Open Access
Add to BibSonomyAbstract
We investigate the properties of a self-pulsating fiber cavity based on cascaded regeneration. The mechanisms that govern the number of oscillating pulses in the cavity, the pulse peak power, the pulse width, the wavelength tunability as well as the generation of sub-picosecond pulses are identified, analyzed and quantified. We find that the described self-pulsating cavity enables the oscillation of quasi transform-limited pulses with a pulsewidth of 4.8 ps at 1540.0 nm when using 0.4 nm non-Gaussian bandpass filters. Sub-picosecond pulses with an autocorrelation width of 471 fs are generated from the same self-pulsating source with modified bandpass filters and the addition of a chromatic dispersion compensator. The number of eigenpulses that oscillate simultaneously in the cavity can be adjusted from 0 up to 29,500 with proper cavity adjustment. This source has dual-wavelength output and can be tuned throughout the gain band of the amplifiers.
©2009 Optical Society of America
1. Introduction
Self-pulsating optical sources are of great interest owing to their application in telecommunication systems as well as for time-resolved spectroscopy. Self-pulsating sources are mostly based on the principle of mode locked laser oscillation. Passively mode-locked sources are achieved by exploiting the nonlinear power transfer characteristics of a saturable absorber (SA) [1], a saturable absorber mirror (SAM) [2,3], nonlinear polarization evolution (NPE) [4–6], an optoelectronic oscillator (OEO) in a fiber extended cavity [7–9], and a nonlinear amplifying loop mirror (NALM) [10–12].
A promising alternative to self-pulsating sources arises from 2R optical regenerators [13–15] in cascade. It was demonstrated theoretically that 2R optical regenerators placed in cascade can lead to the formation of eigenpulses, that is, pulses that remain unchanged between the input and the output of the paired regenerators [16,17]. Recently, we demonstrated experimentally the feasibility of self-pulsation from paired regenerators in a closed loop [18]. In contrast with the passively mode-locked lasers mentioned above, the paired regenerators do not lead to laser oscillation but regenerate circulating pulses twice per cavity round trip. This provides the pulses with an eigenpulse profile that minimizes the propagation losses and that remains stable over time. A distinctive property of this source is that several pulses can oscillate simultaneously in the cavity and in such a case, the spatial distribution of these pulses in the fiber is random but stable through time [18]. Such an aperiodic source can also find applications in nonlinear components characterization since the pulse profile is well defined and pulse buffering as proposed in [18].
In this paper, we present theoretical and experimental analysis of the output characteristics of the self-pulsating source based on cascaded regeneration. In particular, the mechanisms that govern the number of oscillating pulses in the cavity, the pulse peak power, the pulse width, the wavelength tunability as well as the generation of sub-picosecond pulses are identified, analyzed and quantified. The remainder of this paper is organized as follows: we present the experimental setup of this source and principle of operation in section 2. The simulation model is discussed in Section 3. Experimental results are provided in Section 4. Finally, conclusions are drawn in Section 5.
2. Experimental setup and principle of operation
Figure 1(a) shows the paired regenerators in cascade which leads to the formation of eigenpulses, as demonstrated theoretically in [16]. Figure 1(b) shows our contribution of inserting paired regenerators in a closed loop and thus leading to self-pulsation. Overall, the assembly consists of two cascaded regenerators with tunable output signal wavelengths. Figure 1(c) shows the experimental setup which is further simplified compared to Fig. 1(b) by implementing the bidirectional design. The source is assembled out of the following components: a Highly Non-Linear Fiber (HNLF), two low-power Erbium-Doped Fiber Amplifiers (EDFAs), two optical circulators, two optical bandpass filters (BPFs), and five tap couplers to extract the signal output (O) and for monitoring purposes (M). Specifications of these components are as follows. The HNLF has a length=1007 m, insertion loss=1.86 dB, nonlinearity coefficient=12.5 W/-1-km-1, chromatic dispersion=-0.69 ps/nm-km @ 1550.0 nm, and chromatic dispersion slope=0.0074 ps/nm2-km @ 1550.0 nm. The amplifiers have saturated output power=~15 dBm, and both BPFs have the same spectral profile with a full-width at half maximum (FWHM) bandwidth=0.9 nm or 0.4 nm. The center wavelength of BPF2 is fixed at λ BPF2=1540.0 nm whereas λ BPF1 is changed during the experiment. The wavelength difference between both filters is defined as the filter offset, FO=λ BPF2-λ BPF1.
