The optical network market is less welcoming than it once was, which means widely tunable lasers emerging as commercial products have extra pressure to deliver cost, performance, reliability and an acceptable footprint. In the ’90s, visions of the all-optical network danced merrily in the heads of the world’s telecommunications service and equipmentproviders. Now the optical components necessary to transform that all-optical fantasy into a reality are emerging. One of the most critical components — the widely tunable laser — enables network elements to convert one wavelength into another on the fly. A host of component manufacturers have products that are nearly ready for prime time. The trouble is that prime time has been pushed back indefinitely. The economic downturn has all but stopped telecommunications service providers from purchasing next-generation optical transport and switching systems in which widely tunable lasers would debut. “The market is dead right now,” said Jake Weise, vice president of marketing for Bandwidth9 Inc. of Fremont, Calif. “However, I, like most of the colleagues I have, believe that the telecom market is just taking a breather.” While the market catches its breath, interest in and activity surrounding widely tunable lasers remain keen. Approximately 10 component makers are offering or developing products. Each company has a unique approach to making them as cost-effective, reliable and easy to use as the fixed-wavelength distributed feedback lasers that they aim to replace. Several large companies, including ADC Telecommunications Inc., Bookham Technology plc, Intel Corp. and Nortel Networks Corp., have offered up widely tunable lasers, but it’s the 1- to 3-year-old start-ups that are creating most of the buzz. Several, such as Agility Communications Inc., Bandwidth9, Iolon Inc. and Santur Corp., are shipping or sampling products. Picarro in Sunnyvale, Calif., is among those that remain in stealth mode. Although carriers are not spending money on new switching or transport gear, their interest in tunables remains high. “We are deploying narrowly tuned lasers with great success,” said Allen Shipman, senior manager of network planning for Broadwing Communications Inc. in Cincinnati. “And we are very interested in having a wider band of tunability.” If the market recovers slowly, use of widely tunable lasers may reach parity with narrowly tuned devices. All things being equal, Shipman said he would rather just deploy widely tunable lasers. Based on discussions with system vendors, he said he does not expect Broadwing to begin lab tests of widely tunable laser-equipped transport and switching systems until later this year or early next year. In the meantime, the company has worked out what he refers to as a “manual” tunable laser program with its vendors. “We pull a card out and send it back to the system vendor to get a laser with the frequency we need, and they send it back to us in 24 to 48 hours,” he explained. Shipman thinks that widely tunable lasers will find their initial niche in ultralong-haul systems, where demand for bandwidth on routes varies more — and more often — than it does on long-haul or metropolitan routes. Broadwing’s announced ultralong-haul vendors, including Corvis Corp. of Columbia, Md., are all interested in widely tunable lasers, he said. Decisions, decisions, decisions Just as tunable lasers provide carriers with the ability to choose a new transport wavelength, laser suppliers are providing system manufacturers with a variety of technologies. The choices fall into four broad categories: distributed feedback, distributed Bragg grating and external-cavity lasers and vertical-cavity surface-emitting lasers (VCSELs). The distributed Bragg grating laser, formed here by metallorganic chemical vapor deposition, is a more mature and common tunable technology. But cheaper, more easily manufactured vertical-cavity surface-emitting lasers appear to be gaining market share. Courtesy of Agility Communications. The base technology of tunable lasers is important to system makers and carriers. However, it is not the only factor that determines which laser they select for their switches, add/drop multiplexers or transport systems. “From our perspective, our package size and drive requirements don’t change. It’s the tunability functionality that we are after. You could use a variety of different technologies to get that,” said Jeff Livas, vice president of systems technology for Ciena Corp. in Linthicum, Md. “At this point in time, the most important criteria to measure widely tunable lasers against fixed-wavelength lasers are cost, performance, reliability and footprint.” Although the cost and performance in C-band tunables are close to that of fixed-wavelength distributed feedback lasers, there is a wider gap in their reliability and footprint, he added. In the near term, Ciena plans to release systems with 8- and 16-channel narrowly tuned