Abstract：Super luminescent diode (SLD) and amplified spontaneous emission (ASE) light sources are often used in fiber optic gyroscopes. Their wide spectrum can effectively suppress the influence of polarization noise, back reflection and Rayleigh scattering. Ripple, as an important index in SLD and ASE light sources, mainly coming from the laser part of broadband light source that is not completely suppressed, impacts the accuracy of fiber optic gyroscope. In the military standard, ripple is defined as the range of mode amplitude near the peak wavelength of the spectral curve, with dB as the unit. At present, ripple measurement is mainly based on the logarithmic spectrum observed by human eyes. However, in human eye observation, the subjective factors of a human and the external interference may lead to subjectivity and contingency of the results, and measurement errors cannot be ruled out. Additionally, the spectrum in logarithmic coordinates cannot accurately reflect the spectral characteristics. Because of this, the present study analyzed the ripple in the asymmetric spectrum of the broadband light source, compared the advantages and disadvantages of multi-peak Gaussian fitting and polynomial fitting. Finally, adopted the polynomial fitting method. Through theoretical analysis and experimental verification, it was revealed that the quintic polynomial fitting can meet the requirements. Besides, a new method which can precisely calculate the ripple based on the triple standard deviation principle was proposed. AQ6370D spectrometer and a self-made “Spectral Ripple Fitting” software were utilized to experiment with ASE and SLD light sources, and the results obtained were in good agreement with the observation data of human eyes. On this basis, repeated validation experiments were carried out, the five experimental data of the same light source were recorded, and their ripple coefficients were calculated respectively. Compared with the human eye observation results, the experimental results demonstrate that the spectral ripple coefficient algorithm can well eliminate the influence of subjectivity and contingency, indicating that it is feasible. With the self-made “Spectral ripple Test” software, the average wavelength, peak wavelength, center wavelength, spectral width, ripple, linear coordinates and logarithmic coordinates of the spectrum and their fitting curve expressions can be obtained on the interactive interface, which makes the experimental analysis more convenient and speedy, thus significantly improving the experimental efficiency. It has been proved that the spectral ripple algorithm based on asymmetric polynomial fitting proposed in this paper can effectively judge the light source ripple, improve the calculation efficiency, and eliminate the influence of subjectivity and contingency, showing a broad application prospect. Furthermore, it is a supplement to the national military standard as well.
