SOAs still awaiting big chance

Th Cip7p2

In March 2004, Lightwave ran an article in this space entitled “SOA Debate Rages On.” The lead paragraph concluded with the following assertion: “While SOAs [semiconductor optical amplifiers] are now commercially available from several companies, some industry insiders believe they have not yet made the leap from lab curiosity to network reality.” Despite the recent emergence of what could be several viable applications for the technology, SOAs still have not made their much anticipated migration from the lab to the field.

Alistair Poustie, systems and applications engineer at the U.K.-based Centre for Integrated Photonics (, believes the barrier to SOA deployment is two-fold. First, he says, there is an industry-wide misconception of the performance parameters that SOAs can achieve, and second, there has been a lack of real applications that leverage the strengths of the SOA.

Several years ago, state-of-the-art SOAs featured a noise figure of around 8 or 9 dB, a far cry from the 5-dB noise figure of the erbium-doped fiber amplifier (EDFA) with which the SOA competes in certain applications. CIP says a number of its SOAs have noise figures below 7 dB, inching ever closer to the performance parameters of bottom-end EDFAs.

As for the second hurdle slowing SOA momentum-a lack of applications that leverage the technology’s strengths-that too is changing, contends Poustie. “One of the big advantages of SOAs is that you can gate them on and off very fast,” he says, “and these sorts of systems are now starting to emerge. Even in access networks, people are looking at burst-mode-based systems, so that’s something that’s of interest.”

One of the biggest market opportunities may well be in the WDM-PON space, which has generated interest from carriers worldwide and actual field trials in the Far East, most notably by Korea Telecom. A WDM-PON architecture delivers a single wavelength, both upstream and downstream, to each residence and/or business, providing far greater bandwidth than current BPON/EPON/GPON deployments.

While a WDM-PON implementation is technically feasible today using distributed feedback (DFB) lasers or narrowband tunable lasers, it is cost prohibitive to put a WDM laser in every residential optical network terminal (ONT). As such, system vendors have begun to explore less expensive alternatives, one of which may be the reflective SOA (RSOA).

Traditional SOAs are two-fiber devices designed to have very low reflection at each facet, explains Ian Lealman, SOA product manager at CIP. They are considered traveling wave devices. Reflective SOAs, by contrast, feature a curved waveguide to provide very low reflection from the front facet and fairly high reflection from the back facet. “What you actually need [in an RSOA] is a chip with very low front-facet reflectivity, but a cleaved or high reflecting rear facet,” he says. “It tends to be more a variation in device geometry, whereas a normal SOA would be an angled stripe device.”

The RSOA in the ONT receives a reference wavelength from the OLT, remodulates the wavelength, amplifies it, and sends it back to the OLT. The real advantage of the RSOA is that it is essentially colorless, enabling it to operate at any wavelength.

Fujitsu Network Communications ( has announced its intention to develop a WDM-PON system with SOA-based ONTs. This architecture is more feasible for mass deployment, says market development manager Randy Eisenach, because you would no longer need dedicated lasers in each ONT. The ONT “simply receives the signal, retransmits it, and bounces it back toward the OLT,” he explains. “We think there are a couple of technologies required to do that cost effectively. You do have to amplify the signal, so you want a semiconductor optical amplifier. We have the technology based on quantum dots that we think will get us to the semiconductor optical amplifier at a very low cost point such that we can do it uncooled,” he adds.

While he admits that WDM-PON will never attain price parity with a BPON or GPON ONT, an SOA-based architecture “shrinks that gap tremendously,” Eisenach reports.

At press time, Korean broadband equipment provider Corecess ( introduced a commercially available RSOA-based system, which the company is touting as “a next-generation DWDM-PON system.” Its RSOAs, developed by the Electronics and Telecommunications Research Institute (ETRI), are implemented in SFP and GBIC form factors. “Colorless operation is required to make DWDM work for the access,” noted Yeong-bong Son, Corecess’s executive vice president of sales and marketing, in a press release announcing the product. “ETRI’s knowledge and technology of advanced optics inspired Corecess to implement this new type of gigabit DWDM-PON.”

While the only real implementations to date have occurred in Korea, the folks at CIP, which has developed an RSOA optimized for WDM-PON applications, report interest from both Europe and North America, at least on an R&D level. The ratification of some sort of WDM-PON standard would further the development of the technology, notes Lealman.Th Cip7p2

CIP’s monolithic indium phosphide-based RSOA features a curved waveguide architecture that results in ultralow front-facet reflectivity, making the device ideal for use as a colorless modulator in WDM-PON systems.

Meanwhile, nonlinear SOAs, which fall into the traveling wave camp, also are experiencing a slight uptick in interest, thanks to the emergence of 40-Gbit/sec optical switching applications, such as regeneration and wavelength conversion, say the folks at CIP. The vendor’s nonlinear device has an extremely fast gain recovery that the company believes makes it ideal for use in high-speed applications, and its low reflectivity enables the device to be passively aligned in an integrated optical subsystem. Though Lealman is quick to note that the use of nonlinear SOAs at 40G is “still mainly laboratory based,” he also adds that, “things are starting to happen. People are starting to think about field trials and moving up the chain of interest in that type of application.”

Of course, heightened interest in using SOAs for these more forward-looking applications does not necessarily mean widespread deployment is imminent. As Poustie notes, SOAs are not a one-size-fits-all device, and the industry is still determining which SOA design works best for what application.

As a result of this slow-to-materialize demand for SOA devices, many of the vendors who touted products before the telecom crash have disappeared. Today, only a handful remain, most noticeably CIP; Princeton, NJ-based Alphion; and Glasgow-based Amphotonix Ltd., which emerged with the intellectual property and assets of the now defunct Kamelian Ltd.

In fact, the market selection is so thin, optical module supplier Kailight Photonics ( resorted to building its own SOA because it could not find an existing product on the market that fit the needs of its 40-Gbit/sec Tunable All-optical Signal Regenerator (TASR) product. “For our 2R regenerator, we needed an SOA that was highly nonlinear and had very quick rise and fall times, and the best way of getting that was just building it ourselves,” explains Neil Salisbury, vice president of marketing and business development.

Kailight Photonics is using its homegrown SOA not as an amplifier but as a nonlinear device to perform the cross-phase modulation necessary for all-optical signal regeneration, which Salisbury claims is “the wave of the future. 40G has taken off,” he adds. “People are getting really serious about 40G. There are deployments; it is more than just trials now.”

For those vendors who have long awaited the emergence of applications that would take advantage of the SOA’s unique attributes, this is good news indeed.

Meghan Fuller is senior editor at Lightwave.

More in PON/Systems