Open Access
Add to BibSonomyAbstract
Experimentally we find a 10 dB input power dynamic range advantage for amplification of phase encoded signals with quantum dot SOA as compared to low-confinement bulk SOA. An analysis of amplitude and phase effects shows that this improvement can be attributed to the lower alpha-factor found in QD SOA.
©2010 Optical Society of America
1. Introduction
SOA have attracted new interest in the last few years due to their ability to amplify signals across the whole spectral range from 1250 nm up to 1600 nm at reasonable costs [1]. A new and interesting question has thereby been the ability of SOA to amplify phase-encoded signals. As a result, it has been shown that the constant envelope of differential phase encoded signals provides higher tolerance towards SOA nonlinear impairments such as cross-gain (XGM) and cross-phase modulation (XPM) compared to on-off keying (OOK) formats [2]. This higher tolerance has its limit once the SOA is operated in saturation where nonlinear impairments reduce the input power dynamic range (IPDR) even for phase encoded signals [3,4]. While quantum dots (QD) as an active medium in SOA have been shown to extend the IPDR for OOK formats, their suitability for differential-phase encoded signals has not been studied up to now. QD SOA offer low alpha-factor [5], ultra-fast QD gain response (~1 ps) [5], greatly expanded gain bandwidth (~120 nm) [6], high gain (>25 dB) [7], large IPDR for OOK signals, and high burst mode tolerance [8].
In this paper, we show for the first time the input power dynamic range improvement for a 28 GBd non-return to zero differential quadrature phase-shift keying (NRZ-DQPSK) signal amplified in a 1.5 µm QD SOA. The IPDR is improved up to 10 dB compared to a low confinement bulk SOA especially designed for amplification. This enhancement found for QD SOA is attributed to the reduced phase error on the differential encoded phase signal, due to the lower alpha-factor. The IPDR of the QD SOA is 20 dB at a bit error ratio of BER = 10−9 and exceeds 32 dB for BER = 10−3.
2. QD and bulk SOA characteristics
A comparison for phase encoded signals of two SOA with different active media requires similar device performance. Figure 1(a) shows comparable fiber-to-fiber gain, noise figure [9, 10] and saturation powers of the 1.5 µm QD SOA device (6 layers of InAs/InP quantum dashes) and the bulk SOA operated with the same current density [11]. The low optical confinement (20%) bulk SOA is especially designed for linear applications. For both devices, the gain peak is around 1530 nm and the 3 dB bandwidth is 60 nm. In all OOK experiments, both devices show the same performance and comparable time constants.
Fig. 1 Comparison of QD and bulk SOA characteristics. (a) Fiber-to-fiber gain, noise figure (NF) and saturation power levels for a 1.5 µm QD SOA (black) and a bulk SOA (blue) with a low optical confinement of 20%. For equal current densities all characteristics are comparable. (b) Peak-to-peak (P2P) phase changes of the bulk SOA compared to the QD SOA as a function of the channel input power. The phase changes are measured as XPM of a 33% RZ-OOK 40 Gbit/s “1010” data sequence at a wavelength of 1557.4 nm on a cw signal at a wavelength of 1554.1 nm. The average input power of the cw (ch. 1) and data (ch. 2) channels is always adjusted to be equal, thereby defining the channel input power. (c) Measured phase changes of bulk SOA versus QD SOA from (b). For all input power levels the phase effect of the QD SOA is less than the phase effect of the bulk SOA by a factor of 0.58.
Figure 1(b) shows the peak-to-peak (P2P) phase changes of the QD and bulk SOA as a function of channel input power. The phase changes are measured as XPM of a 33% RZ-OOK 40 Gbit/s “1010” data sequence at a wavelength of 1557.4 nm on a cw signal at a wavelength of 1554.1 nm. The average input power of the cw (ch. 1) and data (ch. 2) channels is always adjusted to be equal, thereby defining the channel input power. The phase change is measured using the frequency resolved electro-absorption gating technique (FREAG) based on linear spectrograms [12]. In Fig. 1(c) the measured phase changes of the bulk SOA are plotted versus the phase changes of the QD SOA using the results from Fig. 1(b). For all input power levels the phase effect of the QD SOA is less than the phase effect of the bulk SOA. The ratio of the alpha-factors is obtained by a linear fit of the data to αQD/αBulk = 0.58.
