Modern fiber optic telecommunications systems employ wavelength division multiplexing (WDM), in which multiple wavelengths of light carry information through the optical fibers. Although nearly all deployed systems use multiple transmitting lasers, one for each wavelength, researchers worldwide are exploring ways to obtain multiple wavelengths from a single laser. Figure 1. The transmission of the intracavity modulator is different for different wavelengths. T is the transit time for one circuit around the ring resonator.One approach to this challenge is to design a mode-locked laser that produces simultaneous pulse trains at different wavelengths. Several such lasers have been designed and operated in the past quarter-century, but a collaboration of researchers from Centre National de la Recher-che Scientifique institutes in BesanÇon and Metz in France has proposed a ring fiber laser that produces simultaneous mode-locked pulse trains at each of three wavelengths in the C-band.Figure 2. When the modulator is placed inside the laser resonator (above), it produces an output pulse train consisting of alternating pulses at l1 and l2 (below).The concept of simultaneous mode-locked pulse trains, first demonstrated in 1979, is relatively straightforward. It requires an intracavity modulator whose transmission depends on wavelength (Figure 1). When this modulator is placed inside a resonator whose gain bandwidth includes both l1 and l2 from Figure 1, it generates two simultaneous, interleaved pulse trains, one at each wavelength (Figure 2).Figure 3. The experimental arrangement used an unbalanced Mach-Zehnder interferometer as the modulator and a commercial erbium-doped fiber amplifier as the gain medium.In a 1982 demonstration of the concept, researchers then at Cornell University in Ithaca, N.Y., used a simple phase modulator inside a diode-laser resonator to generate two interleaved pulse trains at two wavelengths. One pulse train passed through the phase modulator at a maximum value of phase modulation (when the frequency shift was zero), and the other passed through at a minimum of phase modulation (when the frequency shift was again zero). Mode-locked pulses cannot pass through a phase modulator between the minimum and maximum of phase modulation because they would be frequency-shifted out of resonance with the resonator's longitudinal modes. Figure 4. This plot shows the calculated peak transmission wavelength of the unbalanced Mach-Zehnder as a function of time (a). The researchers hoped to observe interleaved pulse trains at different wavelengths, each passing through the interferometer during its peak transmission at that wavelength (b).The researchers in France employed a more sophisticated modulation approach to generate three simultaneous, interleaved pulse trains at three wavelengths. They used an unbalanced Mach-Zehnder interferometer as a loss modulator to mode-lock the laser (Figure 3). One arm of the interferometer was 40 µm longer than the other, and applying an electric field to the electro-optic crystal in the arms could change the refractive index of both. They calculated that the peak transmission wavelength of the interferometer would vary with time when an appropriate electric field was applied to the arms (Figure 4a). When they placed the interferometer inside the resonator, they hoped to see several interleaved pulse trains, each passing through the interferometer when its transmission was the maximum for that wavelength (Figure 4b).Figure 5. The laser produced three interleaved pulse trains at three wavelengths: approximately 1543, 1550 and 1563 nm.But in the laboratory, factors such as the wavelength dependence of the erbium-doped fiber amplifier's (EDFA) gain and competition among longitudinal modes for the homogeneously broadened gain introduced complexity. Nonetheless, the scientists observed interleaved pulse trains at different wavelengths (Figure 5), although the pulse train timing was not entirely consistent with their calculations. They improved their timing of the pulses by adding a conventional loss modulator to the resonator.In the future, they plan to focus on stabilizing the pulse trains, controlling the wavelengths better and obtaining pulse trains at more than three wavelengths.