Fig. 1. (a). Scheme of paired regenerators in cascade. (b). Scheme of paired regenerators in closed loop. (c). Experimental setup of the self-pulsating source. (d). Spectrum representation at various point of the setup to illustrate the operation principle of the source in SP and PB regimes. O: Laser Output, M: Monitor output, C: Circulator, PSD: Power Spectral Density, FO: Filter Offset.
The experiments are typically performed as follows: Both central frequencies of BPF1 (λ BPF1) and BPF2 (λ BPF2) are superimposed to enable CW operation. Then λ BPF1 is wavelength-shifted until self-pulsation occurs while λ BPF2 is held constant. Signals from the output and monitor couplers are sampled and analyzed at various values of λ BPF1 using a digitizing oscilloscope and an optical sampling module with bandwidth=30 GHz, an autocorrelator, an optical spectrum analyzer, and a power meter.
This self-pulsating source operates among three regimes: continuous-wave (CW), self-pulsating (SP) and pulse-buffering (PB) [18]. In the SP regime, self-pulsation occurs spontaneously from the amplified spontaneous noise (ASE) of the amplifiers. In contrast, the PB regime does not allow pulses to ignite from ASE but can only support the oscillation of pulses that already circulate or that are being injected in the cavity. Figure 1(d) schematically illustrates the operation principle of this self-pulsating source in SP and PB regimes from an optical spectrum processing viewpoint: A pulse propagating in the HNLF (from C2 to C1) experiences self-phase modulation (SPM) spectral broadening (as observed from M1) before offset spectral filtering with BPF1. The regenerated pulse (as observed by O1) is then amplified by A1 and propagates through the HNLF (from C1 to C2) before passing through the second regeneration stage. The pulse circulating in the cavity is then regenerated twice per round trip.
The SPM based 2R regenerator comprises an Amplifier + HNLF + BPF and provides a nonlinear power transfer function in the cavity. This nonlinear power transfer function acts as the pulse shaper and stabilizer, and plays a role similar to a saturable absorber. The cavity is not considered as passively mode-locked, since there is no locking of modes due to the wavelength shifting of the circulating pulses. For this reason, the use of the term “laser” is avoided to describe the self-pulsating cavity. Pulses that are generated spontaneously from ASE are regenerated twice per round trip and evolve into eigenpulses, maintaining their amplitude and profile after each round trip.
3. Simulations
A theoretical model of this self-pulsating source was developed and validated by comparing the experimental results with computer simulations. Simulations allow us to follow the signal profile as it propagates from a regenerator pair to the next. A theoretical analysis exploring the impact of different regenerator parameters has already been performed in [16,17] and thus this paper is focused on the theoretical analysis of the regenerator pair using physical parameters as close as possible to those of the experimental setup.
The HNLF combines simultaneously the effects of nonlinearity and chromatic dispersion. These are described mathematically with the nonlinear Schrödinger equation
and solved with the split-step Fourier method [19]. A is the slowly-varying envelope of the THz oscillating electric field, α is the attenuation coefficient, β2 and β3 are the chromatic dispersion coefficients of the first and second order, respectively, γ is the nonlinearity coefficient, z is the longitudinal coordinate in the HNLF, and T is the time relative to a frame of reference moving at the group velocity.
The EDFAs are modeled as a gain medium characterized by a small signal gain (G0) and a saturation input power (PSat) [20]. The output power from such a medium can be expressed as
where POut and PIn represent the output and input power, respectively. Both PSat and G0 are extracted from experimental characterization of the EDFAs.
After the HNLF and the EDFAs, a third important component of the simulation is the BPFs spectral profile. Both BPF1 and BPF2 have the same spectral profile shape and FWHM of 0.4 nm or 0.9 nm. In the rest of this paper, the spectral profiles of the BPFs used in the simulation are implemented directly from experimental characterization unless specifically stated. The experimental characterization of both the EDFAs power transfer function and the BPFs spectral responses are shown in Fig. 2.
Fig. 2. Experimental characterization of the EDFAs (Left figure) and the bandpass filters (Right figure). The amplifiers were characterized at 1540.0 nm. Both BPF1 and BPF2 have the same spectral profile and FWHM of 0.4 nm or 0.9 nm.
