By MEGHAN FULLER
A popular misconception of fiber Bragg gratings (FBGs) is that they are an emerging technology, when, in fact, they have been shipping since 1997. A key component in WDM pump modules, particularly at 980 and 1480 nm, FBGs have seen their WDM market share dwindle in the current economy, prompting manufacturers to diversify their product lines and apply their grating expertise to new applications.
An FBG is a filter used in such applications as channel separation or signal conditioning. The filter is written into the core of the fiber via the interference of two ultraviolet (UV) beams from a UV laser. This interference pattern forms a periodic refractive index change longitudinally along the fiber for a length of about 10 mm. Each refractive index change or jump acts as a series of reflectors, reflecting back a small amount of light. But because there are typically tens of thousands of these index changes in a row, they add coherently; light traveling down the fiber will reflect strongly off the grating structure.
In other words, FBGs transmit most of the light going through the fiber except for a narrow stop band where the light is reflected back up the core in the direction of the signal entering the fiber. Typically, for a 1-cm-long grating, the bandwidth of the reflected signal is about 0.4 nm.
FBGs are in-fiber devices-the light never leaves the core of the fiber-which makes them inherently reliable, says David Psaila, director of technology for FBG products at JDS Uniphase (Melbourne, Australia).
Of course, FBGs have their disadvantages, which are primarily related to scale. Because the gratings work in serial (they cascade), insertion loss aggregates across the entire device. Arrayed-waveguide gratings (AWGs), on the other hand, are parallel devices, so they do not have a problem with aggregating insertion loss. FBGs also compete with thin-film filters (TFFs), which, like AWGs, offer lower price points than FBGs. The cost of an FBG scales linearly; every time a new channel is added, another FBG must be added as well. Also, FBGs must be used with circulators, which themselves can cost upwards of $500.
That said, Adam Reeves, senior analyst at Communications Industry Researchers (CIR) Inc. (Charlottesville, VA) and author of a new report, "Fiber Bragg Gratings: Winners and Losers in a Crowded Market," believes that the newest high-performance FBGs outperform TFFs and AWGs on a single-channel basis.
FBG supplier glut
There are at least 12 companies that manufacture FBG-based DWDM filters-many of which received significant funding over the last two years. However, the market for such FBGs is relatively small, as few system and subsystem manufacturers favor FBGs over other alternatives, admits Reeves. And Ciena, arguably the largest consumer of FBGs, manufactures its own.
Five vendors-Corning/Pirelli, MRV/Luminent/ FOCI, Spectra Physics, Superlightwave, and Ionas-have already exited the market, and Reeves believes others may be forced to follow suit. "Companies are going to have to differentiate themselves based on the quality of their products, the long-term viability of their business," he contends. "I am not convinced that any company that is focusing solely on DWDM FBGs is ever going to be able to survive."
However, there are certain steps FBG vendors can take to increase their chances of survival. First and foremost, says Reeves, they should leverage their FBG expertise to tackle new market segments, which could mean moving vertically in the market and integrating their FBGs into more complete products and subsystems. Most vendors, however, have chosen to diversify their product lines to encompass other FBG-based components such as gain-flattening filters (GFFs), wavelength lockers, and dispersion compensators.
The first GFFs were done with TFFs, says Erick Pelletier, product management director at TeraXion (Quebec). However, new requirements for higher accuracy and lower error functions favor FBGs. TFFs cannot achieve the necessary shapes with a single filter design; they need multiple filters to meet the complex shape requirements.
"Another advantage of the FBGs in GFF is that each grating is written individually as opposed to the TFF method of creating gain-flatteners where you create a whole batch that has roughly the same shape," explains JDS's Psaila. "When you start concatenating in a long amplifier chain, if all the filters have exactly the same profile, any small systematic error that deviates from the target shape a supplier has specified will accumulate, and you'll end up with a large error at the end of the line. FBGs avoid this because they are not made in a batch process."
A variance of the GFF is the shape correction filter, used to correct deviations in erbium-doped fiber-amplifier (EDFA) flatness after long-haul and ultra long-haul systems have been tested or deployed. Due to variations in the erbium concentration in EDFAs, there is usually a residual nonflatness to the gain structure that requires a second-order correction filter. FBGs have emerged as the technology of choice for this application, says Psaila. "The benefits of FBGs are (1) they can meet that shape, (2) they can do it with fairly low loss, and (3) because of the automated computer-controlled technique by which these gratings are written, we can turn complex filters around in a couple of weeks," he explains.
FBGs are also emerging as options for wavelength lockers or stabilizers, particularly in the 980- and 14xx-nm windows. FBGs are particularly important in Raman amplifiers, because light has to be pumped 100 nm away from the DWDM wavelength; locking in the center wavelength is therefore critical.
