Polarization mode dispersion compensation of chirped Bragg gratings used as chromatic dispersion compensators

Compensation of Polarization Mode Dispersion (PMD) induced by a chirped Bragg grating is performed using a short length of highly birefringent fiber. A recirculating loop experiment demonstrates complete PMD compensation of the chirped Bragg grating. This simple method can be extended to deterministic PMD compensation of any components. Introduction: Pulse broadening induced by PMD is now an important limitation in long-haul optical communication systems operating at high data rate. For this reason, PMD of optical fiber links, including its optical components, has recently become a major concern. Fiber optic components of short lengths display deterministic PMD resulting in a propagation delay between the two Principal States of Polarization (PSP). A Chirped Bragg Grating (CBG), used as a chromatic dispersion compensator, is such a component exhibiting high deterministic PMD [1,2]. In a CBG, the fiber birefringence combined to the chirp of the grating causes the two PSP to be reflected at different positions along the fiber grating. The resulting PMD thus induced by a CBG is usually as high, or even higher, than the PMD of the associated fiber link. A high spectral resolution measurement of CBG induced PMD displays oscillations that result from the presence of ripples on the group delay curve. However, the mean PMD is mostly constant over the CBG bandwidth. This deterministic PMD can easily be compensated using another component of opposite deterministic PMD. In this paper, we present mean PMD compensation of a CGB using a short length of Highly Birefringent (HiBi) fiber. To our knowledge, these results are the first demonstration of PMD compensation of a CBG. PMD compensator: The chromatic dispersion compensator is a CBG exhibiting a delay curve with a slope of -1273 ps/nm over the 0.15 nm wavelength span of interest. The CBG can therefore compensate the chromatic dispersion induced by nearly 73 km of standard single mode fiber. The standard deviation of the delay curve ripples is 21 ps. The PMD of the reflected signal from the CBG is measured with the Poincaré arc method [3] and a mean delay of 3.7±0.3 ps between the PSP is M. Rochette et al., “Polarization mode dispersion compensation... ”, Electron. Lett., vol. 36(4), pp. 342 (2000). 2 obtained. The PMD compensator is made of a short length (194 cm) of HiBi fiber (B = 5.68e-4) providing the same polarization delay as the CBG. The HiBi fiber is inserted after the CBG with a λ/2 waveplate beetween the two components (fig. 1). The λ/2 waveplate is adjusted to couple the PSP of the CBG to the HiBi fiber. The PMD compensator could also be placed before the dispersion compensator or even between the CBG and port 2 of the optical circulator in order to reduce by two the required HiBi fiber length. It is also possible, without using a λ/2 waveplate, to directly splice the CGB and the HiBi fiber by taking care of aligning the opposite PSP. The residual PMD of the compensated CBG is below the measurement resolution of 0.3 ps. The CBG and its PMD compensator (λ/2 waveplate + HiBi fiber) are inserted in the mid-stage of a dual-stage Erbium-Doped Fiber Amplifier (EDFA) [4]. The performance of the PMD compensation is evaluated in a recirculating loop experiment. Recirculating loop experiment: The recirculating loop, illustrated in fig. 2, is made of 75.8 Km of standard single mode fiber. A gain switched DFB laser provides 70 ps pulses and is modulated at a rate of 1.2 Gb/s. The output wavelength is finely tuned to the smoothest portion of the CBG delay curve (~1550.9 nm). The pulses propagating in the recirculating loop are controlled via acousto-optic switches. Pulses are inserted in the loop and extracted from the loop using a 3 dB coupler. Two polarization controllers, one placed after the DFB laser and the other placed in the recirculating loop, insure that the polarization state of the pulses at any given position in the loop remains the same at every revolution. Furthermore, the polarization controllers are adjusted to equally excite the two PSP of the CBG to highlight the effect of PMD on the pulses. At the loop output, the signal is first analyzed with a polarizer to monitor the evolution of the polarization state from turn to turn. This allows a fine adjustment of the polarization controller inside the loop. Afterwards, the loop output is set to time-gate the pulseshapes exiting from the loop after a chosen number of revolutions, using an electro-optical switch. This optical signal is sent to a 25 GHz photodetector then providing the RF signal to a 40 GHz oscilloscope. Acquisition of one pulseshape implies an average of 32 sampled pulses by the oscilloscope. The data points represent the average of at least 4 measurements. The pulsewidth measurement error is ±0.3 ps. M. Rochette et al., “Polarization mode dispersion compensation... ”, Electron. Lett., vol. 36(4), pp. 342 (2000). 3 Experimental results: The pulsewidth evolution as a function of distance is presented on fig. 3. Without the chromatic dispersion compensator, the pulsewidth is 450 ps after one revolution through the 75.8 Km loop (not illustrated). Inserting a CBG compensates for chromatic dispersion at the expense of additional PMD. After 25 revolutions in the loop (1895 km), the mean pulsewidth is 283 ps. Once the PMD compensator is inserted, the mean pulsewidth is reduced to 190 ps indicating a cumulative PMD compensation of 93 ps. When the polarization is controlled in the loop, the expected polarization delay induced by the grating after N revolutions simply adds as τ=3.7*N ps, where 3.7 ps is the delay between the PSP induced by one reflection from the grating. In fig. 3, the expected pulsewidths with PMD compensation is calculated by taking the measured uncompensated pulsewidths minus the expected polarization delay. We observe that the experimental pulsewidth compression matches well the expected value calculated from the polarization delay of the grating. These results confirm complete PMD compensation of the CBG. The PMD of the fiber loop itself (<τloop>=0.36 ps), which has a random influence on the pulsewidth measurements, is accounted for in the measurement standard deviation illustrated in fig. 3 as error bars around the mean values. Even with PMD compensation, the pulses broaden as propagation distance increases. This results from residual chromatic dispersion since the delay slope of the dispersion compensator does not perfectly match the dispersion of the fiber loop. Conclusion: We have demonstrated a simple and efficient way to compensate the average PMD induced by a CBG by using a short length of HiBi fiber. Complete PMD compensation of the CGB was demonstrated after 25 successive reflections corresponding to a propagation distance of 1895 km in a recirculating loop. The same technique could be extended to compensate the deterministic PMD of any optical component. To our knowledge, this experiment is the first demonstration of a PMD free chromatic dispersion compensation using CBG. M. Rochette, S. LaRochelle and P.Y. Cortes (COPL, Université Laval, Québec, Canada, G1K 7P4) M. Guy, J. Lauzon (INO, 2740 Einstein, Sainte-Foy, Québec, Canada, G1P 4S4)