1. Introduction

Multi-wavelength fiber laser sources have attracted several research interests due to their potential applications in wavelength-division-multiplexing, optical sensor networks, and testing systems [1,2]. Numerous mechanisms have been employed to construct this laser scheme [3–6]. Amongst them, a multi-wavelength Brillouin-Raman fiber laser (MBRFL) has been regarded as a potential and feasible solution with several important advantages. These are such as stable operation at room temperature, flat and large gain bandwidth, compatibility with fiber properties, and design simplicity. Different BRFL structures have been proposed to generate multiple wavelengths with 10 or 20 GHz spacing through linear or ring cavity resonators [7–10]. However, for MBRFL with 20 GHz spacing the need for a wider flat amplitude bandwidth becomes more crucial with the advent of higher S-OSNR. In our earlier work [9], a MBRFL with 20 GHz spacing was reported through Raman forward pumping (FWP) scheme which yields 195 lasing lines with 26 dB OSNR. The larger channel spacing combined with limited bandwidth has obviously placed a constraint on the number of Stokes lines achievable in the structure. One of the major challenges in MBRFL is to achieve a uniform Stokes lines with wider spacing over a broader bandwidth with excellent OSNR. Although utilizing multiple Raman pump sources dubbed as a bidirectional or higher order pumping is a feasible method of achieving this, it introduces higher cost and complexity [11–14]. In order to resolve this conundrum, here for the first time we propose a new cavity design that investigates the effect of Raman pump power distributions on MBRFL spectra utilizing only single pump source. The performances of MBRFL are evaluated by incorporating various pump coupling distributions along with the fiber longitudinal structure. In order to achieve multiple Brillouin Stokes lines (MBSL) with 20 GHz spacing and excellent qualities especially in terms of OSNR and higher peak power discrepancy between odd- and even-order lasing lines in comparison with the results attained in our previous study [9], the optimization of coupling ratio, Brillouin pump (BP) power, BP wavelength, and Raman pump power (RPP) has been carried out thoroughly.

2. Experimental setup and principle of operation

The configuration of a linear cavity MBRFL is depicted in Fig. 1. The BP is provided by an external-cavity tunable laser source (TLS) with maximum power of 5 dBm and a 200 kHz linewidth. Besides this, a Raman pump unit (RPU) that has a maximum power of 1000 mW and operates at 1455 nm wavelength regime is also employed. In addition, a 7.2 km long dispersion compensating fiber (DCF) serves as a Brillouin-Raman gain medium which has a nonlinear coefficient, $\gamma$ of 7.3 (Wkm)⁻¹ and an insertion loss of 5 dB. The interaction length of this DCF fiber also induces a tight confinement attribute due to its small effective area, $A_{eff}$ of 20 µm². This ascertains the formation of distributed laser reflectors through strong Rayleigh backscattering (RLBS) responses which is further strengthened by high germanium (GeO₂) ion doping in the fiber (GeO₂ molecules exhibit a larger Raman gain peaking near 13.1 THz) [15]. The main purpose of this assessment is to investigate the effect of Raman pump power distributions along with two fiber end-facets with various percentages. Thus, the proper coupling ratio is founded so that the MBSL with excellent features is achieved. In the laser structure, a bidirectional pumping (BiDP) design is realized by splicing the RPU to the DCF through two 1480/1550 nm wavelength selective couplers (WSC1 & WSC2). A set of couplers available in our laboratory with numerous ratios such as 5/95, 10/90, 20/80, 30/70, 50/50, 70/30, 80/20, 90/10, and 95/5 are connected individually between these components as illustrated in Fig. 1. In this case, the numerator of this coupling ratio, $CR$ denotes the growth of RPP propagation towards the FWP direction at random separation. Consequently, the denominator implies the corresponding decline of power distribution in the backward pumping (BWP) direction. The concept “forward” and “backward” indicate similar and opposite the Raman pump power distributions with respect to the BP photons as illustrated in Fig. 1. Alternatively, for the configurations that employ 100% FWP and 100% BWP, the couplers are excluded from the system and the RPU is connected directly to either WSC1 or WSC2 accordingly. In addition, In the setup two optical isolators that imply an isolation value around 40 dB in the range of 1530-1560 nm are used to reduce any back-reflection light that can interrupt the lasing stability [15]. For evaluations of optical output, an optical spectrum analyzer (OSA) with a 0.02 nm resolution is employed.

