Spatial Heterodyne Raman Spectroscopy (SHRS) is a new type of Raman spectroscopy detection technology, which has the advantage of high throughput, high spectral resolution, high sensitivity and no moving parts. SHRS can meet the high-sensitivity detection requirements of weak Raman scattered light, and can also obtain clear and sharp Raman spectra. For Raman spectrometers, fluorescence is an inevitable background signal. The fluorescence intensity and the Raman intensity are approximately inversely proportional to the fourth power of the wavelength, so the excitation wavelength of near-infrared light has lower fluorescence than visible light. The excitation wavelengths of near-infrared light are mostly 785 nm, 830 nm and 1 064 nm, of which the shorter 785 nm has larger fluorescence. Although the 1064 nm excitation light has a weaker fluorescence, it requires the near-infrared InGaAs focal plane. Compared with visible detectors, it has higher noise, lower sensitivity and resolution. Therefore, this article chooses the wavelength of 830 nm as the excitation light for Raman spectroscopy detection, and its fluorescence is lower than that of 785 nm. On the other hand, the visible detectors can be used for high-sensitivity detection. For the excitation wavelength of 830 nm, this paper designs, simulates, develops and tests SHRS. The Littrow wavelength of the spectrometer is 842 nm, the theoretical spectral sampling interval is 2.96 cm^-1, and the theoretically Raman shift range is 171.71~3 031.04 cm^-1. The spatial heterodyne interferometer adopts integrated adhesive technology. To increase the throughput, the field-widened prisms are added to the interferometer. The field angle tolerance of the interferometer is selected to be ±2° to ensure the contrast of the interferogram in actual work, and the corresponding contrast of the ideal interferogram is better than 0.98. The fringe-imaging lens group selects a double telecentric lens group with a magnification of 1. The telecentric configuration guarantees the uniform illumination of the image surface, and the symmetrical structure can effectively balance aberrations and further enhance the stability of the system. A checkerboard target is used to test the processed fringe-imaging lens group. The measured average magnification is 1.001 9 and the relative distortion is 0.19%. The Kr lamp is used as the input light of the system to verify the design parameters of the SHRS prototype. According to the positions of the two spectral lines 877.675 nm and 892.869 nm of the Kr lamp and the corresponding Raman shift, the actual spectral sampling interval is 2.918 2 cm^-1. The smaller value compared with the design value is mainly due to the dispersion of the field-widened prism. The actual Littrow wavelength is 841.95 nm, which is close to the theoretical value. The detector selected in this paper does not respond to light with a wavelength greater than 1 000 nm, so the actual Raman shift range is 171.01~2 048.19 cm^-1. The design parameter and the simulation of the system are verified. In the Fourier transform of the interferogram to the spectrogram, apodization is needed to suppress the side lobes, and different apodization functions have different degrees of spectral line broadening, resulting in different actual spectral resolution. In rectangular function apodization, the spectral resolution is about 1.207 times the theoretical spectral sampling interval. The effective spectral resolution of the SHRS prototype is 3.35 cm^-1. An important parameter to measure the performance of Raman spectrometer is the Signal-to-Noise Ratio (SNR) of the Raman spectrum. We choose the peak intensity of the Raman spectrum after removing the baseline as the signal intensity, and the standard deviation after removing the baseline from the Raman spectrum peak area as the noise, and use the ratio of the two as SNR of the measured Raman spectrum. In the experiment, the excitation light power is 500 mW, and the integration time is 10 s. First, the standard Raman sample cyclohexane is tested. SNR of the main Raman peak at 795.5 cm^-1 is 913, and SNR of the weakest Raman peak at 1 341 cm^-1 is 15. It can be verified that the SHRS prototype has good Raman spectrum measurement capabilities, as well as high sensitivity and SNR. Secondly, the solid samples calcium carbonate, calcium sulfate and potassium sulfate are tested. These samples are all strong Raman active substances, and the Raman spectrum peaks of various substances can be accurately identified, and SNR of the main Raman spectrum peaks is greater than 300. Finally, experiments are carried out on 75% alcohol solution, glycerin and glucose powder. The Raman activity of these samples is relatively weak, and there are obvious baselines in the measured Raman spectra, indicating that there is a certain fluorescent background in the spectra. However, a clear and accurate Raman spectrum is still obtained, and the main Raman spectrum peaks of various substances can be accurately obtained, and SNR of each spectrum peak is greater than 20. In general, SHRS has higher detection sensitivity and better stability and can meet the analytical requirements of Raman spectroscopy detection. It has certain advantages in the Raman detection of high-fluorescence background substances and has certain development potential in biomedicine, food safety, geological prospecting, planetary exploration, etc.