PTB >> PTB Insights: Improving Economics at Optical Networks’ 10 Gb/s Sweet Spot

The communications industry is currently facing a critical juncture as increasing bandwidth demands and convergence opportunities create a “sweet spot” at 10 Gb/s. However, significant challenges are still hampering the widespread deployment of 10 Gb/s optical networks. One critical hurdle is the difficulty of creating economically viable modulated-light sources that can reliably deliver 10 Gb/s transmission for distances of 40 km, 80 km, and beyond.

Until now, externally modulated lasers were the only viable solution for transmission in the 1550-nm band, which is used for extended reach and DWDM optical networks at 10 Gb/s. External modulation dissipates more power than direct modulation because the laser is turned on to full power at all times and the external modulator attenuates this to create the modulated signal. Externally modulated lasers using a distributed feedback (DFB) laser with a Lithium Niobate modulator, for example, are large, expensive, and require a high power driver. Monolithic or hybrid integrated electroabsorption modulated lasers (EMLs) are also expensive, especially for distances up to 80 km, and prohibitively expensive for narrow grid DWDM. In addition, EMLs have limited output power, which can increase the need for additional erbiumdoped fiber amplifiers (EDFAs) or more expensive EDFAs.

Projected volumes for 10 Gb/s transceivers are insufficient to support the development required to overcome the low production yields associated with monolithic EMLs. This means that EMLs will be only marginally capable of meeting volume and cost requirements for SONET and DWDM performance at 80 km, a key requirement for telecom.

The use of a new Planar External Cavity (PLANEX) approach overcomes these obstacles through innovative design and manufacturing techniques that economically produce directly modulated External Cavity Lasers (ECLs). The approach combines a Planar Lightwave Circuit (PLC) containing an integrated Bragg grating and a high-performance, low-cost gain chip. Optimizing the lowcost integration of ECLs on silicon wafers using proven high-yield processes and standard CMOS fabrication tools enables a dramatic reduction in the cost of implementing 10 Gb/s links for 40 km, 80 km, and DWDM.

The PLANEX approach offers three key advantages: a low-cost, highly scalable volume-oriented production environment; reduced cost-of-ownership and migration from TDM to DWDM; and higher performance over longer distances.

Production Environment
A key tenet of the production model is to eliminate the low-yield constraints of complex fabrication processes in the early part of the production chain and to leverage high-yield low cost components in the final assembly. The use of pre-tested components that are readily outsourced allows for improved yields and lower costs, along with the ability to scale-up production volumes efficiently.

By using low cost, outsourced gain chips, the PLANEX approach eliminates the need for expensive and underutilized internal InP fabrication facilities, which are a major cost in conventional EML production. The gain chip is integrated with a low-cost PLC containing a Bragg grating, which is fabricated on a silicon wafer using well-established, volume-oriented processes.

The gain chip functions as an optical amplifier and the Bragg grating acts as both a partially reflecting mirror and a wavelength locker. This enables the laser to output a controlled chirp waveform at a very well defined wavelength. The large cavity size of the ECL provides a device that exhibits low reflection sensitivity, and, in some applications, can operate without an optical isolator. In contrast, EML implementations require an isolator to overcome their inherent high-reflection sensitivity.

The flexibility of the PLANEX ECL design makes it possible to further automate assembly using Silicon “Optical Bench” technology, allowing for efficient integration of the optical laser driver, gain chip, ball lens, and other components on a single silicon platform containing the gain chip and waveguide grating.

As shown in Figure 1, the miniature size and high degree of integration inherent to the PLANEX approach allows implementation of complete optical transmitter modules that are compatible with the smallest TOSA form factor specified by the XFP and XMD MSA for 40 km and 80 km. At the 40 km distance, the ECL approach offers a better than 30% cost advantage over comparable EML devices. Scalability of the high-yield model provides even greater cost benefits for 80 km and DWDM applications.

Deployment & Ownership
PLANEX technology offers a high degree of manufacturing flexibility for producing wavelength-specific DWDM devices, which enables carrier and network administrators to more efficiently migrate from TDM to DWDM, using the same basic component design and TOSA form factor. Unlike DFB and EML manufacturing, where the wavelength decision must be made at the early stage of wafer processing, the ECL wavelength is set during the packaging process using pretested components. This also helps to reduce the lead-time for DWDM ECLs and simplifies the carriers’ inventory management.

Since the gain chip used in the ECL has a broad bandwidth (with a single gain chip covering almost the entire Cband), all DWDM channels in the Cband can be supplied by using a high-volume gain chip and different (low-cost) PLCs. This mitigates the need to test and qualify additional modules and also eliminates the cost of adding a separate wavelength-locker component. In contrast, EML devices typically need a wavelength locker because of their higher sensitivity of wavelength to temperature variations, which is typically ~100 pm/°C, or 10X the sensitivity of the ECL.

Performance
The PLANEX approach improves performance by tightly controlling chirp, maintaining consistent eye and Bit-Error-Rate (BER) characteristics.

EML chirp is transient by nature; however, monolithic EML devices exhibit undesirable additional “adiabatic chirp” due to internal reflections, which limit transmission distances. Hybrid EML devices address this issue through the addition of an optical isolator between the DFB and modulator to eliminate chirp, but pay the penalty of added complexity and cost. In comparison, planar ECLs exhibit low, reproducible chirp characteristics that have no adverse impact at 40 km. The same basic devices can be manufactured to provide reproducible “negative chirp” for reliable transmission at 80 km, with costs comparable to 40 km.

Planar ECL product implementation has demonstrated performance characteristics that consistently meet or exceed the specified performance of industry standards such as Telcordia GR235 OC192 and 10 Gigabit Ethernet. BER measurements taken back-to-back (40 km and 80 km) with the ECL setup for the same operating conditions and an extinction ratio of 8.2 dB are shown in Figure 2. The results are within target performance specification, showing a small dispersion penalty of 0.2 dB at 40 km and 2 dB at 80 km at 10-10 BER.

These results represent the industry’s first demonstration of a directly modulated ECL that can transmit 10 Gb/s through 80 km of SMF; thus, confirming the viability of this approach for TDM and DWDM optical transmission systems. In addition, continued development of PLANEX-based hybrid ECLs and Silicon Optical Bench integration provides the technology platform and building blocks for more advanced and more integrated next generation products.



This article was written by Radu Barsan, President and CEO of Redfern Integrated Optics (RIO) based in Santa Clara, CA.

Dr. Barsan has 25 years of commercial experience in semiconductor and optical components development, operations, and general management — Phaethon Communications, Cirrus Logic, AMD, and Cypress Semiconductor — and a Ph.D. in Applied Sciences from the Catholic University of Louvain, Belgium. For more information, call (408) 970-3500 or visit www.rio1.com.