New Working Group tackles 40G short-reach serial interfaces

March 26, 2007
MARCH 26, 2007 By Meghan Fuller -- Today marks the first meeting of a new working group established to address the need for a cost-effective 40-Gbit/sec serial transponder for short-reach applications.

MARCH 26, 2007 By Meghan Fuller -- Today marks the first meeting of a new working group established to address the need for a cost-effective, 40-Gbit/sec serial transponder for short-reach applications. Dubbed the Serial Short Reach 40-Gbit/sec Transponder (SSR-40) Working Group, the new group includes founding member Kailight Photonics, along with Inphi Corp., Picometrix, Sanmina-SCI, SHF Communication Techologies AG, and Sierra Monolithics.

One of the biggest issues facing the 40-Gbit/sec market today is the need for a low-cost, short-reach (i.e., less than 2-km) interface for client-side connections.

This isn't the first time a group of vendors has collaborated to address this issue, however. In November 2006, nine vendors, including Aeluros, Broadcom, Emcore, Finisar, Infinera, Juniper Networks, MergeOptics, Tyco Electronics, and Vitesse, created the X40 multisource agreement (MSA) to develop a multirate optical transceiver to support a range of 40-Gbit/sec links. (See "New X40 MSA Defines Hot-pluggable, Multi-rate 40-Gbit/sec Optical Transceiver".)

Unlike the SSR-40, which espouses a serial approach, the X40 MSA focuses on a parallel or inverse multiplexing technique in which the 40-Gbit/sec signal actually is carried on four channels of 10 Gbits/sec.

"One of the problems with inverse multiplexing in general is that 1) you end up burning four channels on a fiber to carry [the traffic] across the fiber, and 2) you end up placing the full burden of the 40-Gbit cost on the network side," explains Neil Salisbury, executive vice president of North American Operations for Kailight Photonics (search for Kailight Photonics).

The SSR-40 contends that an inverse multiplexing approach actually increases the cost of 40-Gbit/sec transmission in the long run because many of the same components are currently used in both short-reach (i.e., network side) and long-reach (i.e., client side) transponders. "If you say, 'Okay, for network-side interfaces, we're going to use 40-Gbit/sec components, but for the client side, we're only going to use 10-Gbit/sec components,' what you end up doing is spreading a much larger burden of the cost across just the network-side applications," says Salisbury.

The X40 MSA's inverse multiplexing approach takes advantage of 10-Gbit/sec transponder volumes to lower the cost of 40-Gbit/sec transmission and thereby ignite additional 40G deployments. But the SSR-40 Working Group believes that the use of 10-Gbit/sec components for client-side applications ultimately will inflate the cost of 40-Gbit/sec components for network-side applications and could effectively delay the widespread rollout of 40G.

The SSR-40 aims to develop a serial short-reach interface that is more economical than the parallel or inverse multiplexing approach—with or without 40G volume shipments, says Salisbury.

"Based on what we hear from carriers and system vendors, we will get there by volume," he contends. "But the trick is, are there other things we can do besides just volume? Are there things we can do in the design, the tolerance, the yield ratios?" he muses.

The SSR-40 Working Group, which meets for the first time today, has challenged itself to answer this question.


Visit The SSR-40 Working Group