Pluggable DWDM transceivers lower metro costs, increase density
Metropolitan networks are the fastest growing segment for DWDM technology. In its annual WAN optical components report issued in January 2006, Ovum-RHK forecasts 14% category growth for metro DWDM optical components from 2004 to 2010. This growth is driven by the competition between multiple-system operators (MSOs) and regional Bell operating companies (RBOCs) to provide broadband services to the masses. Such carriers could not easily adapt their infrastructures to the dynamic demands of end users. Therefore, simplifying and expanding these networks using reconfigurable optical add/drop multiplexers (ROADMs) has been a growing trend. This ROADM deployment has increased demand for cost-effective WDM components. Ovum-RHK predicts the multireach DWDM market will overtake the classic long-haul DWDM market this year.
Metro DWDM systems emphasize cost, density, and flexibility. These parameters are strongly dependent upon optical transceiver technology. System manufacturers are focusing upon pluggable and tunable optical transceivers to enable these features in metropolitan DWDM networks.
Pluggable DWDM optical transceivers offer several significant advantages for system manufacturers and end users:
• Density: A system can potentially accommodate up to 48 DWDM 2.5-Gbit/sec small-form-factor pluggable (SFP) modules, up to eight DWDM XENPAKs at 10 Gbits/sec, or 16 DWDM XFPs at 10 Gbits/sec in typical 1-U chassis.
• Upgradeability: Network operators can increase the number of populated transceiver slots as network demand increases. They can add additional wavelengths by plugging additional optical transceivers into their systems.
• Flexibility: The pluggable transceiver allows the network operator to manage and replace wavelength channels for each transceiver slot. This simplifies the upgrade, maintenance, and repair of the DWDM network. XFP, SFP, and XENPAK are front-panel hot pluggable, which allows an upgrade or change to the DWDM network without interruption of service.
• Cost: Both the bill of materials and engineering time are reduced by buying standardized pluggable transceivers versus internal discrete designs.
In addition to the classic OC-48 or OC-192 applications, other protocols such as Gigabit Ethernet (GbE), 1/2/4-Gbit/sec Fibre Channel, and 10-GbE need to be supported by WDM networks. The resulting increased bandwidth demand results in higher-density DWDM network systems requiring small, flexible, low-power, and low-cost DWDM optical transceiver modules.
Line-card density directly translates to module size, which is usually limited by power consumption. A DWDM signal needs to be transmitted at a constant wavelength. In a transmission system with 100-GHz (0.8-nm) wavelength spacing, the laser wavelength has to stay within ±100 pm around its nominal wavelength.
Unfortunately, over the greater than 10-year lifetime of DWDM transmitter operation, the laser diode has a tendency to exhibit wavelength drift. If the drift is uncontrolled, there is a risk the wavelength could drift into the neighboring ITU channel and disrupt the network. The solution has been to include a wavelength locker to eliminate the risk of wavelength drift.Historically, DWDM optics have been relegated to higher-cost and larger-dimension packages to include this wavelength locker. A wavelength locker provides an active feedback loop to monitor the laser wavelength of the DWDM transmitter over time. The temperature of the laser diode is adjusted via a thermoelectric cooler (TEC) based upon the active feedback loop of the locker. Typical wavelength lockers split the backreflected optical beam into two photodiodes. One of the beams is passed through a wavelength-dependent etalon-type filter while the other impinges directly upon the other photodiode. This scheme was typically constructed using micro-optics, which introduce space constraints for the DWDM laser package.
Recently, advances in InP growth techniques and materials processing have produced finer control of the laser diode construction. These refinements have enabled manufacturers to limit the drift over time and characterize and predict lifetime drift profiles. This performance has been proven over a series of accelerated aging tests of the laser diodes in which the chips were run at high set temperatures and high drive currents.
This excellent characterization has enabled construction of fixed feedback loops that mitigate the wavelength drift over the DWDM transmitter’s lifetime. Thus, the large optics-based active feedback loop of the wavelength locker can be replaced by a fixed electrical feedback circuit. This advance enables the transition to smaller, lower-cost laser packages (see Figure 1).
In addition to better laser chip aging characteristics, new laser structures also enable higher laser set temperatures and therefore lower TEC power consumption. Since a 1°C shift in temperature equals an approximately 0.1-nm shift in wavelength, the transmitter optical subassembly (TOSA) and transceiver aging characterization represent additional challenges (see Figure 2). Thermistor aging as well as increased laser bias current over lifetime must be considered.
The following factors change the wavelength of a DWDM transceiver during its lifetime:
• Laser diode bond to carrier — wavelength shift due to stress relaxation after die bonding/soldering.
• Thermistor accuracy and aging.
• Laser diode bias current drift due to aging.
• TEC control loop aging.
• Shift from uncontrolled startup condition to steady-state operation (no interference with neighboring channels allowed).
Other features that transceiver companies would like to integrate include pilot tone to suppress stimulated Brillouin scattering, wavelength monitoring, and even a variable optical attenuator to adjust optical power levels among various channels. Future advances in laser chip technology could enable narrowband or even fully C-band tunable DWDM SFP transceivers. Tunability over a few ITU channels could be made possible by using high-temperature (“uncooled”) electroabsorption distributed-feedback (EA-DFB) lasers, which can operate over a large temperature range and can therefore be tuned to different ITU channels.
For 10-Gbit/sec data rates, DWDM pluggable modules in the XENPAK form factor have been available for some time. As an example, one such module uses a TOSA with an InGaAsP-based EA-DFB laser structure and electrical control circuitry. The DWDM XENPAK is used in metro Ethernet DWDM systems, providing a XAUI electrical interface.
Now, some transceiver companies are in the final stages of DWDM XFP development. The XFP 10-Gbit/sec form factor is protocol agnostic and can be used for telecom as well as for Ethernet systems. Moving from the larger XENPAK to XFP, the DWDM TOSA size has to be reduced further. On the receiver side, a limiting amplifier is used rather than a larger, higher-power-consumption automatic gain control circuit. The XFP enables a density of 16 links per line card rather than eight using XENPAKs. However, systems with high port density face significant thermal challenges.
For lower-data-rate DWDM applications, such as GbE, OC-48, and 1/2/4-Gbit/sec Fibre Channel, DWDM SFP modules are rapidly replacing more expensive, less flexible butterfly-laser package designs. Modules with a maximum power consumption less than 1 W allow system designs with up to 48 ports per line card. DWDM SFP modules are available with an adjustable receiver decision threshold, a feature used in DWDM systems to optimize the receiver performance during operation, given the actual network noise and dispersion conditions.
Innovation in optical transceivers will support the cost, density, and flexibility demands of DWDM networks. Continued investment in laser technology will enable further power consumption and size reduction of the DWDM transmitter. With the shrinking of the laser packaging, pluggable transceivers will incorporate additional DWDM services and functionality.Josef Berger is senior product marketing manager at Opnext Inc. (www.opnext.com).