High-performance pump lasers drive Raman amplification
Advancements in the sophistication of 1480-nm and 14xx-nm pump lasers will support next-generation ultralong-haul, long-haul, and metro amplifiers based on erbium-doped fiber amplifiers and Raman technologies.
For optical amplifiers to succeed in the marketplace, they must offer a compelling value proposition in two stages. The first stage is to meet essential performance demands, and the second is to conserve customer resources. In the first stage, the most important performance criterion is reach per link—the amount of gain available per amplification stage less the amount of noise added per stage. Depending on reach, the optical amplification market has five categories: ultralong-haul (ULH), long-haul (LH), regional, metro core, and metro access.
In the ULH/LH market, several system vendors are developing distributed wide-band Raman amplifiers, high-gain-flatness erbium-doped fiber amplifiers (EDFAs), and high spectral-efficiency networks. Next-generation amplifiers designed for ULH/LH applications will need to do much more than provide the typical 17 to 23 dB of gain observed between the input and the output of a line amplifier. The amplification solution needs to provide high gain with low accumulated noise, careful dispersion compensation, and low gain ripple. With this in mind, significant gain is required to compensate for mid-stage internal access incorporating passive and dynamic loss elements such as dispersion-compensation fibers, gain-flattening filters, add/drop multiplexers, and the usual losses associated with splicing and diagnostics.
Reliable high-power pump lasers with good power efficiency are needed in ULH/LH applications to optimize amplifier design for parameters besides the link power budget. The amplifier designs developed before 2001 incorporated the 14xx- (1405- to 1520-nm), 1480-, and 980-nm pump lasers with power levels of typically less than 200 mW. The pump lasers used for the next generation of optical networks will typically have more than 200 mW of power, with most of them capable of 300 mW. An intermediate generation of pump lasers, currently being manufactured and marketed, will not see significant deployment.
Pump lasers with high output powers in the 14xx wavelength range are required for Raman amplification of C- and L-band signals. In addition, high-power 1480-nm pump lasers are used to pump the power amplifier sections of multistage, high-channel-count EDFAs. Multiple vendors have announced high-efficiency pump lasers with power levels ranging from 300 to 500 mW in anticipation of these needs. This will allow amplifier designers to have multiple sources for the next generation of high-power 14xx pumps. Pump lasers with power levels of more than 1 W at multiple wavelengths and more than 700 mW in the fiber have been demonstrated (see Fig. 1).1 In addition to the single-laser discrete packages, various vendors are pursuing hybrid integration of pump modules that provide more than 400 mW. For example, two polarization-multiplexed 400-mW pump-laser chips can provide 720 mW from a single pump-laser package at one wavelength with a depolarized output beam. Various vendors are also developing wavelength-multiplexed pump-laser modules.
In addition to reach, there is an essential level of adaptability, flexibility, and control needed for adequate network monitoring and management of the amplifiers. As a result, designers are emphasizing dynamic and static solutions for precise gain flatness and gain-transient suppression. To enable such performance at the amplifier level, the pump lasers have to be predictable, with few tracking errors, steady tracking ratios, and good spectral stability.
High-stability packaging of pump lasers is essential to achieve predictability and control. The litmus test for stability is the tracking error measurement, where the monitor photodiode current is held constant, the ambient (case) temperature is varied between 0°C and 70°C, and the variation in fiber output power is recorded. For example, some pump lasers can exhibit exceptionally low fluctuation of fiber power with changing case temperature, with tracking errors of ±2% compared to typical values of ±10%.
Another important requirement for Raman pump lasers is a high polarization-extinction ratio at the output of the polarization-maintaining (PM) fiber pigtail. Raman amplifiers require a low degree of polarization to accommodate random fluctuations in the polarization of the incoming signal. The current practice in the industry is to use polarization-beam combiners to maintain a low degree of polarization. For efficient polarization multiplexing with beam combiners, the pump power from the PM fiber pigtail must have a high degree of polarization.