4. Results and discussion
In this section, we demonstrate the mechanisms that govern the operation of the pulsed source in terms of the following: the repetition rate and number of pulses that are circulating in the cavity; the peak power; the pulse width; the generation of sub-picosecond pulses; the dual wavelength operation and wavelength tunability.
4.1 Repetition rate and number of pulses in the cavity
This self-pulsating source does not have a fixed repetition rate since the pulses self-seeded from ASE are numerous and are randomly spaced in the cavity, as can be observed on the oscilloscope. However, by properly decreasing the cavity length as well as the gain in the cavity, it is possible to have a single pulse per round trip, and this is consolidated by the simulation results as shown in Fig. 3. Figure 3(b) and Fig. 3(c) show the output of the same incoherent wave after circulating 40 times in the cavity, with gain of the EDFAs set to 120.0 and 105.7 for Fig. 3(b) and Fig. 3(c), respectively. Other parameters for the simulations are as follows: BPFs bandwidth=0.9 nm, λ BPF2=1540.0 nm, and FO=4.0 nm. In our experiment, single pulse per round trip can’t be achieved because the cavity is too long. The round trip time in the cavity is 10.4 µs as calculated from the ≈ 2 km fiber propagation length. However, the 1007 m of HNLF used in the cavity can be replaced by a shorter length of medium which has larger nonlinearity such as ~1 m of chalcogenide or bismuth fiber, thus leading to shorter cavity round trip time and probably the operation of single pulse per round trip. Although pulses are randomly spaced in the cavity, simulations have shown that the eigenpulses that oscillate in the cavity maintain their relative position and keep a stable profile due to the operation of the regenerators.
Fig. 3. (a) Input incoherent wave. Output after circulating 40 loops when (b) EDFAs gain=120.0 and (c) EDFAs gain=105.7.
The number of pulses that are circulating in the cavity can be calculated from the ratio of the average energy that flows through one fixed point in the cavity in one round trip time and the energy of a single eigenpulse at that specific point. By integrating the experimental spectrum after the HNLF, the average power after the HNLF in the cavity can be retrieved, and the average energy is then calculated by the product of the average power and the round trip time. The energy of a single eigenpulse is retrieved from simulation by fitting the broadened spectrum between simulation and experiment. Figure 4 shows the retrieved number of pulses as a function of FO when 0.9 nm BPFs are used with λ BPF2 fixed at 1540.0 nm. Increasing FO leads to a decreasing number of pulses stored in the cavity. This monotonous decrease leads to the conclusion that pulses are being eliminated in this process and thus the number of remaining pulses indicates the maximum pulse storage capacity of the cavity. The decrease of the number of pulses can be understood from the fact that larger FO requires more spectral broadening, and thus larger peak power of eigenpulses. However, the amplifier provides a relatively constant amount of power in the cavity, so the peak power of eigenpulses will increase at the expense of the number of eigenpulses. In contrast, when FO is decreasing in PB regime, the number of stored pulses remains almost constant as it is below the maximum storage capacity. The trivial decrease of the number of pulses is caused by the perturbation that is introduced to the cavity when changing FO. When FO is further decreased and the cavity enters into the SP regime, the number of pulses will increase. The largest storage capacity of this self-pulsating cavity was found to be 29,500 pulses for FO=2.0 nm. The number of pulses for FO<2.0 nm could not be retrieved as the spectral broadening was insufficient to make a theoretical fit.
Fig. 4. Experimentally retrieved number of pulses as a function of FO with 1 nm step size. An increase of FO leads to a decrease in the maximum number of oscillating pulses. However, the number of pulses remains relatively constant when FO is decreasing, since it is below the maximum capacity. The number of pulses increases when FO is further decreased and the cavity enters into SP regime.
