Performance analysis of correlation techniques for noise measurements

Abstract

The cross-correlation technique makes it possible to perform noise measurements with a sensitivity that would otherwise be unreachable, well below the noise floor of the amplifiers. Not all noise contributions from the amplifiers can however be eliminated or even just attenuated by cross-correlation: therefore it is important to take into consideration the detailed characteristics of the DUT (Device Under Test) and of the amplifiers when setting up the measurement system. Here we discuss the relative advantages of the different (“series” and “parallel”) configurations coupled with our technique for the accurate evaluation of the transimpedance between the noise source to be measured and the amplifier output. In particular, we show (i) the importance of the comparison between the real and the imaginary part of the cross-spectrum due to the asymmetry of the correlation amplifiers and (ii) how to estimate the maximum number of averages in the cross-spectrum evaluation that leads to an actual advantage from the point of view of the measurement accuracy. Finally we discuss the issue of shielding from external spurious signals, whose relevance is often underestimated