Abstract: The joint amplifying system based Erbium Doped Fiber Amplifier (EDFA) and Praseodymium Doped Fiber Amplifier (PDFA) has been regarded as one of the promising technologies in optical core networks. However, due to the complex structure and high cost, it is unpractical to optimize the performance for these two amplifiers by experiments. In this study, to overcome this problem and enhance the output performance, the operation processes of EDFA and PDFA are modeled by adopting the Optisystem software. Then, according to the spectrum characteristics of the amplifying systems obtained, the analysis in detail is conducted respectively. Consequently, based on the quantitative results in terms of gain, flatness and noises, the optimal parameters with the respect of pumping power and the length of EDF/PDF are determinate and the output performance of these two amplifying system is also improved obviously.
INTRODUCTION
To date, the Erbium Doped Fiber Amplifier (EDFA) has been widely used in some optical systems, for instance, communication, sensing and measurement, etc., which are with the advantages of high gain and low energy consumption (Chang et al., 2006; Lee et al., 2011; Liang and Chen, 2012; Gao, 2012). However, with the tremendous rising in transmission services from users, the traditional EDFA devices maybe not meet such demand for more data due to the less than 40 nm bandwidth. So, the 1.3 μm waveband with the nature of zero dispersion is considered to be applied into the present and future optical communication networks (Giles and Desurvire, 1991). Therefore, the joint amplifying system combining the Praseodymium Doped Fiber Amplifier (PDFA) and EDFA devices has become one of the hot topics by many researchers recently. But, it should be noticed that, the design optimization is still the key problems in the applications of such two fiber amplifiers, because the output performance is dominated by some parameters, e.g., the length and density of doped fiber, the power and wavelength of pumping source, etc. (Quimby et al., 2008). Taking the complexity of the operating process and structure into account, it is very difficult to analyze the output characteristics of EDFA and PDFA with the various parameters by a real experimental system.
In order to overcome this drawback, the methods based numerical analysis are concerned which are regarded as many advantages with the respect of low cost, fast speed, high flexibility and so on. Then, it is alternative that some works based on the simulation way have been reported in the performance analysis of some active optical devices. For instance (Chang et al., 2008; Wysocki et al., 1994) modeled the output features of the superfluorescent fiber source based on the Dynamic Rate Equation (DRE). Deshayes et al. (2008) simulated the features of 1550 nm fiber laser. And the literatures (Giles and Desurvire, 1991; Qin et al., 2005; Tan et al., 2004) had also achieved to study the output performance of EDFA based the classic Giles model. While, note that, most of these works are hard to be understood and operating because of the high communicational complexity of DRE. In contrast, Ref (Nagy and Marincic, 1991) used the SPICE2 to model the performance of pulse light source in a data transmission system. Also, in the previous work, the characteristics of the laser diode (650 nm) (Buyin et al., 2011), optical fiber laser (1550 nm) (Chang et al., 2011) and Fiber Raman Amplifier (FRA) had been analyzed by using the Optisystem (Liu et al., 2013). In this study, according to the method adopted in the manuscript (Liu et al., 2013), a further simulation is developed on the output performance analysis of EDFA and PDFA devices, respectively. Owing to the limitation of pages, the work focuses on modeling the characteristics of the optical amplifying systems in terms of gain, noise and flatness with the variants pumping power and fiber length.
SIMULATION AND ANALYSIS ON THE EDFA SYSTEM
In Fig. 1, the typical EDFA system with forward pumping is demonstrated, consisting of five parts at least, i.e. the Continuous Wavelength (CW) laser array, the pumping source, WDM device, Erbium Doped Fiber (EDF) and Optical Spectrum Analyzers (OSA). In particular, the CW laser array is used to model the three optical signals from 1540-1560 nm, with the same interval (10 nm) and power (-20 dBm). And a semiconductor laser with the center wavelength of 980 nm is chosen as the pumping source, whose output power is 100 mW. Here, assuming the WDM device is ideal and without any extra loss and two OSA devices are applied into the model to monitoring the output of optical signals and Amplified Spontaneous Emission (ASE) noise. Furthermore, for the EDF, its designed parameters are the length is 5 m, the NA value is 0.24 and erbium ion density is 1x1025 m-3, respectively. According to Ref (Giles and Desurvire, 1991; Chang et al., 2008), not only the pumping power, but the length and density of EDF have the obvious effects to the output features of the EDFA system. But, considering the constant erbium ion density in Optisystem, the work is then concentrated on modeling the EDFA system with various lengths of EDF and pumping power.