 

Fig. 1 Schematic diagram of a MBRFL where the dashed boxes indicate the pumping arrangements. In the BiDP schemes, the couplers utilized in forward/backward directions are 5/95, 10/90, 20/80, 30/70, 50/50, 70/30, 80/20, 90/10, and 95/5. In addition, no couplers are included for 100% FWP and BWP as shown in the insets.

Download Full Size | PDF

3. Results and discussion

The principle of multi-wavelength combs generation that is amplified by Brillouin-Raman gain is slightly different in various pump coupling distributions. When providing a total RPP, $P_0$ the Raman gain development along with the fiber longitudinal dimension $z$ can be described as [15],

At the outset of this assessment, in order to study the effect of applying different types of Raman pumping distributions, evaluations on amplified spontaneous emission (ASE) are carried out. These are done when the $P_0$ is set at 1000 mW with the absence of BP signal. All the results obtained are measured at the OSA location as shown in Fig. 1. When employing 100% BWP, $r_f$is taken as 0 and the total pump power $P_0=P_b$ in which the Eq. (1) above can be simplified as,

 

Fig. 2 ASE spectra at different coupling ratios when the RPP is set at 1000 mW and the BP signal is switched off. The $CR$ labels are arranged from top to bottom level in descending-order of output power values.

Download Full Size | PDF

In other hand, when 100% FWP is used, the Raman gain is the highest at the input end of the fiber ($z=0$) before reaching a minimum value at $z=L$ due to its compliance to the exponential decay elucidated in Eq. (3). This justifies the weaker effect of ASE power when comparing to that produced in 100% BWP. However, this value is not too much different to that at other $CRs$ of 70/30, 10/90, 30/70, and 95/5 where the lowest noise level is observed with the incorporation of 50/50 coupler.

Since the ASE noise depends directly on the strength of oscillating Raman Stokes wave in the cavity, several laser characteristics of the MBRFL are investigated when altering the pump coupling distributions. These characteristics include average S-OSNR, wavelength operation, wavelength spacing, as well as Stokes lines count (SLC). Therefore, from Fig. 2, it can be concluded that lower ASE spectra results in reduction of gain competition among self-lasing modes itself and with Brillouin lasing lines which can improve S-OSNR considerably. From this figure, employment of 50/50 coupler represents the lowest ASE noise among all pumping schemes. This also can be understood from Eq. (1) where there is a balance for $RG(z)$ at the input and output end of the fiber which leads to reduction in gain competition and can result in an improvement on wavelength operation especially S-OSNR. However in some cases, a higher Raman gain property may lead to suppression of the oscillating modes that are associated to the RPG which effects on SLC and S-OSNR as will be discussed later. As a result, for all $CRs$the interplay between Raman gain and self-lasing modes has effect on the lasing characteristics.

