Many analyzer measurements have considerable amounts of noise present in them. Except in those cases in which the noise is to be measured, the noise can be considered undesirable. Two basic techniques are used to reduce the noise-filtering and averaging. ^[1] Filtering usually takes the form of an analog filter, but it can also be implemented in digital form, while averaging is almost always done digitally. The two concepts are closely related and are treated here in a unified manner. Both filtering and averaging can be classified as either predetection (before the detector) or postdetection (after the detector). Predetection averaging/filtering reduces the noise present in a measurement while postdetection averaging/filtering reduces the amount of fluctuation in the noise.

10.1 Predetection Filtering

In the spectrum or network analyzer block diagram, filtering can be broken down into two types-predetection and postdetection-depending on whether the filter resides before or after the detector, as shown in Figure 10-1. In most swept analyzers, the detector is implemented as a distinct circuit block, although digital signal processing techniques can be used instead. There is no detector circuit in an FFT analyzer-instead the magnitude detection is done by computing the magnitude of the complex frequency domain data provided by the FFT algorithm. Whether there is or is not an actual detector circuit, the concept remains the same.

Figure 10-1: Predetection filtering takes place in front of the detector while postdetection filtering is performed behind the detector.

Noise

There is always some noise present in...