Various length of EDF: First, in the model as shown in Figure 1, the density of erbium ion at 1x1025 m-3 is kept and the pumping power (denoted by Ppump) is designed as 100 mW. Then, the effects of output performance of EDFA by the various length of EDF are studied. The initial value of the length of EDF (denoted by LEr) is designed as 5 m and the output spectrum features of EDFA are first shown in Fig. 2a by simulation. In Fig. 2(a), the power of the optical signals reaches about 10 dBm. In other words, there is about 30 dB gain obtained. Also, an ideal flatness among the three signals is found which the difference between the maximum and minimum output power is less than 1 dBm.
| Fig. 1: | Simulation model of the EDFA system |
On the other hand, the average amplitude of ASE noise is near to -45 dBm within the range from 1540-1560 nm. Comparatively, when the length of EDF is equal to 10 m, there is a different picture for the signals and ASE noise. As shown in Fig. 2b, the signal at 1560 nm obtains the maximum gain, whose output power reaches over 12 dBm. But, for the signal at 1540 nm, its output power after amplifying is only about 4 dBm. So, the gain difference is 8dB, much larger (4 times) than the case of Ler = 5 m.
Correspondingly, the spectrum of the ASE noise is hardly fluctuated at the band between 1540 and 1560 nm, though the average amplitude of the ASE noise is still in the level of -45 dBm. The similar numerical results are shown in Fig. 2c, when the length of EDF is 15 m. Also, from Fig. 2c, the gain difference is further enlarged and the value reaches 16 dB. Especially for the signal at 1540 nm, there is a negative gain (about -4 dBm) occurred. That means the amplifying function of the EDFA system has been lost at the 1540 nm waveband. And according to the spectrum of the ASE noise, the center of gain has also changed into 1560 nm in this case. Therefore, by comparing the numerical results obtained in Fig. 2, the optimal output performance in terms of gain and flatness is achieved when Ler = 5 m.
Various pumping power: The output characteristics of EDFA are then analyzed with the various pumping power. Here, the density of erbium ion is still at 1x1025 m-3 and the parameter Ler = 5 m is selected. With the initial pumping power Ppump = 100 mW, the simulation results are shown in Fig. 3. Particularly, the black lines in the left of all three pictures are the output of pumping source with the wavelength at 980 nm. In Fig. 3a, the signal power (red parts) is about 10 dBm and the average ASE noise is -45 dBm when Ppump = 100 mW. Comparatively, for the case of Ppump = 150 mW, the signal power is raised about 2 dBm, but the ASE noise is still -45 dBm. So, the value of SNR of the EDFA system is also increased 2 dB. But, with the continuous raising for Ppump, e.g., Ppump = 200 mW (Fig. 3c), the SNR of EDFA system is not further improved due to the larger ASE noise. In other word, the optimal performance is obtained when pumping power is 150 mW.
| Fig. 2(a-c): | Gain and noise spectrum of the EDFA system with various length of EDF, (a) Ler = 5 m, (b) Ler = 10 m and (c) Ler = 15 m |
| Fig. 3(a-c): | Output spectrum of the EDFA system with various pumping power, (a) Ppump = 100 mW, (b) Ppump = 150 mW and (c) Ppump = 200 mW |
SIMULATION AND ANALYSIS ON THE PDFA SYSTEM
The simulation model of the PDFA system is shown in Fig. 4. Similar to the EDFA system, it is also involved in five parts: CW laser array, pump laser, WDM device (i.e., ideal Multiplexer), Praseodymium Doped Fiber (PDF) and two OSA devices. But, it should be notice that, the designed parameters in the PDFA system are totally different. For instance, the wavelength of pumping source is 1017 nm. And the wavelengths of the three signals generated by the CW laser array are 1300, 1310 and 1320 nm with the same input power at zero dBm, respectively. Further, though the initial length of PDF is 5 m, the density of praseodymium ion is lower one level (1x1024 m-3) than that of the EDFA system. Again, the value of NA is raised to 0.364.