In order to observe the effect of different pumping propagations on consecutive Brillouin–Rayleigh Stokes lines, the output spectra of MBRFLs is demonstrated in Fig. 3(a) and their magnified views are shown in Fig. 3(b). For all experiments in this assessment, the BP power and BP wavelength are set at 5 dBm and 1555 nm, respectively while the RPP is fixed at 1000 mW. As shown clearly in Fig. 3(b), different pumping distributions lead to different MBRFL spectra since pumping ratios play an effective function on lasing lines features. From Fig. 3(b), for the 100% BWP configuration, as the BP signal with the wavelength of ${\lambda }_{BP}$ passes throughout the gain medium (DCF) and experiences gain, is backscattered through the non-shifted frequency of Rayleigh and up-shifted frequency of Brillouin scatterings. RLBS of Brillouin pump signal, $\lambda ={\lambda }_{BP}$ and first Brillouin Stokes line (1st BSL), ${\lambda }_{BS1}={\lambda }_{BP}+0.08nm$ are amplified through co-pumping Raman amplification. However, the 1st BSL experiences higher gain than its Rayleigh line and is saturated quicker owing to lower nonlinear threshold power. When the threshold condition is satisfied for both Rayleigh and Brillouin scatterings, the 1st BSL is capable of generating second Brillouin Stokes line (2nd BSL), ${\lambda }_{BS2}={\lambda }_{BP}+2(0.08nm)$ in forward direction. Simultaneously, the 1st BSL is backscattered through non-shifted frequency of Rayleigh scattering. At this point, the Rayleigh scattering acts as a virtual mirror. Both first and second BSL in a similar direction to the transmitted BP signal experience higher amplification through BWP scheme compared with other pumping distributions as illustrated in Fig. 2. In consequence, the successive peak to peak distance between odd- and even-order lasing lines leads to 10 GHz spacing as elucidated in Fig. 3(c) by brown colour. On the other hand, Rayleigh scattering of 1st BSL and subsequent odd-orders Rayleigh components in FWP and BiDP schemes are just slightly amplified due to the lower Raman gain which results in MBSL with 20 GHz spacing as indicated in Fig. 3(c) by green colour.

 

Fig. 3 (a) Spectra output of MBRFLs, (b) the magnified views of spectral features at different RPP couplings namely, 100% BWP, 20/80, 50/50, 90/10, and 100% FWP and (c) illustration of 10 and 20 GHz spacing MBRFL spectra for BWP and 90/10 coupler as an example. For simplification, other $CRs$ results are not shown (RPP and BP power are fixed at 1000 mW and 5 dBm, respectively, while the BP wavelength is set at 1555 nm).

Download Full Size | PDF

From Fig. 3, in 100% BWP scheme as a result of the higher noise factor of ASE as shown in Fig. 2, the generated Stokes combs suffer the most severe S-OSNR reduction across the MBRFL bandwidth compared with other pumping distributions. However, this scheme offers wider bandwidth owing to the highest Raman gain as illustrated in Fig. 3(a). In contrast, the second and consequent even-orders Brillouin components in FWP and BiDP schemes experience lower noise and have extremely higher S-OSNR values. Since the Rayleigh components of odd-orders do not experience a strong Raman gain as much as in BWP scheme so that the energies are utilized only for the formation of the BSL. Thus, the lasing lines attain enough energy to compete with the free running cavity modes which implies with the noise floor reduction. As a consequence, this results in the increase ratio between peak power and the noise floor which elucidates their higher property of S-OSNRs when comparing to that for 100% BWP as shown in Fig. 3(c). From Fig. 3(b) it is observable that when implementing a 20/80 coupler that slightly reduces the RPP in the backward direction, the Rayleigh components of odd-order Brillouin Stokes are stronger as compared to those at other couplers (50/50, 90/10, and 100% FWP). The results are reasonable when relating to the dependency of Raman gain to the pumping distributions as manifested in Eq. (1). In addition, it is also clear that implementing a 50/50 coupler yields the lowest noise floor, which agrees well with the lowest ASE spectrum attained in Fig. 2. Therefore from the results obtained to achieve 20 GHz spacing MBRFL spectra, the investigations of the FWP and BiDP schemes on the lasing characteristics are suggested. However, before any performance analysis on Stokes combs features are carried out, proper configurations for 20 GHz spacing MBRFL need to be also specified. This is done to find better schemes which offer higher peak power discrepancy between odd- and even-order Stokes lines. Thus, the right structures will be those with at least 21 dB discrepancies between odd- and even-order lines to achieve an improvement in BS combs characteristics compared with the results achieved in our previous study [9]. Figure 4 shows the relation between average discrepancy between peak power of odd- and even-order BSL with fraction of RPP toward FWP ($r_f$). This measurement is done by subtracting the average peak power values of odd- and even-orders BS combs in decibel scale. For clarification, the increment of $r_f$ at WSC1 from 5% until 100% is equivalent by the increasing $CR$ from 5/95 to 100% FWP scheme. During measurements, all pumping characteristics are maintained at the previous values. From the graph, it is observed that for the cases 50/50, 90/10, and 100% FWP, MBRFL offer higher difference between odd- and even-order Stokes lines which are more preferable. Thus, an analysis on lasing characteristics of these three schemes is only performed for the entire assessments.

