Noise Measurements

In most cases, two detectors are used for measuring noise-like signals, either a sample or average detector. A sample detector provides a single sample in each trace point on the display. The sample is taken between each trace point as the spectrum analyzer is swept and displayed in the next trace point. The average detector averages, on a linear power scale, all samples between two trace points. The average value is then displayed in the next trace point. With noise-like signals, peak or normal detectors should not be used, since these detector types do not provide a good statistical distribution of the signal and can yield inaccurate power measurements. Peak detectors (Max detectors) are primarily used to measure CW signals.

One of the major challenges in measuring noise-like signals is the variance in measurement results caused by the nature of the signals themselves. Several averaging techniques can be used to reduce this variance. Most notable are trace averaging, video bandwidth (VBW) averaging, and averaging with the average detector. While these techniques can be combined, normally only one is needed.

Trace Averaging

Trace averaging is simply averaging each trace point from one sweep to the next. When trace averaging is turned on, the detector is auto-coupled to the sample detector. Because it can be used with any detector type it is one of the most common averaging techniques and is supported on almost all spectrum analyzers.

VBW-to-RBW Ratio

Another approach is to reduce the analyzer’s VBW to less than the resolution bandwidth (RBW Resolution Band Width (RBW or ResBW): specifies the minimum frequency bandwith that two separate frequency spectra can be resolved and viewed seperately. For FFT (digital) based VSA's the process is equivalent to passing a time-domain signal through a bank of bandpass filters, whose center frequencies correspond to the frequencies of the FFT bins. For a traditional swept-tuned (non-digital) spectrum analyzer, the resolution bandwidth is the bandwidth of the IF filter which determines the selectivity.). In many spectrum analyzers, the VBW-to-RBW ratio defaults to 1, providing a small amount of averaging. In most cases, a VBW-to-RBW ratio of 0.1 sufficiently reduces the variance in the majority of signals to an acceptable level.

Average Detector

Often times, a better approach is to use the average detector. When the average detector is selected simply increasing sweep time applies more averaging. Increasing the sweep time also increases the number of samples taken and averaged between sweep points, thereby reducing signal variance.

Power Vs. Log Scale

A key tip when measuring a noise-like signal is to make sure you are averaging on a power (RMS) scale and not a logarithmic scale. This is because the log of the average is not equal to the average of the log. Averaging a noise-like signal on a log scale can cause an error as great as -2.51 dB. Simply set the Average Type to Power (RMS) to ensure accurate measurements.