set in a resistor to convert its resistance noise into noise voltage by Ohm's Law: DeltaV=DeltaR*Iconv, the converted noise thus obtained does not track DeltaRte (its resistance noise in Thermal Equilibrium, TE) but DeltaR, that is: a resistance noise out of TE due to Iconv itself. Thus, backgating noises in the channel of resistors (i. e. Field-Induced Resistance Noise, FIRN) found by this method always are noises out of TE. The way the Lorentzian DeltaRte of a resistor is converted by Iconv into nine decades of resistance noise DeltaR with 1/f spectrum is the lesson we give on this unexpected spectral change that we could express as: "To measure is to disturb, particularly in resistance noise measurements".