A Bristol, U.K., semiconductor startup, Phyworks Ltd., has managed to keep a low profile since it was founded in January 2001. That may now be about to change, because the company has reached a significant milestone: tapeout of its first products -- two Physical Layer (PHY) chips that could be used in a wide variety of networking equipment.

Phyworks says that its first products will make a big splash in the market for electrical solutions for handling 10-Gbit/s transmission speeds over existing fiber. "We are addressing the question of how you deploy electronics to make 10-Gbit/s communications less expensive," says Nick Weiner, Phyworks CTO.

"Less expensive" in this case, means four times less expensive. Phyworks forecasts that the average price of 10-Gbit/s transponders will come down from today's $2,000 to $500 in the next "two to three years." So Weiner figures the cost of the integrated circuits within these modules will have to come down from $400 to less than $100. That's the cost-level Phyworks is targeting.

Phyworks is pursuing two approaches to cutting costs, according to Weiner, who says these will be built into separate chips for now but could be integrated in the future. The first uses electronic dispersion compensation, or equalization -- using electronic wizardry to correct for impairments in the optical signal such as Chromatic Dispersion and Polarization Mode Dispersion (PMD). The other implements Forward Error Correction (FEC) -- adding extra bits to the signal that make it possible to check and remove errors at the receiver.

Both techniques are usually seen as methods of increasing the transmission distance in optical links, says Weiner. But the other way of looking at them is as a means of reducing the cost of the transmitter and receiver optics, because it becomes possible to use cheaper components designed for shorter reaches. Phyworks sees short-reach applications as the main market at present, although it contends that its techniques can also be applied to the long haul.

As noted, Phyworks' dispersion-compensating chip and FEC chip recently achieved tapeout; in other words, the designs can be sent to a foundry for manufacturing. Sampling will start in early 2003, the company says. The product roadmap then calls for these chips to be integrated with other functions, such as the physical layer or framers, depending on customer requirements.

Another startup, Big Bear Networks, has made quite a song and dance about using electronic dispersion to overcome fiber impairments (see Big Bear Promises Picnic). However, Phyworks views heavyweight chip vendors like Applied Micro Circuits Corp. (AMCC) (Nasdaq: AMCC) and Broadcom Corp. (Nasdaq: BRCM) as its main competitors. AMCC, for instance, introduced a family of PHY chips with integrated dispersion compensation earlier this year (see AMCC Premieres SuperPHY Family).

Wiener claims that Phyworks will have "the lowest power, smallest size, and comparable or better performance" than competing solutions. However, since the products aren't yet ready, and no product specifications are available, it's difficult to corroborate these claims right now. And it will be tough competing with solutions that reached the market much earlier.

One area where the startup is offering a few details is in the performance of the FEC. Phyworks will offer default G.709 FEC, says Weiner, as well as a proprietary FEC, which has about 50 percent more gain but still runs at the same line speed as the standard version.

Running at the same line speed is important, says Weiner, because it makes the FEC chip compatible with existing commercial line-card chips. But it also makes for an even bigger boost in performance. Many other advanced FEC schemes that claim to give higher performance also run at higher line rates, so some of the gain from the FEC has to offset the increased dispersion that comes from using a faster optical signal, reducing the overall advantage.

Phyworks was born out of Microcosm Communications Ltd., another U.K. startup developing communications chips for optical networks, which was acquired by Conexant Systems Inc. (Nasdaq: CNXT) in January 2000 for $160 million.

Once Microcosm had been absorbed into Conexant, some of the executives left to set up Phyworks with $1.25 million of their own money. This was used for market research, developing a business plan, and assembling the beginnings of a design team, together with CAD tools. Gary Steel, founder and former CEO of Microcosm, sits on the Phyworks board.

The company closed a $10.1 million first round of funding in February of this year, with three venture capital companies -- Add Partners, Atlas Venture, and Prelude Ventures.

— Pauline Rigby, Senior Editor, and Peter Heywood, Founding Editor,
Light Reading www.lightreading.com