lasers at the beginning of 2003. “We would like to be able to take our standard products today, which have single-frequency distributed feedback lasers on their circuit boards, and pull them off and stick a widely tunable laser on it,” Livas remarked. Shipman said that distributed Bragg gratings seem to have the lead right now, but that VCSELs, which are cheap and easy to manufacture, seem to be gaining market share in all areas of laser deployment. However, he added, VCSELs’ low output power makes them an unlikely technology choice for ultralong-haul applications. Power/cost trade-off Atoga Systems of Fremont, for instance, selected a distributed Bragg grating laser from Agility for use in two transport platforms that deliver Ethernet or time-division multiplexing service to multiple customers via existing Sonet rings. The device uses an electroabsorption modulated laser that can transmit a 2.5-Gb/s signal up to 200 km, said Arlon Martin, vice president of marketing for Agility in Santa Barbara, Calif. When Atoga was developing its platform, Agility’s distributed Bragg grating-based laser was more mature than external-cavity lasers or VCSELs, explained P.G. Menon, vice president of marketing for Atoga. However, the company is working with other component manufacturers that make VCSEL-based devices. “We are not particularly tied to distributed Bragg gratings. VCSELs can work as well. If you use a VCSEL, there is a power issue, but it’s cheaper. Distributed Bragg gratings give you more power, but they are more expensive,” Menon said. Where more power is needed, VCSELs could be used in conjunction with erbium-doped fiber amplifiers. Although amplifiers are expensive, when combined with VCSEL-based widely tunable lasers in metro transport systems, the total system cost is “pretty much a wash,” he said. n general, Menon said, widely tunable lasers provide system vendors and their service provider customers with three important advantages: • Less-expensive spares and less inventory. • The ability to add wavelengths and increase data rates on a link without having to roll a truck. • The ability to adapt the network to traffic patterns as they change throughout the day. For instance, a service provider could use tunable lasers to configure its network to carry data between multiple sites for a customer during the day and then reconfigure it to transport traffic from multiple locations to one or two storage sites at night. Testing 1, 2, 3 Although Atoga’s customers were skeptical about tunable lasers a few years ago, they have grown more and more comfortable with widely tunable lasers after having seen them handling live traffic in the company’s laboratory. Agility is promoting its widely tunable laser modules with the fact that they have met and have exceeded the reliability assurance requirements contained in Telcordia Technologies’ GR-468-CORE document. This, Martin claims, enhances their chances of success with the regional Bell operating companies. Agility expects local carriers to start deploying these lasers in their networks by early 2004. Most other tunable-laser makers are beginning or completing Telcordia testing. Telcordia qualification increases customer comfort with the new technology, but it is not a “do or die” proposition. “Ideally, we try to make sure all of the equipment we deploy meets the highest industry standards, but certification of this type is not always a requirement,” said Broadwing’s Shipman. “For instance, we have seen fixed-wavelength lasers from two different vendors go through the same testing, and yet we still see different failure rates.” San Jose, Calif.-based Iolon, which is in the final stages of applying Telcordia qualification testing to its Apollo family of external-cavity lasers, snagged customer Innovance Networks before testing was done. “Iolon is using well-established components put together in a new configuration,” said Alan Solheim, chief technology officer for Innovance. The Piscataway, N.J., company is using Iolon’s externally modulated laser in its Agilecore product portfolio for tuning across the L-band. He said that external-cavity lasers offer higher power and spectral purity. They have been more difficult to build than distributed Bragg gratings or VCSELs, but he said that Iolon has solved some of those issues by using a wavelength locker and the microelectromechanical systems (MEMS) expertise borrowed from the disk-drive industry. Innovance has other sources for widely tunable lasers, which Solheim declined to name. However, he said that Picarro and Santur have offered some interesting solutions that have caught his eye. Santur, another Fremont company, made its debut at the Optical Fiber Conference in Long Beach, Calif., in March. It offers what is essentially an integrated array of distributed feedback narrowly tuned lasers, said Julian Osinski, vice president of product marketing. Its C-band device contains 12 integrated distributed