 

Fig. 4 Peak power difference between odd- and even-order BSL as a fraction of RPP toward FWP ($r_f$) (RPP = 1000 mW, BP wavelength = 1555 nm, BP power = 5 dBm).

Download Full Size | PDF

In order to investigate the effect of RPP on average S-OSNR along the entire MBRFL comb for the proposed schemes, plotted graph in Fig. 5 is illustrated. This figure depicts an average S-OSNR as a function of the RPP increment from 800 to 1000 mW. In this case, parameters such as the BP power and wavelength are set at 5 dBm and 1555 nm, correspondingly. By taking into account that the higher RPP leads to higher SLC, analysis on the larger values of RPP are only carried out [9]. It can be inferred from this figure that with the increment of RPP, the S-OSNR is reduced gradually at all pumping configurations due to the spectral broadening phenomenon on each lasing line [16]. From this figure also, $CR=50/50$ is the optimum value that results in the highest S-OSNR at all RPP values. The equal distribution of Raman gain at the input and output end of the fiber justifies this condition. In this case, the gain competition among the self-lasing modes is the lowest, thus the minimum amount of energy is sufficient to repress the free running modes and its remaining is invested to increase Brillouin Stokes peak power (SPP). Therefore, a reduction in the noise floor and increase in SPP leads to better S-OSNR properties. In this case, the average S-OSNRs of MBRFL are 31.5, 30, and 29 dB at 800, 900, and 1000 mW RPP respectively. In addition, the 90/10 coupler offers better S-OSNR compared to the 100% FWP scheme. Although, it (90/10) exhibits higher Raman gain as well as higher free running modes compared to that 100% FWP, its energy is sufficient to suppress self-lasing cavity modes which results in more S-OSNR improvement.

 

Fig. 5 Evolutions of S-OSNR against RPP increment, the $CR$ labels are arranged from top to bottom levels in descending-order of S-OSNR values (BP wavelength = 1555 nm, BP power = 5 dBm).

Download Full Size | PDF

Next, in order to study the effect of RPP on Stokes lines numbers of the proposed pumping distributions, the graph is plotted in Fig. 6(a). Similar to the prior evaluations, the BP power and its corresponding wavelength are fixed at 5 dBm and 1555 nm, respectively. The results obtained demonstrate that the SLC increases with the RPP increment for all $CR$ cases due to the more energy transfer from the pump to the BS signals. Therefore, this leads to the formation of more BSL within the accessible Raman gain bandwidth.

 

Fig. 6 (a) Number of output channels versus RPP increment at different$CRs$, (b) threshold power as a function of $r_f$ (RPP = 1 W, BP power = 5 dBm, BP wavelength = 1555 nm).

Download Full Size | PDF

In general, a threshold level is satisfied when the intracavity gain surpasses the loss in the systems. In addition, the loss is caused by the light transmission along the sample surface. It is noted that to achieve a higher number of channels, the configuration should be able to provide high gain with lower stimulated Brillouin scattering (SBS) threshold properties together. Therefore from this figure, the 90/10 coupler offers the highest SLC among all pumping schemes at all RPP values since it experiences high gain and low SBS threshold as shown in Figs. 2 and 6(b) correspondingly. Moreover, a lower amount of Stokes line numbers in the case of 50/50 coupler corresponds to its lower Raman gain characteristics. Thus, due to the low availability of energy, higher order Stokes lines are not able to overcome the cavity loss and reach the SBS threshold; thus, no more Stokes lines are generated.

