Evaluating Noise Sources By Gaussian Distribution

A typical Gaussian distribution is generally plotted with the number of sigma, or standard deviations, shown on the horizontal (X) axis (see figure, left). When measuring the performance of a noise generator, a critical factor in assessing its usefulness is to compare its distribution with that of the ideal mathematical distribution shown above. Using the histogram feature on a spectrum analyzer makes it possible to assess the "Gaussian behavior" of the source under test. Such a histogram (see figure, right) was used to show recorded hits over time superimposed over the time-domain profile of a noise source. The peak-to-peak variation measured over time allows one to establish the "peak factor"—a measure of how many standard deviations the source reaches. Other factors, such as symmetry about the mean and distribution within the standard deviations, can also be analyzed to determine the Gaussian quality of a noise source under test.