4.2. Output eigenpulse power
An in-depth analysis of the power transfer function (PTF) was performed to explain the origin of the eigenpulse peak power selected by the cavity. Figure 5(a) shows PTFs of one regeneration stage comprising the HNLF and BPF using the 0.4 nm filters and with FO=2.0 nm or 4.0 nm; Fig. 5(b) shows the case for 0.9 nm filters. For Fig. 5(a) and Fig. 5(b), an external source with an input wavelength λ in=1540.0 nm and TFWHM=6.2 ps was used for the characterization. The oscillations in the PTFs come from the peaks and valleys in the SPM broadened spectrum [14], and a BPF with a broader bandwidth will reduce such oscillations. In fact, it can be seen from Fig. 5(a) and Fig. 5(b) that the oscillations in the PTFs have been reduced when using 0.9 nm BPFs, compared to 0.4 nm ones. The agreement between PTFs from experiments and simulations is excellent. It is also worth noting that when it comes to the discussion of eigenpulse in the following sections, it is referred to the pulses that are self-seeded from ASE in the cavity. Under the condition that the BPFs bandwidth=0.9 nm, λ BPF2=1540.0 nm, and FO=4.0 nm, the simulated peak power and energy for the eigenpulse at O2 is 49.6 mW and 1.66×10-13 J, respectively. To further verify the results, the loop is opened at the point after O2, and the simulated eigenpulse is used as the input pulse to calculate the PTF for this opened loop. The PTF for the opened loop from the input of A2 is obtained and shown in Fig. 5(c), and as can be seen, the PTF has a square-like profile due to the amplifiers. The peak power of the eigenpulses self-stabilizes at the crossing point where the input peak power equals to the output peak power, which is 49.6 mW, in agreement with the peak power simulated for the eigenpulses. As shown in Fig. 5(d), the eigenpulse at output O2 was also measured using an optical sampling module with 16 ps impulse response time connected to an oscilloscope. By integrating the trace shown on the oscilloscope, the energy extracted for one eigenpulse is 1.48×10-13 J, which compares well to the simulated energy for one eigenpulse of 1.66×10-13 J.
Fig. 5. (a). PTFs of HNLF+BPF when BPF bandwidth is 0.4 nm, input wavelength is 1540.0 nm, and FO=2.0 nm or 4.0 nm. (b) PTFs of HNLF+BPF when BPF bandwidth is 0.9 nm, input wavelength is 1540.0 nm, and FO=2.0 nm or 4.0 nm. (c) Simulated PTF for the opened loop from the input of A2 using the eigenpulse obtained when BPFs bandwidth=0.9 nm, λ BPF2=1540.0 nm, and FO=4.0 nm. (d) Experimental pulse profile from an oscilloscope after 13.4 dB attenuation. The insets in (a), (b), and (c) show the setup used to obtain the PTFs.
4.3. Output eigenpulse profile
The spectral profile of output eigenpulses is mainly determined from the profile of the BPFs in the cavity, and in this paper, the analysis is focused on the eigenpulses profile in the time domain. Figure 6 shows the autocorrelation trace of output eigenpulses at O2 when λ BPF2=1540.0 nm and FO=4.0 nm, both theoretically and experimentally, with reasonable agreement between the two. As shown in Table 1, when the bandwidth for BPF1 and BPF2 is 0.4 nm, the FWHM of the eigenpulse temporal profile in simulation is 4.8 ps, with a corresponding FWHM after autocorrelation of 8.7 ps, compared to 9.6 ps in experiment. In contrast, when the bandwidth for BPF1 and BPF2 is 0.9 nm, the FWHM of the eigenpulse temporal profile in simulation is 3.1 ps, with a corresponding FWHM after autocorrelation of 4.2 ps, compared to 4.3 ps in experiment. Thus, as expected, shorter pulses are formed by using spectrally wider BPFs. Note that the differences between the experimentally characterized spectral profiles of the 0.4 nm BPFs and 0.9 nm BPFs lead to the different conversion factors between the pulse width and autocorrelation width. After the offset spectrum filtering by the BPFs, the eigenpulse temporal profile is determined by the combination of the inverse Fourier transform of the filter transmission spectrum and the phase-shift accumulated from chromatic dispersion and SPM. Chromatic dispersion and nonlinear phase shift will contribute to a residual chirp at the output O1 and O2, which means that the output pulse is not transform-limited.
Fig. 6. Experimental and simulated autocorrelation trace of output eigenpulses for 0.9 nm filters and 0.4 nm filters when λ BPF2=1540.0 nm and FO=4.0 nm.
Table 1. Output pulse width with different BPFs bandwidth
Figure 7 presents the simulation results on the pulse width versus filter bandwidth. The experimentally measured filter profile cannot be fit by a single function (e.g., Lorentzian or Gaussian) but here a Gaussian filter profile is used as a common link to all simulations. The red solid line in Fig. 7(a) shows the pulse FWHM as a function of the filter bandwidth. As can be observed, the pulse width decreases with increasing filter bandwidth, up to a certain bandwidth for which the pulse width then increases with increasing filter bandwidth. The black solid line in Fig. 7(a), on the other hand, shows the time-bandwidth product of output eigenpulses as a function of the filter bandwidth. The time-bandwidth product increases with increasing filter bandwidth. This indicates an increasing residual chirp contained at O2. It is also found in simulation that this increasing residual chirp can be compensated by adding an external chromatic dispersion compensator after the BPF2. The red dashed line and blacked dashed line in Fig. 7(a) show the pulse width and time-bandwidth product after external dispersion compensation, respectively. In this case, the pulse width decreases with increasing filter bandwidth and the value of time-bandwidth product is constant around 0.44, which indicates that the compressed pulses are nearly transform-limited. Figure 7(b) shows the simulated eigenpulse profile after BPF2 with/without external dispersion compensation when the filter bandwidth is 500 GHz. The dispersion compensator used was 45.5 m of single mode fiber (SMF) with D=17 ps/nm-km. Pulse FWHM without dispersion compensation is 3.1 ps, compared to 0.9 ps after dispersion compensation, corresponding to a time-bandwidth product of 0.46. The time-bandwidth product after dispersion compensation is not exactly 0.44, which comes from the fact that the output spectrum after the filter is not exactly the same as the Gaussian filter profile.
Fig. 7. (a). Simulated pulse width vs. Gaussian filter bandwidth and time-bandwidth product vs. Gaussian filter bandwidth without/with external dispersion compensation after BPF2, when λ BPF2=1540.0 nm and FO=4.0 nm. (b) Eigenpulse profile after BPF2 without/with dispersion compensation when Gaussian filter bandwidth is 500 GHz.
4.4. Generation of sub-picosecond pulses
Using numerical simulations, an optimal pulse-compression is determined at M2, where the spectrum is broadened up to 20 nm after SPM spectral broadening in the HNLF. Figure 8(a) shows the simulated RMS pulse width after external dispersion compensation at M2 vs. length of SMF used for dispersion compensation. The parameters are as follows: λ BPF2=1550.3 nm, λ BPF1=1553.3 nm, and BPFs bandwidth=0.9 nm. It is found in simulation that 38.1 m of SMF can give small RMS pulse width as well as good waveform quality after dispersion compensation. The simulated pulse and phase profile before and after dispersion compensation are presented in Fig. 8(b). Due to the large nonlinearity and small normal dispersion in the HNLF, the pulse profile has a square like shape before dispersion compensation, which is a typical phenomenon of optical wave breaking [19]. After dispersion compensation, the pulse width is compressed down to the sub-picosecond regime, and the quadratic phase becomes almost flat at the center of the pulse.
Fig. 8. (a). Simulated RMS pulse width after external dispersion compensation at M2 vs. length of SMF used for dispersion compensation when BPFs bandwidth=0.9 nm, λ BPF2=1550.3 nm, and λ BPF1=1553.3 nm. (b) Simulated pulse and phase profile before and after dispersion compensation with 38.1 m of SMF.
The ability to generate sub-picosecond pulses after the HNLF was also demonstrated experimentally. Using the same parameters as in the simulation, the experimental spectrum at M2 is given in Fig. 9(a), and the output pulses at M2 were externally compressed with 36.3 m of SMF. In Fig. 9(a), the two peaks localized around 1550 nm correspond to the filtered ASE from the amplifier. In Fig. 9(b), the blue circle is the autocorrelation trace after dispersion compensation in experiment, the red line is the autocorrelation trace after dispersion compensation in simulation, and the black line is the autocorrelation trace of the transform-limited pulse calculated from the spectrum shown in Fig. 9(a). The compressed pulses have an autocorrelation width of 471 fs. The wings in the autocorrelation trace come from uncompensated nonlinear phase shift that is induced by SPM as well as uncompensated higher order dispersion, which results mainly from the third order dispersion (TOD) generated by the HNLF, the SMF in the cavity, and the SMF used for dispersion compensation.
Fig. 9. (a). Experimental spectrum at M2 when BPFs bandwidth=0.9 nm, λ BPF2=1550.3 nm, and λ BPF1=1553.3 nm. (b) Autocorrelation trace with external dispersion compensation after M2 in experiment, simulation, and calculated from the spectrum in (a).
4.5. Dual wavelength operation and wavelength tunability
The wavelength tunability of the self-pulsating source was tested throughout the C-band. For this purpose, both filters are tuned accordingly and operation is analyzed for various values of the FO parameter. Figure 10 shows spectra for the self-pulsating source operating from λ BPF2=1535.0 nm up to 1560.0 nm with constant FO=-2.0 nm and BPFs bandwidth=0.9 nm. The wavelength range that can be tested is actually limited by the tunability of the BPFs in the experiment. The profile of typical SPM broadened spectra, as manifested by the four traces shown in Fig. 10, shows that the self-pulsating source can operate anywhere throughout the C-band; however, the source properties differ from one set of BPFs wavelengths to another. For multiwavelength applications, this self-pulsating source could provide more than two distinct wavelengths by increasing the number of regenerators in cascade.
Fig. 10. Spectra taken from the monitor output M1 showing self-pulsating operation over the C-band with constant FO=-2.0 nm and BPFs bandwidth=0.9 nm.
Table 2 shows the range of values of FO enabling the CW, SP, and PB regimes [18] for each of the selected values of λ BPF2 throughout the C-band, and the results are reproducible. Overall, the self-pulsating source can be operated in CW, SP, or PB regimes at any arbitrary wavelength of the C-band by using the appropriate adjustment of FO. The three regimes of operation are observed for positive as well as for negative values of FO, however, the absolute values of FO defining the three regimes are different, which comes from the fact that opposite values of FO will make BPF2 filter out the different part of the chirp accumulated from SPM, thus leading to different dynamics of chirp accumulation in the cavity. The range of values of FO enabling the CW, SP, and PB regimes also differ for the two sets of BPFs, considering the larger bandwidth, smaller insertion loss, and steeper edges of the 0.9 nm filters, compared to the 0.4 nm ones.
Table 2. Range of values of FO [nm] enabling the CW, SP and PB regimes of operation when BPFs bandwidth=0.9 nm
Fig. 11. (a). Output autocorrelation width vs. FO when BPFs bandwidth=0.9 nm and λ BPF2=1540.0 nm. (b) Output autocorrelation width vs. λ BPF2 when BPFs bandwidth=0.9 nm and FO=4.0 nm.
Experimental and simulated analysis of the autocorrelation width at O2 vs. FO and vs. λ BPF2 is provided in Fig. 11. The error bars shown in Fig. 11 indicate a ±5% noise superimposed to data as measured on the autocorrelator. As can be seen in Fig. 11, the simulated pulse autocorrelation width falls within the error bars of the experimental characterization, and the pulse FWHM is relatively independent of both FO and working wavelength. Since the pulse FWHM is found to be independent of FO, it can be concluded that the group delay accumulated by the pulse at the output O2 originates mostly from chromatic dispersion and the group delay accumulated from SPM, which should be strongly FO-dependent, must be negligible.
5. Conclusion
We have highlighted the origins of pulse properties of a self-pulsating regenerative fiber cavity. Up to 29,500 pulses can circulate in the cavity when FO is 2.0 nm. The peak power of the eigenpulses is determined from the power transfer function of one round trip, and the spectral profile of the eigenpulses is mainly determined from the profile of the bandpass filters. The eigenpulse after the bandpass filters is close to transform-limited but may need dispersion compensation when bandpass filters with relatively large bandwidth are being used, i.e., from 200 GHz and above with the current setup. Also, the addition of an external chromatic dispersion compensator after the HNLF enables the formation of pulses down to the sub-picosecond regime.
This self-pulsating source has dual-wavelength output and can be tuned throughout the gain band of the amplifiers used. Other gain media such as semiconductor amplifiers or Raman amplifiers could be used as substitutes for EDFAs. Such amplifiers provide the potential for operation in wavelength ranges beyond the C-band.
The unique similarity from one pulse to the next inherent to the eigenpulse formation in this cavity, the possibility of controlling the nearly transform-limited eigenpulse profile, the feasibility to generate sub-picosecond pulses, the multiwavelength operation and wavelength tunability, as well as the simplicity of the design make this optical self-pulsating source an appealing alternative to the conventional self-pulsating sources.
Acknowledgments
This work was supported by the Natural Science and Engineering Research Council of Canada (NSERC) and the Fonds Québecois pour la Recherche sur la Nature et les Technologies (FQRNT).
References and links
1. K. Kieu and F. W. Wise, “All-fiber normal-dispersion femtosecond laser,” Opt. Express 16, 11453–11458 (2008). [CrossRef] [PubMed]
2. C. Nielsen, B. Ortaç, T. Schreiber, J. Limpert, R. Hohmuth, W. Richter, and A. Tünnermann, “Self-starting self-similar all-polarization maintaining Yb-doped fiber laser,” Opt. Express 13, 9346–9351 (2005). [CrossRef] [PubMed]
3. B. Ortaç, M. Plötner, J. Limpert, and A. Tünnermann, “Self-starting passively mode-locked chirped-pulse fiber laser,” Opt. Express 15, 16794–16799 (2007). [CrossRef] [PubMed]
4. M. E. Fermann, L.-M. Yang, M. L. Stock, and M. J. Andrejco, “Environmentally stable Kerr-type mode-locked erbium fiber laser producing 360-fs pulses,” Opt. Lett. 19, 43–45 (1994). [CrossRef] [PubMed]
5. F. Ö Ilday, J. R. Buckley, H. Lim, F. W. Wise, and W. G. Clark, “Generation of 50-fs, 5-nJ pulses at 1.03 µm from a wave-breaking-free fiber laser,” Opt. Lett. 28, 1365–1367 (2003). [CrossRef]
6. O. Prochnow, A. Ruehl, M. Schultz, D. Wandt, and D. Kracht, “All-fiber similariton laser at 1 µm without dispersion compensation,” Opt. Express 15, 6889–6893 (2007). [CrossRef] [PubMed]
7. X. S. Yao, L. Davis, and L. Maleki, “Coupled optoelectronic oscillators for generating both RF signal and optical Pulses,” J. Lightwave Technol. 18, 73–78 (2000). [CrossRef]
8. J. Lasri, P. Devgan, R. Tang, and P. Kumar, “Self-starting optoelectronic oscillator for generating ultra-low-jitter high-rate (10 GHz or higher) optical pulses,” Opt. Express 11, 1430–1435 (2003). [CrossRef] [PubMed]
9. W. W. Tang and C. Shu, “Self-starting picosecond optical pulse source using stimulated Brillouin scattering in an optical fiber,” Opt. Express 13, 1328–1333 (2005). [PubMed]
10. A. Avdokhin, S. Popov, and J. Taylor, “Totally fiber integrated, figure-of-eight, femtosecond source at 1065 nm,” Opt. Express 11, 265–269 (2003). [CrossRef] [PubMed]
11. J. W. Nicholson and M. Andrejco, “A polarization maintaining, dispersion managed, femtosecond figure-eight fiber laser,” Opt. Express 14, 8160–8167 (2006). [CrossRef] [PubMed]
12. Y. Zhao, S. Min, H. Wang, and S. Fleming, “High-power figure-of-eight fiber laser with passive sub-ring loops for repetition rate control,” Opt. Express 14, 10475–10480 (2006). [CrossRef] [PubMed]
13. P. V. Mamyshev, “All-optical data regeneration based on self-phase modulation effect,” in Proceedings of 24th European Conference on Optical Communications, (Madrid, 1998), pp. 475–476.
14. M. Rochette, L. Fu, V. Ta’eed, D. J. Moss, and B. J. Eggleton, “2R optical regeneration: an all-optical solution for BER improvement,” IEEE J. Sel. Top. Quantum Electron. 12, 736–744 (2006). [CrossRef]
15. M. Matsumoto, “Efficient all-optical 2R regeneration using self-phase modulation in bidirectional fiber configuration,” Opt. Express 14, 11018–11023 (2006) [CrossRef] [PubMed]
16. S. Pitois, C. Finot, and L. Provost, “Asymptotic properties of incoherent waves propagating in an all-optical regenerators line,” Opt. Lett. 32, 3263–3265 (2007). [CrossRef] [PubMed]
17. S. Pitois, C. Finot, L. Provost, and D. J. Richardson, “Generation of localized pulses from incoherent wave in optical fiber lines made of concatenated Mamyshev regenerators,” J. Opt. Soc. Am. B 25, 1537–1547 (2008). [CrossRef]
18. M. Rochette, L. R. Chen, K. Sun, and J. H. -Cordero, “Multiwavelength and tunable self-pulsating fiber cavity based on regenerative SPM spectral broadening and filtering,” IEEE Photon. Technol. Lett. 20, 1497–1499 (2008). [CrossRef]
19. G. P. Agrawal, Nonlinear Fiber Optics (Academic press, San Diego, CA, 2007), Chap. 2 & Chap. 4.
20. A. E. Siegman, Lasers (University Science books, Mill Valley, CA, 1986), Chap. 10.
