An efficient Erbium-doped fiber amplifier configured in double-pass amplification scheme with chirped fiber Bragg grating as the reflector is presented in this paper. The proposed amplifier architecture is optimized and designed to work under consideration of low pump powers for remotely-pumped applications. The chirped fiber Bragg grating is used to reflect the amplified signal back to the Erbium-doped fiber and at the same time to compensate the effect of fiber dispersion. The proposed amplifier architecture is able to maintain gain of higher than 20 dB for small signals less than -23 dBm with 10 mW pump power only. The integrated function of loss and dispersion compensator in single black box is an attractive solution to be used as pre-amplifier.
© 2006 Optical Society of America
Repeaterless transmission systems utilizing remotely-pumped optical amplifiers have attracted research interest from various research institutes. The advantage of remotely-pumped Erbium-doped fiber amplifier (R-EDFA) is geographically independent which means that the pump laser can be located at the ends of optical fiber transmission. The advancement of span engineering has enabled its deployment for longer distances. This can be achieved owing to extremely high power optical amplifier , large effective area of fiber , ultra low-loss fiber  and highly efficient R-EDFA in double-pass architecture .
Owing to the dispersion effect in optical fibers, dispersion management is required in any optical transmission systems. The amount of accumulated dispersion is linearly proportional to the transmission distance. Therefore, this value is very large in repeaterless transmission system and needs to be effectively compensated to ensure a good quality of signal at the receiving end. Normally, dispersion compensating modules are inserted in repeaterless transmission systems at both transmitter and receiver ends. In this case, the associated loss due to dispersion compensating modules is compensated by discrete EDFAs . Thus, quality of the signal is degraded in this technique due to additional noises from EDFAs. In another option, the dispersion compensating modules can be utilized as Raman amplifier , however the requirement of high pump power to get the benefit of Raman amplification is not feasible in remotely-pumped optical amplifier applications in repeaterless transmission systems.
Chirped fiber Bragg grating (CFBG) has been utilized as one of the dispersion compensating techniques . Owing to its operation in reflective mode, CFBG can be used as a reflector for double-pass EDFAs. The concept has been demonstrated for discrete amplifiers in which two 1480 nm pump lasers are used in the same amplifier box . In this case, the total pump powers of 140 mW are used in the experiment to push the amplifier into its saturation regime. Thus, the amplifier can produce high output powers for longer transmission distances. However, in order to use this amplifier structure for remotely-pumped applications, the requirement of pump power is very critical which is similar to the case of discrete Raman amplifiers previously discussed. For R-EDFA, in order to push the amplifier to operate in the saturation regime, the pump lasers in Watts region must be used at either transmitting or receiving side. In general, there are two major issues of using these high-power lasers in optical fibers; damage of connector end and fiber fuse (waveguide structure defect) . Therefore, there is a need to optimize R-EDFA performance with low pump powers so that the requirement of extremely high power lasers can be relieved.
In this paper, a double-pass optical amplifier with built-in CFBG is analyzed to have an optimum performance at low pump powers for potential use in repeaterless transmission systems. CFBG is utilized to reflect the signal and filter out large amounts of amplified spontaneous emission (ASE) and, at the same time compensates the effect of fiber dispersion. Comparison between the proposed amplifier structure and single-pass R-EDFA is performed to analyze their performance with respect to the strength of signal powers.
2. Amplifier characterizations
The proposed double-pass optical amplifier configuration is shown in Fig. 1. Normally, a reflector is used to reflect the amplified signal back into the EDF. This reflector can be built either from a mirror, Sagnac loop fiber, fiber Bragg grating or fiber loop mirror. On the other hand, CFBG can also be utilized as the signal reflector. The main advantage of having CFBG in double-pass amplifier structure is its capability of compensating fiber dispersion. In this research work, the CFBG is fabricated to compensate a total dispersion of -1327 ps/nm (75 km standard single-mode fiber) with more than 90% reflectivity at 1550.3 nm. Its full width half-maximum is measured around 0.5 nm with high isolation of more than 20 dB for 100 GHz grid spacing.
A piece of Erbium-doped fiber (EDF) is used that has an absorption coefficient of 9.2 dB/m at 1530 nm, a numerical aperture of 0.21 and a cutoff wavelength of 1420 nm. The EDF is designed to have an optimum performance for 1480 nm pumping band. Before choosing the right EDF length, the proposed R-EDFA is tested with a series of EDF lengths from the same batch of fiber. Finally, the EDF length of 13.5 m is chosen to give the highest gain compared to other lengths that are available in the laboratory. A conventional 1480 nm laser diode is deployed in the research work to investigate the performance of the proposed R-EDFA. In this research work, the pump wavelength is not optimized to get benefits from Raman amplification in the transmission fiber. A wavelength selective coupler (WSC) is used to multiplex and demultiplex the signal and pump lights. A circulator (Cir) is used as an isolator and at the same time to separate the input signal from the output signal. It is also utilized to minimize the effect of multipath interference noise in the transmission line.
