Thread Content
It is very important to analyze the frequency components in signals, as they often cause noise in designs; once they exceed the allowable tolerances, it can lead to malfunctions in the devices. In severe cases, it can also cause voltage spikes that damage the devices. If we do not conduct proper testing during the design phase, the above problems are very likely to occur. So how can we analyze the frequency components of a signal? Perhaps people think that only a spectrum analyzer can perform this task, but in fact an oscilloscope can also handle it to some extent. In addition to time-domain analysis, an oscilloscope has an FFT function that can be used for this purpose. FFT is the abbreviation for Fast Fourier Transform. Simply put, FFT is an algorithm that helps us separate time-domain signals, and then convert these separated signals into the frequency domain. At this point, the oscilloscope switches from the time domain to the frequency domain, displaying the relationship between the signal amplitude and frequency. As shown in the following GIF, it is clear how the oscilloscope converts signals from the time domain to the frequency domain. If you are not very familiar with the time-domain to frequency-domain conversion performed by FFT, you can read our previous article titled \"A Brief Understanding of the FFT Fast Fourier Transform Function and Its Applications in Oscilloscopes\". The menu bar of FFT includes options for selecting the type of spectrum to be generated; you can choose to display the amplitude in volts-per-hertz or in dB-per-hertz, which will then be shown on the oscilloscope screen. When FFT is enabled, it can be seen that the time base on the horizontal axis changes from time to frequency, and the units on the vertical axis become V or dB. Below the spectrum type is the selection of the trigger source, which is easy to understand: we choose the channel for which we want to perform the FFT operation as the source. Below the source are four different FFT windows, namely the rectangular window, Hamming window, Blackman window, and Hann window. So why are there different window choices for FFT? Since the FFT algorithm, when calculating the spectrum of a signal, can only obtain information on the sampling points, it is not possible to measure and process an infinitely long signal; instead, only finite time segments are analyzed. As a result, the data information contained in the intervals between samples is lost, which is inevitable and is also known as the gating effect. An oscilloscope performs an FFT transformation on a time record of finite length, and the FFT algorithm assumes that the time-domain waveform repeats continuously. In this way, when the period is an integer, the amplitude of the time-domain waveform is the same at the beginning and the end, so the waveform does not experience any interruptions. However, if the period of the time-domain waveform is not an integer, it causes the amplitude of the waveform at its beginning and end to differ, resulting in high-frequency transient interruptions at the connection points. In the frequency domain, this effect is called leakage. Therefore, to avoid leakage, a window function is multiplied by the original waveform to force the values at the beginning and end to be zero. Different window functions employ different algorithms, and each has its own advantages in various situations. Window functions alter the waveform in the frequency domain, shaping the spectrum in a form that is easier for us to observe. However, they do not eliminate spectral leakage; each window function has its own unique characteristics, and we simply need to choose one based on the requirements of our measurements. Applications of window function effects: The rectangular window is the best type of window for distinguishing frequencies that are very close to each other; however, it performs worst when it comes to accurately measuring the amplitude of these frequencies. It is best to measure the spectrum of non-repetitive signals and the frequency components near DC. This window is used for transients or bursts at the signal level before or after events that are almost identical. The Hamming window has the best frequency resolution for values that are very close to each other; it also offers slightly improved amplitude accuracy compared to the rectangular window. The frequency resolution of the Hamming window type is slightly higher than that of the Hanning window type. Measure sine, periodic, and narrowband random noise. This window is used for transients or spikes at the signal level before or after events with significant differences. The Hanning window is excellent for measuring amplitude accuracy, but it performs poorly in terms of frequency resolution. Like the Hamming window, the Blackman-Harris window is best for measuring frequency amplitude, but it performs worst when it comes to measuring the resolution frequency. Use the Blackman-Harris measurement to find the main single-signal frequency waveform of higher harmonics. At the same time, the following points should be noted during measurement: 1. Since FFT is a mathematical function, for such functions, the more data that is processed, the more accurate the results will be. Therefore, when making measurements, we need to increase the storage depth and the time base as much as possible, so as to achieve a higher frequency resolution. As shown in the two figures below, which compare the results when the time base is set to 200 μs and 2 ms respectively, it is clear that the FFT performance is much better at a time base of 2 ms. However, it should also be noted that a longer duration for the time-domain signal is not always better, as the oscilloscope has a limited storage capacity; the longer the waveform is recorded, the lower the sampling rate, which may lead to distortion of the original waveform. Generally speaking, it is appropriate for a waveform to have a duration of at least 4 to 8 waveform cycles on the time-domain plot. 2. Signals with a DC component or offset can cause errors or deviations in the FFT waveform components; to reduce the DC component, we can opt for AC coupling. 3. When acquiring periodic signals, the average sampling mode should be used to reduce signal noise. It is recommended that the average be no less than 16. FFT can be used in electronic measurement to identify sources of noise interference, test the pulse responses of filters and systems, perform jitter analysis, harmonic power analysis, electromagnetic interference analysis, frequency response analysis, and more.