EDC not yet ready for 10-Gigabit Ethernet applications
UPDATED June 14 -- June 11, 2004 Nashua, NH -- Electronic dispersion compensation (EDC) offers significant promise for 10-Gigabit Ethernet applications, particularly as the heart of the 10GBase-LRM PMD. But both module vendors and chip manufacturers agree that the technology must mature further before that promise is fulfilled, Editorial Director Stephen Hardy reports.
June 11, 2004 Nashua, NH -- Electronic dispersion compensation (EDC) offers significant promise for 10-Gigabit Ethernet applications, particularly as the heart of the 10GBase-LRM PMD. But both module vendors and chip manufacturers agree that the technology must mature further before that promise is fulfilled.
The LRM will offer an alternative to the 10GBase-LX4 PMD, currently the only standard option for 10-Gigabit Ethernet transmission over 300 m of legacy multimode fiber. LX4 combats the dispersion hurdles such fiber presents by splitting the 10-Gbit/sec signal into four lower-speed channels using a WDM-based approach. As Lightwave will discuss in its upcoming July issue, module vendors have found the development of a cost-effective LX4 device a significant challenge. The serial LRM PMD promises a less expensive approach ¿ which means LX4 module developers potentially have a limited amount of time to perfect their offerings before LRM devices appear on the market.
However, the serial approach also most combat the dispersion inherent in legacy multimode -- and EDC appears the most likely choice. Yet, while several XFP module vendors have demonstrated devices incorporating EDC, no one with whom Lightwave spoke said they were ready to choose an EDC chip vendor.
"We're far from choosing a vendor -- we don't feel comfortable enough with anybody," says Christian Urricariet, director of marketing for high-speed optics at Finisar Corp., in a typical comment. "But we feel comfortable that it will evolve very soon to a point that it will."
Part of the issue, of course, is that until the LRM PMD is ratified, it's difficult to say exactly what the EDC device has to accomplish or to measure whether it has succeeded. "There's no conformance test that tells me that what they [EDC suppliers] have done is going to work," offers Ed Cornejo, director of datacom transceivers at Opnext Inc. "Interoperability is a big question, right? So right now, I don't think it's a proven thing yet."
The lack of standards-imposed guidelines has affected what module vendors currently want to see from an EDC device. "Everyone just wants to see something work. And it's a complex enough technical problem so that people don't really care as much about as power, cost, or form factor in this particular phase," explains Dan Trepanier, president and CEO of IC vendor Quake Technologies Inc.
The focus may change by this September, when the IEEE 802.3aq Task Force in charge of developing the LRM hopes to have a draft in place. While performance that conforms to the PMD will be a requirement, features that may vary from module to module include chip count, size, power, and degree of functional integration. For example, Cornejo points out that devices targeted for the XFP form factor will need to provide EDC capabilities while keeping the transceiver within the MSA's power budget. The tight space constraints would point toward providing the EDC function in an integrated fashion, perhaps with the clock and data recovery/retimer function. XENPAK modules offer more leeway for a standalone device, although integration with the XAUI serializer/deserializer is a possibility, Cornejo believes.
Several IC companies are lining up to serve the market. Companies such as AMCC (a two-chip approach that combines EDC with forward error correction), Big Bear Networks (one chip that combines electronic equalization with CDR and SFI-4 demultiplexing), Scintera Networks Inc. (standalone chip), and Phyworks Ltd. (an EDC/retimer combination targeted at XFP modules) have already announced devices. Sources at Mysticom, Quake, and Vitesse Semiconductor confirmed their companies have EDC programs underway. Vitesse plans a family of products by early next year, with Mysticom following a similar path; Trepanier did not offer a timeline for Quake's efforts. Meanwhile, Agilent has an internal EDC program in parallel with their investigation of off-the-shelf devices.
Meanwhile, ClariPHY Communications, a startup based in Orange County, CA, has made joint presentations with other vendors in the IEEE 802.3aq Task Force meetings. While the company has not yet made much of a splash publically, ClariPHY also is working on EDC technology and is in early stage chip development, according to a company source.
Both module vendors and IC suppliers remain bullish about the eventual success of LRM devices in general and EDC in particular. "I think there's a ton of work that needs to be done, but it's happening," says Tom Fawcett, worldwide marketing manager for Agilent's Fiber Optic Products Division, regarding EDC. "There's a strong support infrastructure within the industry to get the work done. A lot of people have a vested interest in seeing it get done up and down the value chain. So I would rate it as on track."
-- S. Hardy