Since R-EDFA is used at a certain distance from transmitter or receiver side, the requirement of low pump power is very crucial. Thus, the objective of this experiment is to determine the operating pump power of the EDFA to be deployed as a remotely-pumped optical amplifier in repeaterless transmission systems. The signal power of -27 dBm is utilized at 1550.3 nm and the pump power is varied from 5 to 40 mW. The experimental results obtained from this experiment are depicted in Fig. 2. Since the output power is proportional to the pump power, it is not the best parameter to optimize the design of optical amplifiers. Owing to this reason, the power conversion efficiency analysis cannot be applied to achieve the objective. Another parameter that can be used to measure the optimum performance of EDFA is gain coefficient. It is defined as the efficiency of optical amplifier to amplify signal restricted to the availability of pump power . In the experiment, the highest gain coefficient of 2.2 dB/mW is achieved around 10 mW pump power. In this pump power range, the noise figure (NF) is measured around 6.8 dB. By doubling the pump power from 10 to 20 mW, the signal gain is increased by 6 dB. This margin can be translated to either a higher received signal power at the receiver side for a fixed propagation loss or longer distances. However, the amount of power needed from a remote pump laser is also doubled for the former advantage. On the other hand, the latter advantage requires a rocketed amount of output from a remote pump laser. Both situations then invite unprecedented problems associated to harmful effects from high power lasers as described in Ref. . Therefore, the optimum performance of double-pass EDFA is selected at pump power of 10 mW for the remotely-pumped applications in repeaterless transmission systems.
In the next experiment, the characteristics of conventional single-pass amplifier are investigated in order to determine the efficiency of the double-pass R-EDFA with CFBG. For single-pass R-EDFA, the CFBG is replaced with an isolator. Since the same circulator can be used as unidirectional isolator, the same insertion loss can be maintained. Thus, the comparison can be made more realistically. From Fig. 3, it can be seen clearly that the gain of double-pass R-EDFA has higher gain than its counterpart for small signal power up to -18 dBm. This is owing to the efficiency of signal amplification of the signal that occurs twice and at the same time, the CFBG is used to filter out the broadband ASE. Thus, the double-pass R-EDFA is relaxed from ASE saturation as compared to the single-pass R-EDFA. However, the noise figure of double-pass R-EDFA is poorer than its counterpart owing to same mechanism of amplification. In this case, the in-band ASE noise within the signal wavelength cannot be effectively filtered out and this ASE noise is also amplified twice together with the signal. Therefore, the double-pass R-EDFA accumulates higher noise within the signal band than its counterpart of single-pass R-EDFA. In order to further evaluate the characteristics of both amplifiers, a figure of merit (FOM) is adopted in the experiment which calculates the ratio of gain to noise figure as proposed in . The calculated FOM with respect to signal powers for both amplifiers is shown in Fig. 4.
For low signal powers, FOM of double-pass R-EDFA is higher than that of single-pass R-EDFA. For double-pass R-EDFA, FOM value gradually decreases as the signal power increases. An intersection point between the curve of single-pass EDFA and double-pass EDFA is found around -23 dBm signal power. The signal power at this intersection point is defined as the critical input power as previously reported . At this point, the value of FOM is estimated around 11.2 dB. In order to validate the usefulness of this FOM in determining the classification of R-EDFA functionality, an experiment of bit error rate (BER) measurement is performed.
3. BER measurement
The experimental setup for BER measurement is shown in Fig. 5. In the experiment, the transmitter is modulated with 2.5 Gbps data using a pseudo-random bit sequence of 223–1 of non-return zero signal. The transmitted signal power is around 0 dBm at 1550.3 nm wavelength and a variable optical attenuator (VOA1) is adjusted to a desired signal power level. In order to evaluate FOM analysis, the signal power into R-EDFA is set at -35, -23 and -15 dBm to represent signal power region of small signal, critical power and large signal respectively. For single-pass R-EDFA, the amplifier is placed in between transmitter and receiver directly. Since the double-pass R-EDFA is constructed with CFBG, the negative value of dispersion (-1327 ps/nm) must be compensated in order to have a dispersion-free signal. Thus, 75 km long of SMF-28 fiber is used before the input of amplifier to fully compensate the fiber dispersion effect, therefore, the BER measurement is only affected by the amplifier characteristics. At the receiver, an optical bandpass filter (OBF) is utilized to filter out the broadband ASE generated from both R-EDFAs. In this experiment, the received signal power is varied by VOA2 and finally, the optical signal is captured by an avalanche photodiode (APD). The converted data is sent to the BER Tester to measure the BER performance accordingly. The back-to-back measurement is used as a reference set for performance evaluation purposes.
