Optical components

OneChip Tries Infinera's Trick

Optical startup OneChip Photonics Inc. wants to do for transceivers what Infinera Corp. (Nasdaq: INFN) did for long-haul Wavelength Division Multiplexing (WDM) transport.

That's to say the Ottawa-based startup has compacted the innards of a fiber-to-the-home (FTTH) transceiver onto a single indium phosphide (InP) chip, producing what it claims will be a cheaper product.

That could be a plus given that FTTH parts -- especially for the optical networking terminal (ONT) that gets deployed at customers' homes -- need to be inexpensive.

"They've got the right idea," says Sterling Perrin, optical analyst for Heavy Reading. "If they can do it, they'll hit a couple of sweet spots. Photonic integration is one. And in fiber access, the ONT costs have got to come down."

The idea is catchy enough that OneChip amassed a $16.5 million round of funding, the second tranche of which was completed last week.

The company announced yesterday that it's raised $19.5 million total in two rounds, from investors including BDC Venture Capital , DCM - Doll Capital Management , GrowthWorks WV Funds , and Morgenthaler . (See OneChip Photonics Raises $19.5M.)

By going for monolithic integration, OneChip is trying to outshine the makers of integrated FTTH transceivers. Designs from companies like ColorChip Inc. , Enablence Technologies Inc. (Toronto: ENA), and NeoPhotonics Corp. (NYSE: NPTN) are based on planar lightwave circuits (PLCs), devices that are integrated but don't pack everything onto one chip, sometimes because they're combining elements made of different materials. (See ColorChip Adds Meteor II, OFC: Mergers Pay Off, and NeoPhotonics Unveils GPON Transceivers.)

OneChip, by contrast, says it can pack everything together: laser, PIN diode, waveguides, and so on.

That kind of monolithic integration is the key to Infinera's DTN long-haul system, where Infinera was able to pack more optical channels into a smaller and less power-hungry space. (See Infinera Declares WDM War.)

But can OneChip really use InP integration to make a cheaper transceiver? Infinera's business -- long-haul optical equipment -- involves inherently high-priced equipment. The company hasn't yet produced a system for the lower-cost metro market, although it's thought about it. (See Infinera Surprises, Targets Metro Access.)

"I was wondering if anybody would even go into the metro because of the cost, which is why it's interesting that OneChip is starting at the access," Perrin says.

Moreover, because lasers can't be made of ordinary silicon, any single-chip transceiver has to be built with a more expensive material like InP.

OneChip swears it can get the InP chip down to a price that's lower than other FTTH transceivers. It's taken active alignment out of the assembly process -- meaning its chips can be built, assembled, and tested without manual intervention, a common cost-cutting direction for optical parts. And it's outsourcing its manufacturing to InP foundries rather than building its own fab.

On top of that, CEO Jim Hjartarson says OneChip's chips are built using techniques that uses fewer steps than even some normal InP chips, which keeps the manufacturing cost down. "Our manufacturing process is quite a bit simpler than what Infinera is doing," he says.

OneChip was founded in 2005 by CTO Valery Tolstikhin, who had previously worked on InP PICs for Metrophotonics. (See MetroPhotonics Shuts Down.) It's got a staff of about 30, some of them snapped up during Bookham Inc. (Nasdaq: BKHM; London: BHM) layoffs in Ottawa.

— Craig Matsumoto, West Coast Editor, Light Reading

HomerJ 12/5/2012 | 4:08:45 PM
re: OneChip Tries Infinera's Trick Indeed. General Tao would not be pleased with OneChip's chicken.
deauxfaux 12/5/2012 | 4:08:49 PM
re: OneChip Tries Infinera's Trick

Here we go again. Hjartarson is a systems company CEO that decided that components must be where all the money is in optics. I guess that he forgot to check out the fact that the average margins in the component business are probably in the mid 20's...percentage wise.

Infinera did the smart thing by breaking up its design into a SiGe chip and a InP chip. Why? Because wire bonds are much cheaper than the yield loss on more complex InP devices. The Infinera guys are successful component experts and understand the issues well. Metrophotonics was an expensive disaster that never delivered any production components....but I'm sure it will work better this time <img alt="0" border="0" class="inlineimg" height="25" src="http://www.socalsoccer.com/forum/images/smilies/roflmao.gif" width="50">

GEPON Xcvrs are cheap, plentiful and the active alignments in the assembly process are inexpensive. The street price is as low as $12 and one of the most expensive parts of the whole thing is the pigtail itself. How much sense does an InP TIA make from a cost perspective at 1G.....or even 10G? The answer is none....absolutely none. The SiGe guys have similar costs from 1G to 10 and can drop prices in an instant to defend their share if they need to do so. Triplexers are more expensive, but the video issues are not trivial and OneChip's approach is unlikely to resolve the optical and electrial crosstalk challenges in a meaningful (from a cost perspective) way.

I hear that big flushing sound of cash going down the toilet

derac 12/5/2012 | 4:08:49 PM
re: OneChip Tries Infinera's Trick I agree although the price point for GPON optics isn't that low given the tighter specs of the GPON system.-á Also, the only domestic market for GPON is the VZ network and they need a 3 color transceiver.
Sign In