Abstract
Mode-locked fiber lasers based on nonlinear polarization evolution can generate femtosecond pulses with different pulse widths and rich spectral distributions for versatile applications through polarization tuning. However, a precise and repeatable location of a specific pulsation regime is extremely challenging. Here, by using fast spectral analysis based on a time-stretched dispersion Fourier transform as the spectral discrimination criterion, along with an intelligent polarization search algorithm, for the first time, we achieved real-time control of the spectral width and shape of mode-locked femtosecond pulses; the spectral width can be tuned from 10 to 40 nm with a resolution of ~1.47 nm, and the spectral shape can be programmed to be hyperbolic secant or triangular. Furthermore, we reveal the complex, repeatable transition dynamics of the spectrum broadening of femtosecond pulses, including five middle phases, which provides deep insight into ultrashort pulse formation that cannot be observed with traditional mode-locked lasers.
Highlights
Because pulse trains achieve excellent performance with a simple laser setup[1], passively mode-locked fiber lasers (MLFLs) based on nonlinear polarization evolution (NPE) have numerous applications, ranging from highresolution atomic clocks[2,3] to optical frequency measurements[4,5], photonic analog-to-digital converters (ADCs)[6], photonic radars[7], fine ranging metrology[8,9], and astronomy[10,11]
By using the TSRPC to control the spectral width and shape of the mode-locked pulses in real time, we revealed the transition dynamics from the narrowspectrum mode-locking regime to the wide-spectrum mode-locking regime and from the triangular-spectrum Q-switched mode-locking (QML) regime to the wide-spectrum mode-locking regime by recording several sets of experienced voltages of the electric polarization controllers (EPCs) that led to the different regimes
The TSRPC can be made even more portable by replacing the DCF with a small optical grating, and its spectral programming resolution can be improved by using an ADC with a higher sampling rate or a medium with large dispersion
Summary
The fitness function for spectral full width at half maximum (FWHM) programming is the spectral width of the current pulse, while the function for spectral shape programming is the normalized meansquared error (NMSE) between the current pulse and the desired pulse This method is different from traditional GA-based automatic mode-locking[13,14,15,16], where the algorithm iterates through all the given iterations and artificially checks if mode-locking is achieved via off-line measurements. This result agrees with the theory that the temporal duration decreases as the spectral width increases. The maximum spectral width is reduced by half to ~20 nm, the experimental results in Fig. 2c, d
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
