Finisar unveils 40-km 10-Gbit XFP transceiver for DWDM apps
At NFOEC last September, Finisar (Sunnyvale, CA) launched a 10-km ultra-small-form-factor 10-Gbit/sec XFP transceiver that occupies one-fifth the space and uses one-half the power of currently available transponders. Nine months later, it is ready to unveil a 40-km version, which it claims is the industry's first 10-Gbit/sec XFP transceiver to support DWDM applications.
The XFP multisource agreement (MSA) was established in March 2002 to develop a smaller-footprint, lower-power device at 10 Gbits/sec for high-density applications. Unlike the transponders currently available or in development—including those established by the XENPAK, XPAK, X2, and 200-pin and 300-pin MSAs—the XFP transceiver does not include an internal multiplexing/demultiplexing function. All multiplexing/demultiplexing is done on the host board; the transceiver communicates with the board via a 10-Gbit/sec serial electrical interface known as XFI.
The typical transponder consumes between 4 and 6 W, explains Christian Urricariet, director of marketing for the high-speed optics division at Finisar. In contrast, the XFP transceiver consumes just 2-2.5 W. Combined with the module's small size, the power savings enables a higher board density; the XFP module supports up to 16 ports per line card.
The transceiver itself is hot-pluggable and fits inside a specially designed metallic cage that is soldered on the host board. The cage contains a clip-on removable heat-sink, which makes contact with the transceiver when it is inserted into the cage. Unlike the XENPAK or 300-pin transponders, the XFP is not directly attached to a heat-sink. "Thus," says Urricariet, "the OEMs can use either the standard heat-sink defined in the MSA standard or design their own heat-sink. It gives them the flexibility to meet the specific power dissipation requirements of their boards, which depends on the size of the board, the geometry, air flow characteristics, etc."
Another distinguishing characteristic of the XFP transceiver is its ability to provide digital diagnostic functions via a serial communications bus with the host board. The transceiver continually measures in real time such key parameters as the transmitted and received optical powers, the transceiver temperature, the laser bias current, and the DC supply voltage, enabling OEMs to perform preventive maintenance on both the transceiver and optical connector.
Unlike its 10-km cousin, which employs a directly modulated 1310-nm distributed-feedback (DFB) laser, the 40-km transceiver features a 1550-nm electro-absorption-modulated laser (EML). "When you go to longer distances—particularly in the 1550-nm window at 10 Gbits/sec—you have to worry about chromatic dispersion," admits Urricariet. An EML exhibits less chirp than a DFB [laser] and its temperature is more easily controlled, making it better suited for longer-distance applications.
The folks at Finisar drove the creation of the XFP standard and claim to be the first to demonstrate a 10-Gbit/sec transceiver in this form factor that supports DWDM applications. And while they contend that Gigabit Ethernet OEMs and even the typically conservative SONET OEMs are extremely interested in their device, how much demand will there be, given the current state of the long-haul market?
"The long-haul is actually not as dead as some people would have you believe," asserts Lawrence Gasman, president and director of optical components research at Communications Industry Researchers (Charlottesville, VA). "On the other hand," he admits, "it's certainly not in good shape right now." That said, a 40-km transceiver could also be used in larger metro networks, and given the price sensitivity of the metro, any device that lowers costs will not go unnoticed.
"It's a good thing to have done," asserts Gasman. "Finisar is well known and has a huge customer base—and people are looking for this kind of product. The metro WDM market has been severely constrained by cost, and maybe this will help the margins a bit. As for the long-haul market, our view is that capital expenditures are going to be pretty rotten for the rest of the year but will begin to pick up a little bit next year, so maybe you will see some traction there," he surmises, adding that almost any optical product is similarly constrained right now by the optical slump.