To confirm our justification on SBS threshold and further analysis, variation in threshold power values as a fraction of RPP toward FWP ($r_f$) is also investigated, as illustrated in Fig. 6(b). During experiments, all pump power parameters are maintained at the previous values and one circulator together with OSA with 0.02 nm resolution is located after the left isolator to observe the generation of 1st BSL. From this figure, it can be inferred that by implementation of 100% FWP, the lowest threshold operation of 152 mW is achieved. This is related to the highest ratio of Raman gain at the start of the fiber end which results in higher inelastic scattering and photon emission. In fact, when the BP signal interacts with the Raman photons at this fiber location, maximum energy transfer from the Raman pump to the Brillouin signal is occurred. Therefore, a higher energy for BP signal is provided that is responsible for introducing a stronger SBS effect and lower threshold alternatively. While, the threshold power is started to increase slowly until$CR=50/50(50\%)$, where its SBS threshold is maintained at 200 mW. This is related to the lower energy transfer from Raman pump to the BP signal. Although the 100% FWP illustrates the minimum SBS threshold compared to other pumping propagations, its lower gain features leads to the generation of less SLC compared to that 90/10 coupler.

As understood from the results obtained in our previous study [9], at lower BP power, the lasing lines count is increased which is attributed to the optimization of Raman gain saturation at lower BP power. Thus, to accomplish our analysis on number of channels for these three pumping schemes, the effect of different BP wavelengths on SLC is also carried out when the BP power is set at its minimum value of −2.6 dBm, while a total of 1000 mW RPP is distributed into the cavity.

Figure 7(a) demonstrates the evolution of the SLC and S-OSNR against the BP wavelength increment. From the graph, by increasing the BP wavelength, the SLC is reduced, while the S-OSNRs demonstrate a diverse trend in comparison with SLC for all schemes. The higher value of S-OSNRs at longer BP wavelength is attributed to creation of the lower number of Stokes lines so that the pump energy is sufficient to repress the self-lasing cavity modes and increases the SPP. Moreover for all pump coupling distributions, the S-OSNR is increased by increments of BP wavelength from 1540 nm to 1565 nm. However, this trend is changed when the BP wavelength is chosen at 1570 nm for coupling ratios 50/50 and 90/10. In these cases, the amount of the energy at the end of the Raman gain bandwidth is shared along the lasing lines in order to elevate their SPP as illustrated in Fig. 7(b). This resulted in lesser energy remains for suppression of free running modes and S-OSNR enhancement. However, for 100% FWP, the S-OSNR follows a similar trend in which the Brillouin S-OSNR increases at 1570 nm as well. This can be related to its higher gain properties compared to other pumping ratios at 1570 nm BP wavelength as shown in Fig. 2. In addition, by employing $CR=50/50$ a maximum of 212 lasing lines are attained when the BP wavelength is set to 1543 nm. In contrast, the wavelength operation is reduced to 1545 nm for 90/10 coupler where 195 channels are produced. Decrement in wavelength operation for this coupler is related to the higher gain competition among self-lasing modes with itself and with lasing lines. Moreover, it is essential to mention that when employing 100% FWP scheme, the wavelength operation is increased where the injection of BP wavelength lower than 1540 nm cannot drive MBRFL comb. This is occurred since the cavity loss is reduced as the WSC2 and coupler are removed from the setup, and the SBS threshold is also the lowest. Thus, the lasing lines with the lowest cavity loss and minimum SBS threshold are able to suppress free running modes and consequently driving MBSL at shorter BP wavelengths. However, the number of channels at the specific BP wavelength (1545 nm and longer than this value), follows a similar pattern as shown in Fig. 6(a) in which the channels count is reduced in descending order of 90/10, 100% FWP, and 50/50.

 

Fig. 7 (a) Evolutions of S-OSNR and SLC as a function of BP wavelength, (b) variation of Stokes peak power against the BP wavelength increment at different pumping ratios. RPP and BP power are fixed at their optimized values which are known as 1 W and −2.6 dBm respectively.

