While averaging is a common technique used to reduce measurement uncertainty, not all power averaging techniques, espe¬cially in spectrum analyzers, give the same results. In fact, differences in averaging techniques can cause considerable error in signal measurements.

Averaging is a common tech¬nique for reducing the measure¬ment uncertainty inherent in all measurements. Performing the same measurement a number of times and calculating the average of the measured values can often reduce the ran¬domness of an experimental result. Many (if not most) instruments perform this aver¬aging automatically. Rather than returning 100 noisy measurements, the instrument is responsible for taking all 100 measurements, calculating their average, and returning just the average. Averaging is so common and conceptually simple that one might assume there isn’t much debate on the correct way to average. However, recent experience has demonstrated that power averaging in spec¬trum analyzers isn’t necessarily straight¬forward. The following discussion explores the issues associated with power averaging in order to help readers avoid making the same mistaken assumptions made by the author. The conclusions presented here are the results of an experiment that involved correlating the power measurements of two spectrum analyzers from different vendors. However, the issues discussed are generic in the sense that they apply to any spectrum analyzer power measurement with some form of post-detection averaging.

Incorrect Assumption #1: To find the average power of a Zero-Span trace or a portion of the trace, average the RMS power.

Averaging is so natural to engineers that it hardly seems to merit presenting the mathematical formula for calculating it. Nonetheless, to get everyone on the same page, let’s refer to Eq. (1). M_AVE is the average of a series of individual measurements taken over N trials of an experiment, where each of those measurements is denoted as M_i: