We demonstrate the use of distributed Raman amplification (DRA) in a dispersion decreasing fiber (DDF) for the efficiency enhancement of adiabatic soliton compression of a dual frequency beat signal. We compress a 40 GHz beat signal generated from a LiNbO3 modulator at a driving RF frequency of 20 GHz into ~ 2.2 ps soliton pulses using DRA in a 20 km DDF. The generation of high quality of soliton pulses from the 40 GHz sinusoidal beat signal is readily achieved with a significantly enhanced efficiency using DDF based DRA, compared to the case of using a DDF without DRA or a DSF with DRA.
©2004 Optical Society of America
All-optical high speed time division multiplexing (OTDM) systems operating at data rates in excess of 40 Gbit/s requires the development of stable sources of high repetition rate ultrashort optical pulses operating in the 1550 nm telecommunication band . Among many other methods one very simple and cost-effective approach to high-repetition rate short pulse generation, is to use adiabatic soliton pulse compression of a dual frequency beat signal. In this approach the dual-frequency beat signal is usually obtained by either two separate distributed feedback (DFB) lasers  or a combination of a single DFB laser and a LiNbO3 Mach-Zehnder modulator . The Mach-Zehnder modulator based dual frequency generation method is likely to be a more reliable and simple approach since the generated two optical carriers are phase-synchronized and the corresponding beat signal has a very low phase noise . In order to achieve high quality output pulses using adiabatic soliton compression of the frequency beat signal it is essential to ensure that the compression process is as adiabatic as possible. In a simple scheme based on dispersion decreasing fiber (DDF) [2,3] the dispersion lengths associated with such sinusoidal beat signal dictate the use of multi-tens of km length scale DDF’s in order to ensure that sufficiently adiabatic compression is achieved throughout the system. For such device lengths background fiber loss can have a significant impact on the compression process and limit the degree of compression that can be achieved for physically reasonable ranges of dispersion variation. In a distributed Raman amplification (DRA) based scheme  multi-tens of km of fiber is also typically required to obtain the net on-off gain needed for practical values of pump power. Recently K. Igarashi et al. proposed the use of a comb-like dispersion profiled fiber followed by DRA in a highly nonlinear fiber for achieving both effective suppression of stimulated Brillouin scattering and adiabatic compression efficiency enhancement . Furthermore, several research groups suggested a novel concept of the use of DRA in DDF to achieve high quality adiabatic soliton compression [7, 8]. This approach enhanced the compression factor for which high quality adiabatic compression was achieved for seed pulse sources carved by an electro-absorption modulator (EAM), and reduced the pump power requirements relative to those required for pure DRA compression.
In this paper we numerically and experimentally demonstrate the use of distributed Raman amplification in a dispersion decreasing fiber for the efficiency enhancement of adiabatic soliton compression of a dual frequency beat signal. We compress a 40 GHz beat signal generated from a LiNbO3 modulator at a driving RF frequency of 20 GHz into ~ 2.2 ps soliton pulses using DRA in a 20 km DDF. The output pulses are then compared to those generated from the DDF without DRA in terms of pulse pedestal level. The same experiments are also performed with a 20 km DSF for the overall performance comparison. The generation of high quality of soliton pulses from the 40 GHz sinusoidal beat signal is readily achieved with an enhanced efficiency using distributed Raman amplification in a DDF.
2. Experimental configuration
The schematic of our experiment is shown in Fig. 1(a). A continuous wave (CW) DFB laser was first phase-modulated at a RF frequency of 200 MHz to suppress stimulated Brillouin scattering (SBS) in the dispersion decreasing fiber used in this experiment. The output beam from the DBF laser was then externally modulated using a LiNbO3 Mach-Zehnder modulator which was biased at a null transmission point and driven at a RF frequency of 20 GHz to obtain a 40 GHz sinusoidal optical beat signal. A low-noise erbium doped fiber amplifier (EDFA) followed by a bandpass filter was used to amplify the modulated output beam and another high power EDFA was then employed to boost the optical power of the beat signal (~20 dBm average power) sufficient to form the first-order soliton at the input end of our DDF. The DDF used in this experiment had a length of 20 km and its dispersion followed an exponentially tapered profile at 1550 nm along the length from 6 to 1.75 ps/nm-km as shown in Fig. 1(b).
The dispersion slope was 0.057 ps/nm2-km. The fiber was originally designed and fabricated for loss-compensated soliton transmission applications and thus the dispersion profile was matched to the power loss of 0.27 dB/km within the fiber . Further details concerning the characterization and fabrication of this DDF are provided in Ref. . Three laser diodes operating at center wavelengths of 1430, 1460, and 1490 nm, respectively were employed for the Raman pump. After combing the three pump laser outputs using a fiber based 14XX/C-band WDM coupler, a total pump power of up to 500 mW could be launched into the DDF in a counter-propagating geometry and this level of pump was observed to be able to provide up to 8 dB of on-off gain within the fiber which was sufficient to compensate for the total background fiber.
3. Numerical analysis
In order to validate theoretically the proposed scheme the above system was modeled as a function of the key parameters of the DDF such as distributed dispersion profile, loss, length, and output Raman on-off gain. We first solved the following set of two coupled equations between the time-averaged signal intensity (Is) and the counter-propagating pump intensity (Ip) to figure out the signal power variation along the DDF .
where αs and αp represent the fiber loss at the signal and pump wavelengths (λs,λp), respectively, and gR is the Raman gain coefficient. The distributed gain-loss coefficient (ρs(z)) at the signal wavelength (λs) which accounts for the combined effect of on-off Raman gain and fiber loss, was then obtained by calculating the following simple equation.
where the pulse amplitude A is normalized such that |A|2 represents the optical field strength within the fiber. β 1 is the group delay. β 2 and β 3 are the first and the second order group velocity dispersion (GVD), respectively. is the nonlinearity coefficient where n 2 = 2.3×10-20 m 2/W is the nonlinear-index coefficient, ωo is the signal frequency, and Aeff = 61 ~ 50μm 2 is the effective area of the fiber. We solved the above equation using the split step Fourier method .
Figure 2 shows the numerically calculated results of 40 GHz adiabatic soliton pulse compression of the dual frequency beat signal using the DDF based DRA. It is clearly evident that the high quality of transform-limited soliton pulses with a ~ 2.2 ps temporal width is readily achieved with such a scheme. Some degree of pedestals of the compressed output pulses as shown in Fig. 2(a) are believed to be mainly due to the non-optimized dispersion profile of the DDF for adiabatic soliton pulse compression with DRA. Note that this DDF was originally designed for loss-compensated transmission of ~ 3.5 ps soliton pulses  without considering its use for adiabatic soliton compression using DRA.
4. Experimental results
We subsequently performed a range of experimental measurements to confirm the numerically predicted results. Figure 3 shows the measured second harmonic generation (SHG) autocorrelation traces of the compressed pulses after the DDF for both the case with and without DRA. High-quality, low pedestal soliton pulses were obtained with DRA in the DDF as predicted, whilst both high pedestal level and low compression factor were observed without DRA. Note that the broad background pedestals in the autocorrelation traces are mainly due to the poor extinction ratio of the LiNbO3 modulator used in this experiment and the non-uniform pulse intensity distribution can be attributed to the residual 20 GHz frequency components of the beat signal caused by the imperfect DC bias setting of the modulator as shown in the optical spectrum in Fig. 4(b). The temporal shape of the output compressed pulses with DRA in the DDF was observed to be a hyperbolic secant with a full width at half maximum (FWHM) of ~ 2.2 ps as shown in Fig. 4(a). The corresponding optical spectrum of the pulses is shown in Fig. 4(b). The solid line over the measured spectrum shows a least squares sech2 fit, from which the spectral bandwidth was estimated to be about 1.1 nm. The time-bandwidth product of the pulses is 0.31, indicating that the output compressed pulses are almost transform-limited solitons. Excellent agreement between theory and experiment was observed in terms of temporal shape, duration, and spectral bandwidth of the output pulses.