Download Full Size | PDF

It is noted that, the discrepancy in wavelength operation for different pumping schemes can be attributed to the failure of BSL in gain competition with self-lasing modes due to inadequate Raman gain. However, 90/10 coupler exhibits higher Raman gain in comparison with couplers 50/50. But, the higher Stimulated Raman scattering (SRS) for this coupler leads to stronger self-lasing cavity modes at the RPG as well. Therefore, the lasing lines are not able to suppress the existence of these unwanted modes that limits their wavelength operation. As a result, the wavelength operation for all pumping configurations is largely depends on the capability of SRS to prevail the self-lasing cavity modes.

Figure 7(b) depicts the evolutions of Brillouin SPP with BP wavelength increment for the proposed pumping schemes. These results are achieved when similar pumping characteristics are injected to the cavity. The graph shows a linear evolution of SPP for all pumping schemes when the BP wavelength increases owing to the distribution of energy among the lower number of Stokes lines. Moreover, the Stokes lines generated via 50/50 coupler have the highest peak power when 1555 nm BP wavelength is injected in comparison with other configurations. This occurs due to the minimum gain competition between self-lasing modes compared with other pumping propagations so that more energy is available for BSL to increase their peak power values. As a consequence, from Figs. 7(a) and 7(b), it can be concluded that there is an interplay for energy sharing among lasing lines to enhance SLC, S-OSNR, and SPP. In other words, the generation of higher lasing lines with wider wavelength tunability is achieved at the expense of lower S-OSNR and SPP such as 100% FWP scheme. However, from all assessments it can be inferred that coupling ratio of 50/50 is the optimum value that leads to the MBSL with the highest peak power discrepancy between odd- and even-order lines (22.5 dB), higher number of channels at 1543 nm BP wavelength, high SPP, as well as outstanding S-OSNR. A total of 212 Stokes lines (3-dB flat amplitude) with −10 dBm SPP and 27.5 dB OSNR are generated as shown in Fig. 8. These features for MBRFL spectrum without any turbulent waves can be attributed to the balance in the pump power distribution.

 

Fig. 8 Spectrum of MBRFL together with its enlarged view when the $CR$ is set at 50/50 (RPP = 1000 mW, BP power = −2.6 dBm, BP wavelength = 1543 nm).

Download Full Size | PDF

From Fig. 8, the lasing performances are superior to that obtained in [9] where incorporation 11 km DCF with Bi-EDF through FWP illustrates 195 channels with 26 dB OSNR and average −13.5 dB SPP at 1000 mW and −2.6 dBm RPP and BP power respectively. In addition, this is a significant achievement especially when utilizing a single-pumped source with good performance on the operating wavelengths, lasing lines count, S-OSNR, SPP, as well as flatness.

4. Conclusion

We have presented the exploitation of variable pumping distributions along with two fiber-end facets through single pump source. This is realized by individually changing the coupling ratios in backward and forward directions. The changes in coupling ratio are the main property that influences the gain competition among self-lasing cavity modes and consequentially lasing properties. These are such as number of channels, S-OSNR, and wavelength operation which can be adjusted to satisfy special requirements from high number of channels with longer wavelength operation to an excellent S-OSNR with 20 GHz spacing. From the results obtained, the configurations namely 50/50, 90/10, and 100% FWP offer at least 21 dB peak power discrepancy between odd- and even-order lasing lines among other pumping ratios. Thus, all the evaluations are carried out for aforementioned structures at different RPP, BP power, and BP wavelengths. The incorporation of a 50/50 coupler satisfies the generation of multiple flat Stokes lines with the highest SPP, excellent S-OSNR, and high number of channels with peak power discrepancy of 22.5 dB between odd-and even-order lasing lines. These are occurred owing to its minimum gain competition among self lasing cavity modes. This configuration strongly offers a few advantages especially enhanced functionality and cost-saving solution due to the utilization of only one Raman pump unit. With better understanding of these issues, more advanced design and performance to this novel class of lasers can be developed which will open up various beneficial applications in the future.