After this function, the array we input is decomposed by Fourier. For the last two parameters, because our sampling array is the original sampling value, bit inversion is generally required, because the FFT calculation requires the input value to be inverted to adapt to the algorithm. arm_cfft_f32(const arm_cfft_instance_f32 * S, float32_t * p1, uint8_t ifftFlag, uint8_t bitReverseFlag) this is the main interface of the fast Fourier transform, the first parameter can be understood as the number of sampling points you input into the FFT The second parameter is the input array the third parameter is positive and negative conversion, generally filled with 0 the fourth parameter is bit inversion enable, generally filled with 1. At the same time, don't forget to add the ARM_MATH_CM4 macro definition. Generally, the single-chip microcomputer is in little-endian mode, so add arm_cortexM4lf_math.lib. The lib file is in the CMSIS\Lib\ARM folder. Just add the lib library of dsp to the project, and then include arm_math.h in the c file that needs to call the library function.
One year later, the dsp library was used in the project, and it was found that it was not necessary to add multiple source codes to the project file as mentioned above. For example, if I am M4, I need to add the ARM_MATH_CM4 macro definition in the precompilation. Then add the macro definition according to your own chip. If not, please go to ST official website by yourself download.įor FFT operation, just add the following 7 files in en.stm32f4_dsp_stdperiph_lib\STM32F4xx_DSP_StdPeriph_Lib_V1.8.0\Libraries\CMSIS to the project.
#Signal part 3 download#
If you are using STM32F4, so coincidentally, just download it directly from my network disk
#Signal part 3 software#
FFT software programming 1.dsp library added Then in this cycle, it is composed of 64 sampling points.Ĭopy the data in the workspace to word, then select all, and replace the newline character "^p" with "," to get an array that can be used directly in C language, as shown in the figure below.Ģ. It should also be mentioned that the simulation step size set in simulink is 0.0003125, the corresponding sampling value is 3200Hz, and the sampling time is set to 0.02 seconds, which is a period of a 50Hz sine wave. Here, the array is directly defined in the microcontroller to store the Matlab waveform signal. The waveforms put together are shown in Figure 2, which is a typical three-phase power in the power grid. Among them, taking the phase of A as the benchmark, B lags behind A by 120°, and C lags behind B by 120°. This is very simple, that is, three sine waves with the same frequency and the same amplitude, but with a phase difference of 120°. In Figure 2, the upper side is the superimposed distorted waveform, and the lower side is the waveform when the three signals are placed in one picture. The model is very simple, as shown in Figure 1, the three sine waves have the same phase, different frequencies, and the amplitudes are 5, 15, and 30 respectively. This waveform is mainly used for FFT calculation for harmonic detection. Harmonic detection constructionīy superimposing 100Hz and 200Hz sine waves with 50Hz sine waves, the superimposed and distorted waveform is obtained.
#Signal part 3 how to#
Today, I will mainly talk about how to use the dsp library to perform fast Fourier calculations in STM32, so as to obtain the spectral amplitude and phase of the signal.ġ. Power systems are often doped with harmonics, and FFT can detect harmonics, which has great practical value. Look at the phase after FFT decomposition
0 kommentar(er)