In order to confirm the enhanced performance of adiabatic soliton compression using the proposed scheme we performed the same experiment with a 20 km long dispersion shifted fiber (DSF) with a GVD of 3.8 ps/nm-km. This DSF was specially chosen among those limitedly available in our lab considering the average GVD of the DDF even if it might not be the optimum fiber for fair comparison with the DDF. However, we believe that it would not be straightforward to induce significant adiabatic soliton compression even in a 20 km long, optimized DSF with a ~8 dB on-off Raman gain which is just enough for fiber loss compensation. These results are summarized in Fig. 5. Although distributed Raman amplification along the DSF initiated an adiabatic soliton train formation in some degree, significant pulse compression process was not observed in terms of compression factor and pulse pedestal compared to the case of DRA in the DDF since most of DRA gain in the DSF was used only for compensation of the fiber loss. Without DRA any pulse compression effect was not observed after the DSF. The output pulses after the DSF with DRA was observed to have a Gaussian shape with a FWHM of 5.8 ps as shown in Fig. 6(a). According to the corresponding measured optical spectrum of the pulses in Fig. 6(b) it is clearly evident that sufficient adiabatic soliton compression process was not induced in the DSF despite the use of DRA.
We have experimentally demonstrated the efficiency enhancement of adiabatic soliton compression of a dual frequency beat signal using dispersion decreasing fiber based distributed Raman amplification. Highly efficient adiabatic compression of a 40 GHz beat signal generated from a Mach-Zehnder modulator into a ~ 2.2 ps soliton pulse train, was achieved using DRA in a 20 km of DDF. The use of a DDF with an optimized dispersion profile and a high extinction ratio LiNbO3 modulator should allow for providing further improvements in terms of compression efficiency and background pedestals. The dispersion profile should be optimized considering distributed Raman gain, fiber loss, beat signal frequency, and beat signal power.
References and links
1. E. Ciaramella, G. Contestabile, A, D’Errico, C. Loiacono, and M. Presi, “High-power widely tunable 40GHz pulse source for 160-Gb/s OTDM systems based on nonlinear fiber effects,” IEEE Photon. Technol. Lett. 16, 753–755 (2004). [CrossRef]
2. A. V. Shipulin, E. M. Dianov, D. J. Richardson, and D. N. Payne, “40 GHz soliton train generation through multisoliton pulse propagation in a dispersion varying optical fiber circuit,” IEEE Photon. Technol. Lett. 6, 1380 – 1382 (1994). [CrossRef]
3. A. D. Ellis, W. A. Pender, T. Widdowson, D. J. Richardson, R. P. Chamberlin, and L. Dong, “All-optical modulation of 40-GHz beat frequency conversion soliton source,” Electron. Lett. 31, 1362–1364, 1995. [CrossRef]
4. E. A. Swanson and S. R. Chinn, “40-GHz pulse train generation using soliton compression of a Mach-Zehnder modulator output,” IEEE Photon. Technol. Lett. 7, 114–116 (1995). [CrossRef]
5. P. V. Mamyshev, S. V. Chernikov, and E. M. Dianov, “Generation of fundamental soliton trains for high-bit-rate optical fiber communication lines,” IEEE J. Quantum Electron. 27, 2347–2355 (2002). [CrossRef]
6. K. Igarashi, H. Tobioka, S. Takasaka, S. Matsushita, and S. Namiki, “Duration-Tunable 100-GHz Sub-Picosecond Soliton Train Generation Through Adiabatic Raman Amplification in Conjunction with Soliton Reshaping,” in Proc. Optical Fiber Communications Conference (OFC 2003), Paper TuB6 (2003).
7. I. Morita, N. Edagawa, M. Suzuki, S. Yamamoto, and S. Akiba, “Adiabatic soliton pulse compression by dispersion decreasing fiber with Raman amplification,” in Proc. Opto-Electronics & Communications Conference, 17P-16 (1996).
8. T. Kogure, J. H. Lee, and D. J. Richardson, “Wavelength and duration tunable 10 GHz, 1.3 ps pulse source using dispersion decreasing fiber based distributed Raman amplification,” IEEE Photon. Technol. Lett. 16, 1167–1169 (2004). [CrossRef]
9. D. J. Richardson, R. P. Chamberlin, L. Dong, and D. N. Payne, “High quality soliton loss-compensation in 38km dispersion-decreasing fibre,” Electron. Lett. 31, 1681–1682 (1995). [CrossRef]
10. N. G. R. Broderick, D. J. Richardson, and L. Dong, “Distributed dispersion measurements and control within continuously varying dispersion tapered fibers,” IEEE Photon. Technol. Lett. 9, 1511–1513 (1997). [CrossRef]
11. G. P. Agrawal, Nonlinear fiber optics (Academic Press, 1995), Chap. 2.