Various length of PDF: Similar to the EDFA system, the effects on the output features of the PDFA system is first studied with various length of PDF. The initial pumping power is 100 mW and the density of praseodymium ion is constant and its value is 1x1024 m-3. Then, the output spectrums are demonstrated in Fig. 5a when the length of PDF (denoted by LPr) is equal to five meters. From Fig. 5a, the bandwidth with high flatness is about 20 nm (from 1300-1320 nm). So, the flatness of three signals is better than that of EDFA obviously. But, comparatively, the gain of the PDFA system is low which is approximate 8 dBm when Lpr = 5 m. On the other hand, the amplitude of the ASE noise in 1300-1320 nm is -65 dBm. Hence, the value of SNR reaches 73 dB which is much better than the case in the EDFA system. Further, through increasing the length of PDF (Ler = 10 m and Ler = 15 m) and the output features are shown in Fig. 5b and c. Comparing the numerical results in Fig. 5a and b, it is observed that the gains of signals are not clear and there is only less than 2dBm increase. Correspondingly, the amplitude of the ASE noise is about -64 dBm. However, in Fig. 5c, a different picture is demonstrated in gain and ASE noise. In the facet of gain, compare to the case of Lpr = 10 m, the signals with the wavelength of 1300 and 1310 nm is almost not affected with the increase of LPr.
But, for the signal at 1320 nm, the value of gain is reduced to 8 dBm. So, the performance of flatness is deteriorated. And the ASE noise spectrum also presents there is a quick attenuation for the energy of pumping source in the right band of 1320 nm, though the amplitude of the ASE noise is still -64 dBm. Consequently, it is clear that the output performance for the case Lpr = 5 m and Lpr = 10 m is predominant to the case of Lpr = 15 m. Moreover, taking the similar output performance and the cost of PDF into account, Lpr = 5 m is the best optimal parameter for the PDFA system.
Various pumping power: Then, the output spectrum of the PDFA system is analyzed with the various pumping power. Similar to the analysis in the section on EDFA, the density of praseodymium ion is designed at 1x1025 m-3, the length of PDF Lpr = 5 m and the initial pumping power Ppump = 100 mW. Then, the numerical results by simulation are shown in Fig. 6. In Fig. 6a, the output power of the three signals is the same and about 8dBm and the average ASE noise is -64dBm for the case of Ppump = 100 mW. In Fig. 6b, when the pumping power is 200 mW, the output power of the signals is raised obviously, reaches 14 dBm.
| Fig. 4: | Simulation model of the PDFA system |
| Fig. 5(a-c): | Gain and noise spectrum of the PDFA system with various length of PDF, (a) Lpr = 5 m(b), Lpr = 10 m, (c) Lpr = 15 m |
| Fig. 6(a-c): | Output spectrum of the PDFA system with various pumping power, (a) Ppump = 100 mW, (b) Ppump = 200 mW and (c) Ppump = 300 mW |
And the corresponding gain is 14 dB and the average amplitude of the ASE noise is about -61 dBm. So, it proves the higher pumping power can lead the improvement of SNR of the PDFA system (e.g., the value of SNR is 72 dB for Ppump = 100 mW and SNR = 75 dB for Ppump = 200 mW). Therefore, it is estimated that a higher SNR would be obtained when Ppump = 300 mW. Unfortunately, according to the results in Fig. 6c, the gain of the signals is not increased, still about 14 dBm and in contrast the average amplitude of the ASE noise is a little increased, maybe 58 dBm. That means that the extra 100mW pumping power stimulates 3 dBm ASE noise, not signals. In the other words, the 200 mW pumping power is matched with the five meters PDF.
CONCLUSION
In this study, based on the Optisystem, the operations of the EDFA and PDFA systems are modeled, respectively. And it is convenient to obtain the output features of the amplifying system in terms of gain, flatness and ASE noise. According the numerical results of output spectrum, the optimal parameters are determined with the respects of the length of EDF/PDF and pumping power. Further, the results also show that a higher gain (about 30 dBm) but with a low SNR is obtained in the EDFA system. Comparatively, the better performance in flatness and SNR is presented in the PDFA system, but with a low gain, about half value of the EDFA system. It is significant of these results for the future application of the joint EDFA and PDFA system.
ACKNOWLEDGMENTS
This study is supported by Heilongjiang Province Natural Science Foundation (No. 201009), China Post Doctor Foundations (No. 2012M520779), College of Heilongjiang Province, Key lab of Electronic Engineering Outstanding Young Foundation (DZZD201003) and Heilongjiang Province Educational Department Science and Technology Project (No. 11511382).