PON has been a topic at OFC in recent years, but for 2005 it has the chance to steal the show, as every optical components vendor seems to want to show off PON prowess. Part of the motivation is that the market is finally expanding beyond Japan, as North American carriers pledge to build out FTTP networks (see Tracking Verizon's FTTP Progress and SBC Sheds Light on 'Lightspeed').
More important is the rise of integrated optics. Startups that have been hammering away at planar lightwave circuits (PLC) -- chip-based optics -- see PON as their chance to finally break into a sizable market. One big ticket here is the diplexer, an all-in-one transponder that includes the two wavelengths, 1310nm upstream and 1490nm downstream, required for PON. And some networks will use a 1550nm wavelength for a cable TV overlay, creating the need for triplexers.
The PLC approach can win out here because the parts are smaller and less expensive. PLCs aim to do for photonics what integrated circuits did for electronics: squash bulky groups of circuitry into a space that's small, flat, and easy to replicate in mass quantities (like a Fox and Jacobs home).
The fact that PLCs can be inexpensive is particularly important given that cost has been a barrier to past PON deployments. "That's where the issue is -- cost," says Ed Cornejo, director of product marketing for Opnext Inc. "There have been parts out there for years, but they cost so much."
Some of the pre-OFC PON pronouncements have included:
- NeoPhotonics Advances FTTP
- Eblana Intros Laser Diodes
- BinOptics to Unveil PON Chip
- Metrophotonics Samples Triplexer
- Intec Signs $15M Billing Contract
Given the number of companies interested, it would seem that diplexer prices are headed straight for the ground. But PON has a saving grace: The components require some tricky engineering. "As I understand it, it's pretty hard to do well," says Tom Hausken, optical components analyst with Strategies Unlimited.
Some would find that counterintuitive, given that PONs sit on the access network, not the core, and carry slower speeds, with 1 Gbit/s only now coming into play. But PONs require a lot of analog engineering, particularly in the receiver electronics when it comes to adding analog cable TV to the mix. And analog electronics require a notorious amount of tweaking in order to deliver peak performance.
"There actually is an opportunity there if you can outperform some of the others," Hausken says.
The PON world doesn't stop at triplexers. Another play here is the splitter, the intermediate waystation that divides a wavelength into multiple feeds, usually 32 or less. Traditionally, this has been done with an array of 1x2 elements cascaded to get the right number of splits. But with PLCs, all those devices can now be collapsed onto one chip, says Chris Pfistner, director of global access business for NeoPhotonics Corp.
"What used to be in a 1-rack-unit, 19-inch-wide box now can be in a device about the size of your index finger," Pfistner says.
Work continues on the single-wavelength transponder side, too. Emcore Corp. (Nasdaq: EMKR) is adapting its cable-TV components to the PON market, and Fiberxon Inc. has announced Gigabit Ethernet PON subsystems. Israeli startups are getting into the mix as well: Colorchip will show off PON transponders at OFC, and BroadLight Inc. plans to announce PON transponders on Monday (see Emcore Intros New Components and Fiberxon Unveils GE-PON Subsystems).
— Craig Matsumoto, Senior Editor, Light Reading