By Kevin Hsu and Mark Jones
As a manufacturing tool, a WDM emulator provides a test source that includes low loss, high channel contrast, clean spectral profiles, tunability, grid positioning accuracy, and high channel counts.
Optical–performance testing over multiple wavelengths is an especially important procedure during manufacture. However, multiwavelength testing usually involves expensive arrays of lasers, attenuators, combiners, polarization scramblers, optical amplifiers, and associated equipment that result in a testbed that is typically inefficient in cost, set–up time, maintenance, and sharing flexibility. A low–cost and compact multiwavelength source—a WDM emulator WDME)—provides a useful tool for this purpose.
A spectrum–sliced source as a WDM emulator has been proposed and used by various industry groups for optical–testing applications. It is not difficult to imagine the many testing applications made simple by WDMEs, including calibration for optical performance monitors, measurement of noise figures and gain flatness in optical amplifiers, performance testing for DWDM multiplexers and demultiplexers, and operational tests for optical crossconnects, reconfigurable add/drops, and switches.
In its simplest form, a WDM emulator consists of a broadband input source followed by a multiresonant optical filter such that the output is composed of periodic narrowband emissions. In the interest of testing applications, the WDME should emulate a DWDM communication spectrum as closely as possible.
A Fabry–Perot filter best represents the essential multiresonant filter element because it can provide a comb spectrum that spans across the input broadband source spectrum. For DWDM testing, it is important that not only the channel spacing be narrow—ranging from 100, 50, to <25 GHz—but the channel contrast must also be high (>35 dB).
A Fabry–Perot filter of finesse (F) 100 can provide a contrast factor of 36 dB. However, at a predetermined DWDM channel spacing, the resultant 3–dB bandwidth could be much too narrow (1 GHz) to emulate a typical DWDM spectrum. Thus, a different approach is needed to provide sufficient contrast factor and yet not compromise too much in bandwidth.
A simple and elegant approach is to cascade two "identical" Fabry–Perot filters to double the contrast factor of a single–stage Fabry–Perot filter, while only reducing the 3–dB bandwidth by a ratio of ~1.56 (see Fig. 1). As an example, a single–stage Fabry–Perot filter with a free spectral range (FSR) of 50 GHz and finesse ~40 has a contrast factor (CF) of ~27 dB and a bandwidth of ~1.25 GHz. By cascading two such Fabry–Perot filters, the resultant dual–stage contrast factor is ~54 dB and bandwidth is ~0.8 GHz.
A critical design aspect arising from dual–stage Fabry–Perot filters is how well the transmission resonance can be matched over a broad spectrum, as well as alignment accuracy to the ITU frequency grid. These requirements impose a very high demand on Fabry–Perot filter fabrication and control. Overall, there are five essential technical criteria that determine the quality of a WDM emulator: low loss, high contrast, dual–stage frequency–matching tolerance, ITU frequency–grid positioning accuracy, and power.
A fiber Fabry–Perot technology platform can provide fiber Fabry–Perot interferometers (FFPI) of extremely low–loss and near–theoretical transmission profiles. Moreover, very accurate and precise fabrication processes can provide free spectral ranges corresponding to the ITU frequency grid, thus leading to highly matched dual–stage fiber Fabry– Perot interferometer resonance.
With an amplified spontaneous emission source as an input, the WDM emulator output is a stable, accurately spaced comb signal that can be adjusted and modified to suit the application. The comb can be shifted up or down by up to 100 GHz using temperature control. The inputs and outputs of the fiber Fabry–Perot interferometer can be connected alone, in series, or in parallel to provide a wide array of test configurations from basic to complex and versatile (see Fig. 2).
Well–matched fiber Fabry–Perot interferometers are critical to ensuring high spectral uniformity across broad wavelength range. Consider a free spectral range of 50 GHz at 1550–nm, in which the order number is ~3780. During optical fabrication, not only must the cavity resonance be monitored to reach a certain tolerance of free spectral range, but the order difference between the dual–stage interferometers must be within one order to ensure sufficient spectral uniformity. For FSR = 25 to 100 GHz, one resonance order difference translates to a cavity length mismatch by ~0.53 μm.
