As telecommunications carriers strive to increase the number of channels in their wavelength-division multiplexing systems from 40 to 80 to 160 and beyond, they’re faced with having to plug in more and more lasers, one for each wavelength; they also need to have enough spares on hand to replace defective ones.
But widely tunable lasers, several of which are starting to ship to customers, could cut down on the numbers of lasers in a system and eventually lead to new switching and routing procedures. For instance, at the National Fiber Optic Engineers Conference (NFOEC; Baltimore, MD; 9–12 July), Blue Sky Research (Milpitas, CA) presented its Programmable ITU laser, a tunable device intended to cover 80 channels across the C-band. The device has an output greater than 20 mW, a frequency accuracy of 3 GHz, and it tunes in less than a millisecond, the company says.
Bob Potenza, senior vice president of marketing and sales and one of the founders of Blue Sky, expects to be shipping the laser in volume by the first quarter of next year. The device uses an external cavity design with material properties controlled by current and voltage to tune the wavelength. An etalon locker locks the signal at the appropriate channel. Potenza would not go into detail on the technology but says that the tuning element is purely electro-optical, with no gratings, moving parts, or microelectromechanical systems (MEMS).
“There seem to be six approaches [to making tunable lasers] that I can count,” Potenza says. “Three in external cavities, one in VCSELs, and two in grated approaches.” An approach using vertical-cavity surface-emitting lasers (VCSELs) is being undertaken by the High-Performance Optical Components Solutions group of Nortel Networks (Brampton, Ontario, Canada). Their device uses MEMS to adjust the mirror on one end of the laser, thus changing the length of the cavity, and with it the wavelength. They’ve demonstrated tuning across 80 channels on the C-band. Tuning speed is typically under 10 ms, output is 20 mW, and wavelength accuracy is ± 2.5 GHz. Tom Dudley, vice president of marketing, says the laser should be qualified by October.
Agility Communications (Santa Barbara, CA) bases its tunable laser on sampled-grating distributed Bragg reflector (DBR) technology. It’s built on an indium phosphide chip with a mirror on each end and divided into four sections. The front and back sections provide coarse tuning; the middle sections provide fine tuning and gain. By applying different levels of current to each of the four sections, the device can tune through 100 C-band channels in less than 10 ms with a wavelength accuracy of ±3 GHz and 4 mW output. Vice president of marketing Arlon Martin says that because the company can integrate additional devices, such as an electro-absorption modulator, onto the chip, they can eliminate other equipment, such as an external lithium niobate modulator, reducing transceiver costs.
From spare to driver
The initial market for tunable lasers, which started out tunable across only four to six channels, is in inventory control for vendors and spares for carriers. For a 48-channel DWDM system, for example, a vendor has to maintain a supply of 48 models of single-wavelength lasers to cover the different channels. It takes three months to build the lasers, and the exact wavelength may not be known until the process is complete; plus, it’s hard to tell in advance what wavelengths will be needed. “From the vendor perspective it’s a nightmare to maintain that kind of inventory,” says Vladimir Kozlov, a senior analyst with market research firm RHK (South San Francisco, CA). With up to a 20% price premium, tunable lasers make more sense, he says.
But widely tunable lasers also hold the potential for new system architectures, in which the lasers are used for routing traffic and for dynamic provisioning, and that’s what the manufacturers are looking toward. Kozlov says it remains to be seen which technology for tunable lasers wins out. Edge-emitting lasers based on a distributed feedback or DBR design are an established technology, but they’re fairly expensive to make and produce lower output powers. VCSELs are cheaper and more powerful, but there is some concern as to how stable and reliable they are, Kozlov says. “The slowdown in the market might actually favor this VCSEL technology because it gives them extra time to address these concerns,” he says.
Martin sees a transition to tunable lasers over the next three to four years and expects the market will be worth $8 billion to $10 billion. “I ultimately believe there is room for a couple of different players for that market,” he says.
(By Neil Savage, OEMagazine, September, 2001)