Japan's NEL Targets Metro
As a spinout of what was once Japan's incumbent carrier, Nippon Telegraph and Telephone, NEL has acquired quite a reputation for making cutting-edge components, notably lasers, Fiber Bragg Gratings (FBGs), thin-film filters, and Arrayed Waveguide Gratings (AWGs).
About one third of NEL's sales come from overseas. And yet the company makes little effort to publicize its achievements outside Japan. It lets its products do the talking at tradeshows like the Optical Fiber Communication Conference and Exhibit (OFC). And it rarely issues press releases in English: The latest one on its Website is dated July 6, 2001.
So, what's going on at NEL? Quite a lot, as Light Reading discovered when it interviewed two executives in its Tokyo headquarters recently.
Unfortunately, NEL doesn't want us to disclose the identity of these gents, so we'll call them Mr. P, handling photonics, and Mr. E, handling electronics. Both of them played important roles in NEL's early development of 10-Gbit/s transmission technology in the late 1990s -- a development that helped enhance the company's reputation (and profits).
"We are regarded as the number one maker in many fields, or at least at the very cutting edge," says Mr. P. "We believe we can maintain this position and remain as a top runner... Our short-term focus is on metro as one of the important markets, but the second is the 40-Gig market."
Mr. P acknowledges the technical problems of running 40-Gig over legacy fiber (see 40 Gbit/s: Ready for Prime Time? and 40 Gbit/s Takes the Slow Lane). But NEL sees huge opportunities in unlit emerging markets, with China looking a particularly bright prospect.
Mr. E adds 10-Gbit/s Ethernet to the list of technologies being targeted by NEL. "The long-haul customers are decreasing, so we have to find new customers, and those are in metro, short haul, and 10-Gbit Ethernet," he says.
The focus on metro has led NEL to begin ramping up production while reducing manufacturing costs per chip. A new 20,000 square meter, 6-inch fab went into mass production last November. NEL won't give its capacity, but says it's four or five times bigger than the capacity of its previous 3-inch fab.
"Some customers had been saying that our products were more expensive than competitors', so we established this new factory," says the mysterious Mr. E. "We are able to get almost four times as many chips from the 6-inch wafer compared to the 3-inch wafer, without increasing the process cost so much."
In terms of specific products, here's what NEL has got in the pipeline:
Arrayed Waveguide Gratings
NEL is already a market leader in this field. It has particularly low-loss AWGs and is shipping AWGs in volume, unlike many other manufacturers (see Photonic Integrated Circuits.
The company's new developments targeting the metro market include an athermal AWG with 100GHz spacing, shown at OFC. Apart from dispensing with a temperature controller, the new AWG is scaleable up to four devices and 64 channels. This means that the customer may add additional chips -- that's to say, groups of channels -- that act to progressively lower costs as more channels are used, says Mr. P. "The unit price per channel for this AWG is the lowest among other technologies when several tens of channels are used," he adds. Full production will start in two months.
NEL isn't the only company developing an athermal AWG. Lightwave Microsystems Corp. demonstrated one at OFC, and at least two other companies offer alternative technology mux/demux devices that don't need temperature control. One is Lightchip Inc. (see Lightchip Intros Athermal Mux/Demux), and the other is BaySpec Inc., which is being acquired by Finisar Corp. (Nasdaq: FNSR) (see Finisar to Buy BaySpec ).
Lightchip makes a big thing out of improving reliability by eliminating temperature control. "It is simply ludicrous for carriers to risk losing terabits of customer data due to the failure of an inexpensive mux/demux heater," says Isadore Katz, Lightchip's president and CEO.
"We think scaleability will be the sales point," says Mr. P. Lightchip is a "potential competitor," he concedes, although it's "not so popular or known" (in spite of issuing more press releases -- and in reasonably good English).
AWGs often form the basis of other components, and NEL demonstrated a couple of examples of this at OFC -- a so-called Uniform Loss Cyclic Frequency (ULCF) AWG and a reconfigurable optical add/drop multiplexer (OADM).
The ULCF AWG "is completely different from previous products in terms of function," says Mr. P. "Previous products [worked by] multiplexing or demultiplexing optical signals. This product will work as a key component for wavelength routing with uniform loss over the band. Although it is at the prototype stage, the function and performance are quite original." (Check out All-Optical Switching Tutorial, Part 2, page 8, for how AWGs can be used to route wavelengths.)
"The cyclic-frequency characteristic is one of the best [ways] to realize NxN routing by N wavelengths," adds Mr. P. Further, the ULCF AWG maintains a loss frequency under 1.5 decibels across a 32-channel, 100GHz sample spectrum, according to NEL.
The reconfigurable OADM is an AWG with peripheral control circuits that dynamically add or drop all or any of up to 16 channels. This means that customers can set up and change their optical network configurations whenever they like. The device is one of the most compact around, claims Mr. P. "Almost all the functions needed are installed within the board. As far as dynamic OADMs are concerned, this is a cutting-edge product."
All sorts of other technologies are also being used to make reconfigurable OADMs. These include MEMS (micro-electro-mechanical systems) (see Onix to Demo Reconfigurable OADM and DiCon Intros 11 New Products ); liquid crystals (see Dynamic Gain Equalizers Diversify); and amplifiers (see OFC's Hot Products).
NEL has got a bunch of laser developments in the works that it expects to become mainstay sales next year.
They include uncooled 10-Gbit/s directly modulated Distributed Feedback (DFB) Lasers for the 1310nm-wavelength market. Mr. P says it's just the ticket for driving down the cost of metro and access equipment. "High bit-rate modulation without an external modulator and without a temperature control device is indispensable from the view point of system cost."
The big issue here is signal purity, which is measurable by eye-pattern performance. Mr. P says NEL often outguns competitors in this respect. However, NEL isn't alone in claiming this. OptronX Inc. has a competing product and claims to have the "best uncooled eye in the business" (see OptronX to Demo Transponder). Purer signals also go farther, and Agere Systems (NYSE: AGR) claims a record on that score -- with a 40 kilometer reach for its uncooled 10-Gbit/s transponder (see Agere Intros Transceivers, Transponders). Agilent Technologies Inc. (NYSE: A) is also in the running on distance records (see OFC Is Highly Transponderant).
NEL is also developing 4nm-wavelength Tunable Lasers that promise "superior" performance, according to Mr. P. It will ship within the next six months.
Other products in the pipeline include an OC768 (40 Gbit/s) transponder for testing applications, an OC192 (10 Gbit/s) super FEC (forward error correction) transponder, a 10-Gbit/s Ethernet PHY module for the metro market, and a 10-Gbit/s media converter chip for the access market. All of these are scheduled for release in late 2003.
— Paul Kallender, special to Light Reading