LX4 vendors fight back
In response to the attention the in-process 10GBase-LRM serial-transceiver/transponder standard has generated, vendors with modules based on the existing 10GBase-LX4 approach have launched a trade group aimed at convincing both systems vendors and end users that there’s life in the old 10-Gigabit Ethernet (10-GbE) PMD yet. LX4 devices are more cost-effective and manufacturable than some would have you believe, the trade group says-and unlike potential LRM offerings, you can get your hands on them now.
The LX4 Trade Group (www.lx4.org) includes trans- ceiver/transponder vendors Emcore (Lombard, IL), Merge- Optics (Berlin), and Opnext (Eatontown, NJ) as well as IC provider Quake Technologies (Ottawa, Ontario) and optical-component/module vendor Wavesplitter Technologies (Fremont, CA). A sixth company was expected to be added to the roster by the time the trade group made its official debut after this issue of Lightwave went to press.
As reported previously in Lightwave (see “LX4 development races LRM standards process,” July 2004, front page), the LX4 was one of the original 10-GbE PMDs. Although it can handle singlemode fiber, the transponder specification aimed primarily to accommodate 300-m distances over legacy 62.5-µm fiber. But due to a combination of technical complexity-the transponder uses a WDM approach wherein four wavelengths of 3.125 Gbits/sec carry the aggregate 10-GbE traffic-and a lack of demand, few such devices had made their way to the field by the beginning of this year. Yet, as major Ethernet switch vendors such as Cisco Systems began to receive end-user requests for legacy multimode support at 10 Gbits/sec, the question of whether the LX4 could do the job economically led to the exploration of a serial approach: the LRM. As currently envisioned,
the LRM would offer a single-wavelength design buttressed by electronic dispersion compensation (EDC) techniques. (For more about the LRM, see Tech Trends, p. 15.) But with a clear signal from major OEMs that legacy multimode support had moved back to the front burner, several module manufacturers dusted off their LX4 designs and worked to reduce cost and complexity. As a result, say members of the LX4 Trade Group, OEMs and end users don’t have to wait until the end of next year or the beginning of 2006 for a viable standards-based solution to their legacy infrastructure problems.
And the trade group is determined to make sure this message reaches the market. According to Bryan Gregory, vice president of marketing for Emcore’s Fiber Optics Business Unit, the group will sponsor interoperability tests and educate both OEMs and end users about the advantages of the LX4. “The world needed to know that this is a robust standard, that it’s supplied by multiple sources, [and] it fits the needs for legacy 300-m fiber,” he says.
Adds Edward Cornejo, director of product marketing at Opnext, “I was getting concerned that there was a lot of [misinformation] being put out into the industry that LRM was here and was real and was available now or later this year-and that’s not the case. We’re very active in the standards process, and there are still many significant issues to work through-not to mention the distance. Right now, they’re at 220 m, versus 300 m, which is what LX4 supports.”
The bottom line for trade-group members, says Cornejo, is that “most of our customers are committed to LX4, but I think there are probably a few, maybe the smaller ones, that weren’t sure because of the information that’s been put out in the industry.”
Some of that information revolves around the complexity of LX4 designs. While each supplier has a different approach-Emcore, MergeOptics, and Eudyna Devices have announced products, with Opnext and others expected to follow shortly-all have made significant advances in packaging improvements and cost reduction, Gregory and Cornejo say.
“If you look at the internals of our module and how it’s manufactured, it’s a fairly straightforward product. I mean, you’ve got four lasers that are in there, so we use four laser sources that are packaged in a way that’s proprietary,” Gregory offers by way of example. “It was certainly a difficult design and we had to put a lot of resources to it, but once we had it finished, we can ramp this thing very nicely.”
“Our approach is basically a single-chip approach with four wavelengths on it,” says Cornejo. “Keep in mind that we’re always running at a quarter of the speed [per wavelength]-we’re not running at 10 Gbits. So the chip-level cost is very similar to what a serial 10-gig device would cost you. [The design] also has done a lot to improve our alignment capabilities. In fact, we do mostly passive alignment; we do have some active alignment, but it’s a very small part of our assembly.”
Gregory and Cornejo point out that LRM designers have a few complexities of their own to overcome, including how to incorporate EDC and a 10-Gbit/sec laser source into a cost-effective package. In fact, both sources say they believe LX4 offerings will be cost-competitive with LRM devices, particularly given their expectation that LX4 devices will be ramping toward volume production when LRM counterparts first become available.
“If what we’re being told is true, once you have an LX4 solution that’s fully qualified and released, it should be shipping in fairly high volumes from the beginning-because the ports are out there already,” says Cornejo. He put the cost target at “no more than an LR,” which is one of the singlemode 10-GbE PMDs, and expressed confidence that LX4 offerings will meet this goal.
Another important milestone will be the availability of both XENPAK and X2 devices. Cornejo asserted that if at least two stable suppliers of LX4 XENPAK and X2 modules have not arrived in the market by the end of next year’s first quarter, the momentum behind the LRM will swell significantly. X2 sales, Cornejo reports, could quickly outstrip XENPAK shipments.