By STEPHEN HARDY
The challenges of overcoming the increased effects of chromatic dispersion, pulse-mode dispersion, nonlinearities, and other impairments get even tougher for developers of 40-Gbit/sec equipment if they can't measure how they're doing. However, test equipment companies looking to support 40-Gbit/sec design find themselves facing some of the same technology-supply hurdles as their potential customers. As Larry DesJardin, high-bandwidth program manager at Agilent Technologies, puts it, "To build a BERT [bit-error-rate tester] at 40 Gbits/sec requires the very same advanced high-speed technology that will be used in next-generation 40-Gbit/sec systems, sometimes even better. So this remains challenging, because we need the very technology that is being tested, but we need it earlier!"
Since 40 Gbits/sec can, in some ways, be thought of as an extension of previous work in 10-Gbit/sec technology, DesJardins reports that many of the same test-system types-such as BERTs, tunable lasers, optical spectrum analyzers, dispersion analyzers, digital-communication analyzers, jitter test sets, and SONET/SDH testers-are necessary when designing at the new, higher speeds. However, this equipment may need new capabilities. To use bit-error-rate testing again as an example, BERTs have traditionally been electrical nonreturn-to-zero (NRZ) instruments. With the use of return-to-zero (RZ) signaling mentioned here later, such equipment must also handle wavelength-tunable RZ and NRZ optics, DesJardins says.
Also, with dispersion and jitter a greater problem at 40 Gbit/sec, a wider range of components must be tested for these factors than were considered in the previous generation of transmission equipment. New types of test equipment that measure both chromatic and polarization-mode dispersion have been developed to make such tests easier to conduct.
However, the unique aspects of 40-Gbit/sec design also lead to unique test approaches. The use of Raman amplification provides an example. "Raman amplifiers require test equipment throughout the entire component chain to be different: very low PDL [polarization-dependent loss] for each of the passive components within the amp, S-band measurements of the Raman pump sources, and new noise figure measurement techniques of the entire amplifier, which includes the fiber," DesJardins explains. "Also, the Raman pumps are of extremely high power, so new power heads had to be invented to accurately measure the power level."
The use of RZ techniques makes testing even more difficult. "First of all, there is no standard 'eye-diagram' for RZ, so we've had to create RZ analysis ourselves," DesJardins explains. "Also, as if multiplying the bandwidth needs by four times wasn't enough, RZ modulation doubles the bandwidth used again, making 40-Gbit/sec RZ signals require eight times the bandwidth to analyze over 10-Gbit/sec NRZ signals. This pushes measurements to the limit that standard OE [optoelectronic] technology can work at. Also, it is now necessary for test equipment such as BERTs to generate optical RZ signals, which is very complex."
Finally, the frequent use of radio-frequency (RF) signal modulation techniques further complicates the test process. "Many equipment manufacturers are developing exotic modulation techniques such as suppressed carrier RZ and related single sideband transmission formats. Soon there may even be duo-binary formats-four levels," DesJardins relates. "Modulated spectrum analysis is one of the key techniques used in the microwave world for analyzing similar methods and will be essential in the optical domain as well. However, standard grating-based optical spectrum analyzers don't come close to analyzing the spectrum of individual transmissions." As a result, new test equipment that uses microwave techniques in the optical domain, coupled with advanced digital signal analysis, have been developed.
Similarly, optical systems developers must approach jitter testing in a different way when they stray into the RF domain. The digital wireline world has a model of jitter that is measured in unit intervals and frequency, DesJardins says. The microwave world has a similar concept to jitter, known as phase noise. Fortunately, there is a direct mathematical relationship between phase noise and jitter, and Agilent has applied microwave phase noise measurement techniques to measure jitter, as well.
As a result, test equipment is now available to help engineers develop 40-Gbit/sec systems. "I do think that the test equipment is getting better," surmises Dr. Katherine Hall, chief technology officer at 40-Gbit/sec system startup PhotonEx (Maynard, MA). Hall reports that when PhotonEx engineers first began their work, they "borrowed" test strategies from her previous experience with high-speed transmission at MIT Lincoln Labs.
She offers optical pulse characterization as an example of the difficulties of testing within a 40-Gbit/sec environment. One way to characterize an optical pulse is to use a nonlinear technique called "auto-correlation." Commercial auto-correlators can be purchased. In operation, an engineer would split the pulse into two arms, recombine the arms on a nonlinear crystal, and vary the length of one arm. That creates a signal at twice the frequency. In essence, the engineer is mapping something out in time by changing its delay in space. Such techniques are common in spectroscopy but involve significant complexity.
"That's a more difficult thing to measure cleanly electronically, because the response of the electronic equipment has to be fast enough to resolve all the details of the optical pulse," Hall explains. "Usually what people do is that they say, "If my photodiode only has a 30-GHz bandwidth and my scope plug-in has a 40-GHz, 3-dB bandwidth, then I estimate the waveform I'm measuring by deconvolving it, realizing that there are those bandwidth limitations. But the accuracy of that would not be as high as an auto-correlation measurement." However, electronic test systems are now providing results with an accuracy closer to those obtained via auto-correlation, she reports.
In the February 2002 Lightwave news article "Furukawa restructures Lucent's 'crown jewel,' " p. 37, to clarify, the TrueWave fiber product is for long-haul and metro applications, AllWave is for metro and access applications, and LaserWave is for premises applications. Lucent's former Optical Fiber Solutions division developed central core and loose-tube cable structures.