Figure 6 shows BER curve in variation with received signal power for single-pass and double-pass R-EDFAs. For -35 dBm signal power, the double-pass R-EDFA performs better than the single-pass R-EDFA. In this case, the power penalty around 1.8 dB is obtained for the double-pass R-EDFA as depicted in Fig. 6(a) at BER of 10-10. However, the BER curve for both R-EDFAs has similar characteristics when the critical signal power of -23 dBm is used in the experiment as illustrated in Fig. 6(b). In this case, there is no significant difference between these two R-EDFAs. Nevertheless, single-pass R-EDFA performs better than its counterpart for -15 dBm signal power as shown in Fig. 6(c). Based on the findings, the evaluation of FOM can be utilized to determine the operating signal power range. In this research work, the functionality of R-EDFA can be classified into two groups; pre-amplifier (signal power < -23 dBm) and post-amplifier (signal power > -23 dBm) which is reflected to double-pass R-EDFA and single-pass R-EDFA respectively.
In this paper, a dual-function Erbium-doped fiber amplifier with CFBG has been successfully demonstrated to compensate fiber dispersion effects as well as to amplify the attenuated signal. The proposed double-pass EDFA is optimized and designed to operate under low pump power (10 mW) for remotely-pumped applications in repeaterless transmission systems. A practical comparative analysis has been conducted between the proposed amplifier configuration and the conventional single-pass amplifier configuration to evaluate the optimal setting with respect to the incoming signal power. This is important in order to determine the suitability of R-EDFA configuration which respects to signal power along with the transmission fiber. Based on the findings, the proposed double-pass R-EDFA gives better performance for small input signal powers (less than -23 dBm), while the conventional single-pass R-EDFA gives better performances at higher input signal powers (above -23 dBm). This justification is also verified using BER measurement method. Therefore, the proposed double-pass EDFA with CFBG configuration gives better performance for the application of pre-remotely-pumped EDFA. In addition, the advantage of the proposed amplifier configuration is the use of CFBG to solve the fiber dispersion problem in order to extend the transmission distance. The contribution of this research work is expected to spark attentions from other researchers to integrate dispersion compensating modules with remotely-pumped optical amplifiers.
References and links
1 . J. P. Koplow , S. W. Moore , and D. A. V. Kliner , “ A new method for side pumping of double-clad fiber sources ,” IEEE J. Quantum Electron. 39 , 529 – 540 ( 2003 ). [CrossRef]
2 . H. Maeda , G. Funatsu , and A. Naka , “ Ultra-long-span 500 km 16 × 10 Gbit/s WDM unrepeatered transmission using RZ-DPSK format ,” Electron. Lett. 41 , 34 – 35 ( 2005 ). [CrossRef]
3 . K. Hogari , K. Toge , N. Yoshizawa , and I. Sankawa , “ Low-loss submarine optical fibre cable for repeaterless submarine transmission system employing remotely pumped EDF and distributed Raman amplification ,” Electron. Lett. 39 , 1141 – 1143 ( 2003 ). [CrossRef]
4 . H. Masuda , H. Kawakami , S. Kuwahara , and Y. Miyamoto “ 1.28 Tbit/s (32 × 43 Gbit/s) field trial over 528 km (6 × 88 km) DSF using L-band remotely-pumped EDF/distributed Raman hybrid inline amplifiers ,” Electron. Lett. 39 , 1668 – 1669 ( 2003 ). [CrossRef]
5 . H. Nakano and S. Sasaki , “ Dispersion-compensator incorporated optical fiber amplifier ,” IEEE Photon. Technol. Lett. 7 , 626 – 628 ( 1995 ). [CrossRef]
6 . S. A. E. Lewis , S. V. Chernikov , and J. R. Taylor , “ Gain and saturation characteristics of dual-wavelength-pumped silica-fibre Raman amplifiers ,” Electron. Lett. 35 , 1178 – 1179 ( 1999 ). [CrossRef]
8 . S. L. Tzeng , H. C. Chang , and Y. K. Chen , “ Chirped-fibre-grating-based optical limiting amplifier for simultaneous dispersion compensation and limiting amplification in 10 Gbit/s G.652 fibre link ,” Electron. Lett. 35 , 658 – 660 ( 1999 ). [CrossRef]
9 . S. Namiki , S. Koji , N. Tsukiji , and S. Shikii , “ Challenges of Raman amplification ,” IEEE Proc. 94 , 1024 – 1035 ( 2006 ). [CrossRef]
10 . E. Desurvire , Erbium-doped fiber amplifiers: Principles and applications ( John Wiley & Sons Inc., New York , 1994 ).
11 . A. W. Naji , M. S. Z. Abidin , A. M. Kassir , M. H. Al-Mansoori , M. K. Abdullah , and M. A. Mahdi , “ Trade-off between single and double pass amplification schemes of 1480 nm-pumped EDFA ,” Microwave Opt. Technol. Lett. 43 , 38 – 40 ( 2004 ). [CrossRef]