Demand for Raman amplifiers has slowed somewhat in the current economy. However, Benoit Lavigne, chief operating officer of Alcatel Optronics and co-founder of Innovative Fibers, an FBG supplier bought by Alcatel in the summer of 2000, contends that "a lot of people still believe in it. When the market improves and Raman amplification picks up, gratings will be necessary for this application."
As the transmission speeds increase, dispersion management or compensation has emerged as a new application for FBGs. In the migration from 2.5 to 10 Gbits/sec, dispersion management issues were mainly addressed by specialty fiber, particularly dispersion-compensating fiber (DCF). Traditionally, fixed FBGs have been regarded as a lower-loss alternative to DCF.
Fixed dispersion-compensating gratings (DCGs) are chirped-the periodicity of the refractive index changes along the length of the fiber-which results in a broader filter. Used mainly in 10-Gbit/sec systems, fixed DCGs can compensate for 20 or more channels simultaneously.
The move to 40-Gbit/sec transmission, however, will require channel-by-channel fine-tuning, claims Alcatel's Lavigne, and both DCF and fixed DCGs are inadequate for such an application. For FBGs to operate optimally in a dispersion compensation capacity at 40 Gbits/sec, they will have to be tunable.
FBGs are highly sensitive to temperature; tunability is directly related to thermal stability. FBGs must be athermally packaged for applications in which stability is critical, such as wavelength locking, because the center wavelength, when left in free air, will shift at a rate of 10 psec per degree Celsius. When a chirped FBG is heated from 0° to 100°C, the center wavelength shifts a full nanometer.
Tunability can also be achieved by stretching the grating, according to Pelletier. Vendors have achieved a 10-nm tuning range using this method.
"Currently, tunable DCGs are being designed-in to major system vendors and will become the initial technology of choice for 40-Gbit/sec dispersion mitigation," contends JDS's Psaila. "It's really the only technology available at the moment to do tunable dispersion compensation at 40 Gbits/sec in single-channel devices."
Niche applications for WDM
According to CIR, FBGs are not expected to capture the largest share of the captive WDM filter market; they may offer better performance specifications than AWGs and TFFs, but the other two technologies are less expensive and therefore more often deployed. Nevertheless, FBGs should find a niche in low-channel-count applications, particularly as the market moves to 50- or even sub-50-GHz channel spacings.
"FBGs are the only technology currently available that can meet those very tight channel spacings," asserts Psaila. "There are no 25-GHz thin-film designs. AWGs have trouble with crosstalk at 25 GHz, and other technologies like etalons are difficult to construct and have trouble with the roll-off as well."
FBGs are limited in terms of channel count, however. Gratings at 50 or 25 GHz function well up to 16 wavelengths, but higher channel counts create insurmountable problems. "At 40 or 80 wavelengths, you will have to go to some kind of AWG or PLC [planar-lightwave-circuit] technology," admits Alcatel's Lavigne. "Otherwise, the insertion losses created by the gratings and the optical circulators necessary to create multiplexers, demultiplexers, or optical add/drop multiplexer modules are too great."
That said, Lavigne says that in some applications the technologies are not mutually exclusive-gratings can be written directly into waveguides. "You could do Bragg grating filters, for example, with a very wide band that drops a group of wavelengths, and then you can further split it up with an AWG with very low insertion loss," he explains. "As WDM functions are becoming more and more complex, people are addressing these issues not with a single technology but actually with a combination of technologies."
CIR's Reeves believes there is a long-term market for FBGs. "All three of the filtering technologies will coexist in the long run," he says. "It's just a question of FBGs trying to grow their segment of the market-meaning low-channel-count, high-performance requirement applications. They need to keep pushing the envelope of high performance."
Startup Sabeus Photonics Inc. (Chatsworth, CA) has recently unveiled plans for a line of fiber Bragg grating (FBG)-based, high-concept optical devices-"designed to do new things in new ways, which required a fundamental rethinking of the physics [of FBG fabrication]," admits Dmitry Starodubov, chief technology officer and director of Sabeus.
In the typical method for creating FBGs, the fiber must be stripped, exposed to ultraviolet (UV) light, then recoated-a process that "compromises the integrity of the fiber," contends Starodubov. He and his team tinkered with the physics of index change, using different wavelengths of light to write the gratings. The standard is 240 nm, but the folks at Sabeus discovered that polymer fiber coatings absorb light starting from 300 nm. Using longer wavelengths of 330-350 nm, Sabeus is able to write gratings through the protective polymer coating and cladding of the fiber, directly onto the core.
According to Starodubov, this process, known as "Cold Writing," is simple, repeatable, and automation-friendly. The company has also developed a computer-controlled machine specifically designed to automate the Cold Writing process and robotically produce the gratings.
The resultant cost savings are just what the industry needs right now, surmises Starodubov. "We need to meet new industry demands," he says, "which is reducing cost without sacrificing performance."