Pump lasers with polarization extinction ratios greater than 20 dB (in other words, a 100:1 ratio between the power in the slow axis and the power in the fast axis of the output PM fiber pigtail) are now available. The next generation of Raman amplifiers may see the debut of all-fiber Lyot-depolarizer technology, originally developed a few years ago for applications in fiber-based gyroscopes. With these all-fiber devices, depolarized 300- to 450-mW pump lasers are likely to be available in the 1350- to 1520-nm range for C-, L-, and S-band Raman amplification.
Stage two of the value proposition is to conserve customer resources. Once the essential performance demands are met, the winning optical-networking products will minimize the total resources, natural and manmade, consumed by the immediate customer and end user. Space, energy, cash, and time are the key resources to conserve, with a degree of uncertainty in either performance or resources required.
Reduced power consumption, predictability, and design flexibility are key points of differentiation for advanced amplifier products. This is especially true for amplification applications in which several high-power pumps are integrated into the gain module. High-gain EDFAs and next-generation Raman amplifiers are likely to have more than four pump lasers. The resulting heat load limits the number of pumps used. Heat load (and the concomitant limitation on compactness of the system) is a consideration for system vendors and service providers.
Novel pump-laser designs can dramatically reduce the heat load from amplifiers by enabling lower thermo-electric cooler power consumption (see Fig. 2). With an increase in the operating temperature of the laser chip (as measured by the thermistor temperature), the TEC can operate at significantly lower input power, thus reducing overall power consumption by 25% or more. Chip designs capable of much higher powers allow the increased temperature operation without sacrificing reliability. For example, a laser chip capable of more than 450 mW from the module—in other words, chip powers exceeding 560 mW—at 25°C thermistor temperature can operate at module power outputs of 200 mW to 350 mW.
The reduced heat load per pump laser can significantly reduce the heat density and can help improve the packing density in amplifier cards and in equipment racks meant for use in service providers' central offices, where space and power are at a premium. Lower power consumption by the on-board pumps in a Raman amplifier block also enables the amplifier to handle additional expansion wavelength channels simply by adding more pump lasers. Designers can use the same principles to develop uncooled, medium-power pump lasers that get rid of the TEC and associated control and drive circuitry, thus lowering the complexity and cost of amplifiers for the metro market.
The next generation of pump lasers will also enable simplification of the gain-module board layout by providing single pump lasers that reliably replace two low-power pump lasers. This simplification is likely to lead to cost savings in the manufacturing of amplifiers without compromising performance. For example, a 420-mW depolarized pump laser can replace two 230-mW pump lasers that need to be multiplexed in a separate pump combiner, thus reducing the number of pump lasers, drivers, and control electronics, and enabling compact amplifiers to be designed.
In the amplification market, Raman amplification is clearly an all-optical technology that enables potentially disruptive service-provisioning flexibility in LH/ULH-based networks. Perhaps more significantly, this amplification solution got out of the starting gate before the current industry slowdown, and has been proven in the field. Raman amplification is also expected to become a central ingredient in 40 Gbit/s (OC-768) LH designs, although first field deployments have not yet occurred. Every major system manufacturer is planning to offer Raman amplification in future ULH/LH systems at 10 Gbit/s (OC-192) and beyond with multiple benefits to service providers, including improved quality of service, quicker provisioning, and reduction in operating expenses.
Advancements in pump-laser technology, wavelength mulitplexers, polarization multiplexers, and diagnostics are rapidly enabling Raman amplification. Because pump lasers account for much of the performance and reliability of the Raman amplifiers, and constitute a significant portion of the cost, energy, and space consumed, the next generation of pump lasers is key to the success of Raman amplification.
The author wishes to thank T. Swanson, D. Garbuzov, J. Connolly, P. York, D. Hayes, and A. Winters for their input and reviews.
- D. Garbuzov et al., OFC 2001 Post-deadline paper PD-13 (May 22, 2001).
Swami Srinivasan is product line manager at Princeton Lightwave, 2601 Route 130 South, Cranbury, NJ 08512; e-mail: firstname.lastname@example.org.