Contact Us
Browse by Category
Analysis of FFT Performance of ZDS2000 Oscilloscope
Sampling rate:
Fs represents the sampling frequency of the signal measured by the oscilloscope. Each channel of ZDS2000 oscilloscope supports 1GS/S sampling rate. The following will describe the related problems based on the highest sampling rate 1GS/S.
FFT points:
N is used to represent the number of sample data used in FFT transformation of oscilloscope. ZDS2000 oscilloscope can perform real-time FFT operation at 4M points. Obviously, to do FFT with 4 million points, it is necessary to have amazing computing power and efficiency.
Frequency resolution:
The frequency of the minimum resolution of the oscilloscope is expressed by Delta F. Assuming the frequency resolution is 10Hz, 10Hz, 20Hz, 30Hz can be distinguished... The frequency points are equal to 10 Hz integer times, but the frequency points of 15 Hz, 25 Hz, 37 Hz and other non-integer times can not be distinguished. If there is a signal like 15Hz in the measured signal, obviously after FFT transformation, the frequency point of 15Hz can not appear in the spectrum, that is, it can not be analyzed. However, the energy of this frequency point will leak to the adjacent 10 Hz and 20 Hz frequency points, which not only causes the signal of 15 Hz not to be distinguished, but also makes the magnitude of 10 Hz and 20 Hz frequency points inaccurate, because the energy of 15 Hz frequency points leaks to these frequency points. The only way to solve this problem is to further improve the frequency resolution, for example, to 5Hz or 1Hz or higher (lower value). When the sampling rate is fixed, the frequency resolution can only be improved by increasing the number of FFT points. The premise is that the oscilloscope has enough computing power and storage depth. That is to say, both of them are indispensable.
In fact, the above parameters in FFT constitute an important relationship, which will run through the whole exposition of FFT in this paper.
Delta f = Fs / N
That is, the frequency resolution is equal to the sampling rate divided by FFT points. Why does an oscilloscope have to do 4 megabit real-time FFT? In some common oscilloscopes, FFT supports only 8K points at most, and even some oscilloscopes have only 1K points. As can be seen from the above relational expression, the highest frequency resolution is only
Fs/N = 1GS/S/8K point = 125KHz
This means that if the measured signal is not an integer multiple of 125K, the spectrum of the signal can not be judged at all. How many signals can be measured in the real world that are exactly 125 K integer times? So the FFT of such an oscilloscope has no practical value. That's why you hardly see any publicity about FFT performance in other oscilloscope publicity materials!
Because of the deep optimization of FFT, ZDS2000 oscilloscope can support 4M point FFT. According to the above relation expression, the frequency resolution of FFT reaches 1GS/S/4M point = 250Hz at 1GS/S sampling rate. Obviously, the oscilloscope can achieve such frequency resolution at 1GS/S sampling rate, that is to say, as long as the measured signal frequency is an integral multiple of 250Hz, ZDS2000 can be accurately identified. Even if the measured signal is not an integer multiple of 250 Hz and there is spectrum leakage, the approximate frequency points of the measured signal can be accurately analyzed at 250 Hz frequency resolution.
Assuming that the measured signal contains two frequency components of 100KHz and 150KHz, as well as some other higher frequency signals, for ordinary oscilloscopes, because the maximum support is only 8K FFT, that is, the frequency resolution is only 125 KHz, it is obviously impossible to distinguish 100KHz and 150KHz frequency signal. Because the frequency resolution of ZDS2000 oscilloscope at 4M FFT is as high as 250Hz, it can accurately distinguish 100KHz and 150KHz frequency signal.
Assuming that we often need to analyze the noise in the circuit, if the frequency resolution of the oscilloscope is not fine enough, the source of noise can not be accurately located anyway. If ZDS2000 oscilloscope is used for analysis, because the frequency resolution of ZDS2000 oscilloscope can still reach 250Hz at 4M FFT and 1GS/S sampling rate, the noise of 20KHz can be accurately detected after FFT. Since the DC-DC switching frequency is exactly 20KHz!, the noise source can be located quickly.