**January 24, 2020**

RF field engineers have relied on spectrum analyzers to detect transient interferers and unwanted hidden signals for each wireless communications generation. As we enter a new era where millimeter wave (mmWave) frequencies are becoming more commonplace due to the rollout of 5G and advanced aerospace/military systems, selecting the proper spectrum analyzer becomes more important. In fact, it can be the difference between quickly and efficiently locating and removing an interfering signal or missing it entirely and having network performance fail to meet key performance indicators (KPIs).

Spectrum analyzers can be segmented into three types – real-time spectrum analyzers (RTSA), swept-tuned models and vector signal analyzers (VSA). Each has advantages and limitations in testing scenarios, but when it comes to interference testing, the RTSA should be the first tool out of the bag.

**Advantages of Real Time Spectrum Analyzers**

An RTSA can continuously acquire samples of elusive and transient signals and perform Fourier transform (FFT) analysis on them. Swept-tune spectrum analyzers and VSAs use fast FFT analysis but follow a serial process flow to acquire samples and calculate the FFT.

The RTSA process flow can acquire a new sample frame while simultaneously performing the FFT on the previous sample frame. This parallel processing requires fast digital hardware and a large memory buffer. RTSA spectrum analyzers perform FFTs at an extremely fast rate. For example, Anritsu’s Field Master Pro™ MS2090A (figure 1), can perform 527,000 FFTs per second for a 512-point FFT.

Most RTSA instruments display the FFT data in a colorful density display, which summarizes the amount of time RF energy is present at any given frequency/amplitude during the capture. This is especially useful in interference analysis for two reasons. First, signals of very short duration will still be displayed on the screen for the user to see. Second, when an interfering signal is hidden inside an intended modulated signal, it will still stand out as a different color and be apparent to anyone monitoring the spectrum.

**Swept-tuned Spectrum Analyzer **

A swept-tuned spectrum analyzer uses an architecture that sweeps its local oscillator (LO) to down-convert the input frequency range to a fixed intermediate frequency (IF). It is then filtered by the resolution bandwidth (RBW) filter and detected. This type of spectrum analyzer can only see a small portion of the frequency span at any moment. It is blind to transient signals that appear when the sweep is scanning a different part of the input frequency range.

**Vector Signal Analysis Process **

A VSA down-converts the signal of interest within a certain bandwidth to a fixed frequency IF. The IF analog signal is sampled by an analog-to-digital converter (ADC) and a stream of time domain samples is used for modulation domain analysis. For spectrum analysis, the time domain samples are transformed into a frequency domain spectrum using the FFT. When the FFT calculation is finished and the results are transferred to the display, the next sample frame is acquired. Although the LO is stationary, the VSA is blind to signal events occurring between sample frames.

**Other RTSA factors**

The key performance metric when evaluating spectrum analyzers, especially for finding transient interference, is the probability of intercept (POI). POI is defined as the minimum signal duration necessary to accurately measure the amplitude of a continuous wave (CW) signal. The smaller the POI, the higher the chance of catching intermittent interference.

POI is affected by several factors, including FFT processing speed, sample rate, window overlap, RBW, and span. Let’s examine a few of these considerations.

**Windowing**

When a sample block is acquired for FFT analysis, the mathematics of the FFT presumes the time domain signal is periodic with a segment equal to the sample frame duration. The samples at the beginning and end of the sample frame create an unwanted discontinuity in the time domain. The frequency domain energy is spread out, rather than concentrated at the original sine wave’s frequency.

Spectral leakage is undesirable in FFT spectrum analysis because the ability to resolve closely spaced frequency components with different amplitude levels is lost. Signal amplitude is no longer an accurate representation of the true signal level because the spectrum energy has spread out.

To eliminate these effects, the segments in the sample frame are multiplied by a window function that smoothly tapers them near the start and end of the frame to zero. Thus, when these modified samples are presented for FFT analysis, the periodic extension does not have a sharp discontinuity and the spectrum leakage is reduced.

**RBW and Span Interdependence**

Windowing also serves to implement the RBW filter in FFT spectrum analysis. When the FFT analysis is performed with a window function applied to the input signal sample record, the effect is equivalent to a bank of parallel RBW filters. Figure 2 shows a simplified representation of the RBW filter band frequency response, separated by their -3 dB bandwidth, across the span of one FFT. You will see that the RBW filter response is incomplete at the start and stop frequencies of the FFT span. Because of this, the usable bandwidth is only about 80% of the full analysis bandwidth of the FFT, which is set by the input sample rate (Fs) times the number of points (N) in the FFT.

**Window Overlap**

Since windowing tapers the time samples at the beginning and end of the sample frame to zero, transient signal events at the edges are lost. Overlapping is used to ensure the capture of these signal events. Each sample frame for the FFT is partially filled with samples that were captured in the previous frame.

For total amplitude accuracy, the signal must occupy the entire area under the window function. Meeting this condition will ensure that at least one sample frame and its FFT will capture the full signal amplitude since the start of the signal event can occur at any time relative to the start of any sample frame (Figure 3).

If FFT analysis is performed using overlapping sample frames, the required signal duration for 100% POI is shorter. If the window length is smaller than the FFT size, the required signal duration will be shorter, thereby improving POI. A wider RBW corresponds to shorter window length and shorter POI.

**Density Display Resolution**

Selecting the appropriate point FFT size depends on the situation. A 512 point FFT size allows for the lowest POI, however the FFT frequency bin resolution is coarser and can cause the display to appear more granular at low RBW settings. The 1024 point FFT size allows a smaller RBW setting for a given span with less display granularity at the cost of a higher POI.

Because a 512 point FFT updates 527,000 times per second, it can show the measurement results in a more meaningful way than an LCD display. A density display shows the color graded intensity of a signal event. The warmer the color, the more frequent the signal event. The density display also can simultaneously show up to six spectrum traces, which represent the detection result for all the FFTs calculated during the acquisition interval.

**Spectrograms Locate Intermittent Signals**

A spectrogram displays spectrum vs. time, making it important in detecting intermittent interferers. After every acquisition interval, a new line is added to the bottom of the spectrogram and the oldest spectrogram line is discarded from the top. The spectrogram has 142 lines and the maximum displayed time record is 142x acquisition time. The spectrogram differs from the density display in that it does not indicate how many signal events happened during the acquisition time.

To learn more about spectrum analyzer architectures and the benefits and limitations of each, download the *Understanding Key Real-Time Spectrum Analyzer Specifications* white paper*.*

## Comments

You can follow this conversation by subscribing to the comment feed for this post.