feedback lasers, and its L-band product contains 14. Each laser tunes in 3-nm increments. A wavelength locker, a MEMS tuning mechanism and lenses are built into the externally modified devices. Using distributed feedback lasers reduces cost structure and improves reliability, he said. “Price is everything. No one is going to adopt technology unless the price points are there.” Picarro proposes that widely tunable lasers must provide plug-and-play performance equal to or better than the distributed feedback lasers that system vendors are using today, and must do so at the same price. Before network engineers adopt widely tunable lasers, the technology must deliver plug-and-play performance and a price equal to or better than the distributed feedback lasers used today. Courtesy of Santur Corp. “The reason why widely tunable lasers are not taking off is not limited to the economic downturn,” said David Hughes, vice president of marketing. “The current generation of tunable lasers does not meet all of the technical requirements that customers have.” He said widely tunable lasers must meet all the criteria listed below: • 20 mW of output power to fit current system designs. • 40-dB side-mode suppression ratio. • 140-dB relative intensity noise. • Tunability across an entire band (40 nm). • 25-, 50- and 100-GHz channel-spacing options. • Small-form-factor laser, control electronics and modulator to avoid system redesign. • Cost parity with existing distributed feedbacks. “Most of all, you want to provide customers with products that are compatible with the way things are being done already,” Hughes added. While suppliers seek to produce the perfect widely tunable laser, many are latching on to a more concrete trend — integration. Large layoffs at the big system houses have decimated in-house design teams that would ordinarily be charged with integrating discrete components into their systems. Today, widely tunable laser products include the diode and its control circuitry. In the future, these lasers will be incorporated into a transponder, said Hendrik van der Meer, product marketing manager for Intel. “If you can integrate the mux/ demux, transmitter/receiver — be it [avalanche photodiode] or pin — into one package, then you saved a Lucent or Nortel a huge amount of designing effort they normally would have had to put forth,” he explained. Intel recently purchased New Focus Inc. of San Jose and is developing a temperature-tuned external-cavity laser for use in transponders. It plans to roll out a widely tunable module as well, van deer Meer said. Thanks to the economic downturn, optical switching and transport equipment manufacturers have had time to evaluate the offerings of several makers of widely tunable laser components. Although mature product has made its way into next-generation systems, service providers have yet to begin ordering and deploying those systems in volume. If and when the market regains its momentum, system makers will be looking for widely tunable lasers that will deliver cost, performance, reliability and an acceptable footprint. They do not want to redesign their existing systems just to realize the advantages these lasers have to offer. Meet the author Annie Lindstrom is a free-lance telecommunications writer based in Cape Coral, Fla. Standards Are Being Crafted When the development of widely tunable lasers for optical networking applications began, the end market was much more forgiving of “disruptive” technologies. In today’s market, most components can benefit from standardization. Hence, for the past year, the Optical Internetworking Forum in Fremont, Calif., has been developing a standard communications control interface and form factor for widely tunable lasers. “We’d like tunable lasers to be fit-and-function replaceable,” said Karl Gass, vice chairman of optical interfaces for the forum’s physical and link layer working group. “That is not realistic today in that we have different capabilities based on different technologies.” Approximately 15 to 20 companies are contributing to an implementation agreement defining the form factor for a standard device. The agreement specifies size, pin location and output power, and also defines how systems will communicate with widely tunable laser modules. “With a standard interface, setup is done once and carriers can talk to any device, regardless of its capabilities,” Gass said. The group plans to have the document ready for formal ratification by the end of the year, although component manufacturers building devices now are following what has been defined to date. One of the biggest issues the group has tackled has been to narrow down the potential communications interfaces for modules. “We want the agreement to last several years, so we have tried to be very forward-thinking in how we specify that interface,” Gass said, adding that power dissipation and size of the form factor have been the other major concerns.