The final resonance alignment is accomplished by thermoelectric control of the individual interferometer. Order mismatch will cause power undulation in the emulator spectrum, and reduce the contrast factor between channels. For example, a one–order resonance mismatch can cause ~4 dB of power undulation across 60 nm.
Aside from order mismatch, the WDME spectral power flatness depends on the input broadband source profile. The fiber Fabry–Perot interferometers provide uniform spectral peaks across the entire C–band and the C+L–bands (see Fig. 3).
To enhance the emulator output power, a straightforward approach is to use a post–optical amplifier. However, this approach can significantly degrade the contrast factor. An intrastage amplifier design is the preferred WDME configuration specifically geared toward higher output power with high contrast factor (see Fig. 4).
FIGURE 4. In separate tests, a WDME displays a different spectral response for different component configurations of a dual–stage FFPI (FSR = 50 GHz and CF 47 dB). Blue trace: peak power of ~ –30 dBm using a single ASE input. Red trace: amplified output spectrum from a post–EDFA, resulting in a peak power of ~ –4 dBm and much reduced contrast factor. Green trace: amplified output spectrum by using an EDFA between the dual–stage FFPI, resulting in a peak power of ~ –10 dBm and a high CF ~50 dB.
AWDM emulator can be used in numerous applications in manufacturing or system design (see Fig. 5).
Optical–performance monitors: calibration and testing. Optical–performance monitors (OPMs) play a critical role in monitoring and managing the operational integrity of DWDM systems. To provide accurate measurements in channel power, wavelength, and optical signal–to–noise ratio, OPMs must be properly calibrated and tested through the manufacturing process. Because of the enormous range of parameter space (a combination of power, wavelength, and environmental conditions), testing by a step–wise tunable laser can be very time–consuming. Testing with multiple input channels is particularly essential to understand the measurement ability of OPMs under a loaded DWDM situation.
A WDM emulator can dramatically enhance testing capability and reduce testing time by simultaneously presenting an array of calibrated wavelengths across the spectral range of interest. WDM emulators are currently used to test OPMs constructed from scanning–filter or grating/diode–array technologies. Of particular interest is the fact that the ITU frequency–grid emission from the WDME can be used for optical alignment and algorithm verification of the grating/diode–array pair, thus greatly enhancing the fabrication and characterization capability.
Optical amplifiers: gain flatness, power equalization, and noise figures. Gain flattening and channel power equalization through optical amplifier chains are critical to DWDM system performance. These performance parameters should also be tested under a realistic condition where multiple channels are present. The WDM emulator, with its depolarized multichannel outputs, is an effective and low–cost source for rapid measurement in this application. Similarly, it is also of great interest to measure noise figures in optical amplifiers with DWDM input.
A typical setup for the input source would consist of an array of distributed–feedback lasers combined through a chain of multiplexers, polarization scramblers, and amplifiers. A much more economical WDM emulator, having inherently depolarized multichannel outputs and high contrast factor, would serve particularly well for noise–figure measurement using the amplified spontaneous emission interpolation technique.
Multiplexers and demultiplexers: channel crosstalk measurements. By injecting a multiwavelength input to a mux/demux, the multiplexing properties such as crosstalk, transmission profile, and insertion loss can be characterized by an optical switch and an optical spectrum analyzer (OSA). Because the signal spectral profile from the WDM emulator follows a clear theoretical shape, the actual transfer function of the mux/demux can be reliably extracted from the OSA measurement.
Optical crossconnects, reconfigurable add/drop modules, and optical switches: operational verification. Optical crossconnects, reconfigurable optical add/drop modules, and optical switches, as the critical building blocks for future dynamic optical networks, are widely pursued by the communications industry. Characterization of these devices can pose serious challenges in test equipment availability, allocation, testing time, and cost, particularly when port counts evolve beyond 16 × 16 to 64 × 64 or more. This challenge can be eased by using WDM emulators as multiple–channel test inputs, starting by using one WDME shared by an optical switch, to using a full set of WDMEs each modulated by a unique data stream. In this way, each wavelength at each output fiber can be traced to the originating emulator and thus the input fiber.
Kevin Hsuis vice president of R&D and Mark Jones is a research engineer at Micron Optics, 1852 Century Place NE, Atlanta, GA 30345. Kevin Hsu can be reached at email@example.com.