OSAs migrate from the lab to the field to the manufacturing floor
BY MEGHAN FULLER
For roughly 15 years, optical spectrum analyzers (OSAs) had been used for niche applications-mostly in laboratory settings-until the advent of DWDM made them a widely used, general-purpose instrument. Today, OSAs are "the lynchpin of the whole of fiber-optic testing," claims Bal Ganjoo, senior industry analyst at Frost & Sullivan (San Antonio, TX). According to Ganjoo, the worldwide market for OSAs, which totaled $215 million in 2000, will jump to more than $1 billion by 2007, driven by increased deployments in the field and on the manufacturing floor as well as by the need to replace older devices with more specialized, higher-performance instruments.
An OSA measures light power or intensity as a function of optical wavelengths or frequencies. It scans a given wavelength range from left to right and displays the power levels within discrete wavelength bands, explains Wolfgang Schmid, director of fiber-optic and strategic marketing at Acterna (Germantown, MD). "Suppose we are starting out at 1500 nm and we are ending at 1600 nm," he says. "The OSA takes a power sample at 1500 and displays it on a graph. Then it moves a tenth of a nanometer to the right and displays the next power level and so on."
With the addition of a broadband light source, OSAs can also be used to measure both passive and active devices, including erbium-doped fiber amplifiers, Raman amplifiers, and optical filters. Its key function, though, is to characterize wavelengths in a DWDM system. Recent technological advances in DWDM systems-including the addition of more wavelengths-have increased the performance requirements of OSAs.
To increase capacity on existing networks, carriers have begun to add more wavelengths to their systems, which in turn tightens the space between each wavelength. According to Matt Paoni, OSA product manager at Nettest (Utica, NY), 12.5-GHz or 0.01-nm channel spacing has been accomplished in the lab. Most DWDM systems today are deployed at 50 GHz or 0.4 nm. "You can quadruple the number of channels you put on a system, which is just mind boggling," he says.
As systems become denser, spectral resolution is increasingly more critical. For example, says Mark Albert, product marketing manager for optical parametric testing at Tektronix Inc. (Beaverton, OR), "if you want to look at an optical filter for those WDM signals, you need to look at it in much more detail as the channels get closer together." In the past, typical spectral resolution had been 100 to 60 pm, he explains, but today's OSAs are achieving resolution bandwidth of 50 to 10 pm, rendering OSAs made just two or three years ago almost obsolete.
The OSA's dynamic range has also improved as channels have become more numerous. Most systems today are deployed in the C-band, or the 1520-1570-nm range, but there has been increased interest in both the L-band and the S-band. According to Paoni, the recent development of special fiber that eliminates the water peak has enabled metro DWDM links in the S-band or the 1460-1520-nm range.
OSAs are also becoming more specialized; one size no longer fits all, claims Bita Nosratieh, product marketing manager for the OSA line at Agilent Technologies Inc. (Palo Alto, CA). "While it's true that OSAs are general purpose [instruments], many different customers now use them-component manufacturers, system manufacturers, and even service providers," she says. "You no longer can offer the same product to everybody and meet their needs."
OSAs are now being tailored for specific applications. "For the metro, relative measurement is important, so the wavelength linearity of the OSA is critical," claims Graham Sperrin, product marketing manager at Anritsu Co. (Richardson, TX). "For the core network, what is more critical is the absolute wavelength. So we're finding in specific markets, certain OSAs have been very much tailored. One for the metro, one for the core, and another one-a general-purpose one-for the lab where one minute they could be looking at the C-band, and then the next minute they could be looking at a Raman pump laser 100 nm outside the C-band."
When DWDM migrated from the lab to the field, OSAs had to follow suit, claims Ganjoo. Packing lab-grade performance into a lightweight, portable device suitable for the field proved to be a challenge for test equipment vendors; however, today's field OSA is now "nearly as good as a lab OSA," claims Acterna's Schmid.
According to André Dusablon, product manager of OSAs at EXFO Inc. (Quebec City), another trend that "will accelerate in the coming years because of the need to lower the cost of optical components is that the OSA will be more targeted to manufacturing applications than lab applications."
Within the manufacturing test field, there are two major trends: integration and automation. While component manufacturers must take traditional OSA measurements like wavelength versus signal power and optical signal-to-noise ratio (OSNR), they must also look at polarization-mode dispersion, chromatic dispersion, and other parameters, which requires using several different test boxes. One solution to this problem is to create a test rack in which different modules are coupled to a central computer. However, that is not the ideal scenario, claims Agilent's Nosratieh, because such test racks are costly and unwieldy, creating congestion in the workstation.
"I think the expectation is going to be for the OSA itself to do more than just power versus wavelength," she says. The industry is already heading in this direction. Agilent recently unveiled an OSA capable of taking time resolve chirp measurements, for example, and Anritsu has just released OSAs in the form of a DWDM tester and WDM network tester, both of which do more than the typical OSA.
"Another parallel trend is that these OSAs will also increasingly become incorporated into overall bigger test systems, which will be easier to manage and probably smaller in footprint," asserts Ganjoo. "They will be one automated system that does many functions, so manufacturers will not have to buy OSAs separately, or power meters separately, or tunable lasers separately."
When asked about what kind of impact these systems will have on the standalone OSA market, Ganjoo asserts that vendors will be able to migrate their sales of high-performance OSAs very easily into these automated test systems.
Automated OSAs are also being developed to help manufacturers lower costs and reduce time-to-market. While the semiconductor manufacturing industry is almost completely automated, the optical component manufacturing process is "surprisingly manually oriented," claims Ganjoo. He believes that OSAs will become increasingly more automated and more intelligent in the future.
"Rather than taking a measurement and sending that data to a central computer to analyze it, users are demanding, 'Let us have intelligence right there on the instrument itself. Let it tell us right away, hey, this is happening, this is not right.' Taking the data somewhere else, analyzing it, and coming back the next day takes too much time," explains Ganjoo. "There will definitely be a demand for more intelligent OSAs."
Even in a sluggish economy, the market for OSAs remains solid. The big telecommunications companies may not be building new networks, contends Dusablon, but they still need to upgrade their existing infrastructure, and that will create opportunity for OSA manufacturers.
"Several OSAs that we see in the portable instruments will have to be replaced as a result of new systems with tighter channel spacing. So even if the optical market does not expand as much as people would have seen nine months ago, there is still good growth in the market because of replacing and improving the actual systems in the field," he claims.
As switched network deployments become more numerous, dual-port OSAs will be needed to measure wavelengths that are constantly added and dropped. Further down the road, the advent of all-optical systems will push the requirements for OSAs even higher.
"The market is definitely growing," contends Sperrin, "and prices are going to be pushed down-the way OTDRs went from being used to manufacture fiber and then were pushed into the field and then extensively used as smaller units that were deployed. But not at the penalty of performance-that's the tradeoff," he adds.