3. Experimental setup
The power penalty caused by an SOA for NRZ-DQPSK data is investigated using the experimental setup shown in Fig. 2 . It comprises two data signals at 1554.1 nm (ch. 1) and 1557.4 nm (ch. 2), which are de-correlated by 69 bit. The power levels of both channels are adjusted to be equal before launching them into the device under test (DUT). After amplifying both data signals in the DUT, the 1557.4 nm channel is blocked by a band-pass filter while the BER of the remaining data channel is analyzed. The DQPSK receiver (Rx) consists of a delay interferometer (DI) based demodulator followed by a balanced detector and a bit error ratio tester (BERT). In the experiment, no data encoder circuit was employed. To allow bit error ratio measurements, the error detector is programmed with the expected data sequence, which allows a pseudo-random bit sequence (PRBS) length of up to 29-1.
Fig. 2 Experimental setup. Two 28 GBd NRZ-DQPSK channels are equalized and de-correlated using 0.5 m of fiber. The average power of both channels is varied and launched into a bulk or QD SOA. A single channel is selected, amplified and demodulated in a DQPSK receiver (Rx), consisting of a delay interferometer (DI) followed by a balanced detector. The electrical signal is then analyzed using a digital communications analyzer (DCA) and a bit error ratio tester (BERT).
The IPDR for amplification of one and two 28 GBd NRZ-DQPSK channels is studied by evaluating the power penalty for two selected bit error ratios of BER = 10−9 and BER = 10−3.
4. Dynamic range improvement in QD SOA for PSK signals
We define the input power dynamic range (IPDR) as the range of input power levels with less than 2 dB power penalty compared to the back-to-back case. Figure 3 shows the power penalty as a function of the SOA channel input power for one and two channels for a specific BER. Figure 3(a) and (b) show an IPDR improvement of 5 dB for the single channel and >10 dB for the two channel case for a BER of 10−3, respectively. Figure 3(c) and (d) show around 5 dB IPDR improvement for the QD SOA compared to the bulk SOA for one and two NRZ-DQPSK channels at a BER of 10−9. The filled symbols correspond to the I-channel, whereas the open symbols represent the Q-channel. The QD SOA exhibits a large IPDR of around 20 dB for BER = 10−9 and exceeds 30 dB for BER = 10−3. We attribute the increased IPDR in the two-channel case to the effect of cross-gain modulation (XGM), which reduces the gain for each channel. This leads to larger saturation power levels, less patterning effects and thus smaller power penalties. The results are summarized in Table 1 .
Fig. 3 Power penalty vs. channel input power levels. The input power dynamic range (IPDR) is defined as the range of input power levels with less than 2 dB power penalty compared to the back-to-back case. Red arrows indicate the IPDR enhancement of the QD SOA (black) over the bulk SOA (blue). (a), (b) QD SOA improve the IPDR at a BER of 10−3 by 5 dB and >10 dB compared to bulk SOA for one and two 28 GBd NRZ-DQPSK channels, respectively. (c), (d) For a BER of 10−9, the QD SOA IPDR is enhanced by 5 dB. The filled symbols correspond to the I-channel, whereas the open symbols represent the Q-channel.
Table 1. Input power dynamic range (IPDR) at 2 dB power penalty for bulk and QD SOA. In all cases, the QD SOA shows a significant IPDR enhancement compared to a conventional bulk SOA.
To illustrate this advantage of QD SOA over bulk SOA, Fig. 4 shows the observed eye diagrams of the demodulated 28 GBd NRZ-DQPSK Q-channel for high SOA input power levels of 6 dBm. The comparison of the back-to-back eye diagram in Fig. 4(a) for high input power to the QD SOA eye diagrams in Fig. 4(b) shows no signal degradation at optimum receiver sensitivity. In contrast, for high input power to the bulk SOA, the eye opening in Fig. 4(c) is reduced, and the signal quality is significantly degraded.
Fig. 4 Demodulated NRZ-DQPSK eye diagrams for high SOA input power levels of 6 dBm. (a) Back-to-back eye diagram of 28 GBd Q-channel at optimum receiver sensitivity. (b) No signal degradation for QD SOA high input powers. (c) Reduced eye opening and signal quality degradation for SOA.
5. Dominant sources for impairments of DQPSK signals in SOA
The unexpected IPDR enhancement found in QD SOA needs explanation. As is well known from OOK formats, SOA can introduce strong amplitude distortions like overshoots [13] and patterning effects [14] when operated in saturation. For ideal phase encoded signals, these effects should be strongly suppressed [2].
Figure 5(a) shows all possible transitions in the constellation diagram of DQPSK signals. Instead of pure phase modulation (all states on the circle), most practical implementations generate some transitions (“B”, “C”) using amplitude modulation in order to increase the long-term stability of the transmitter. Figure 5(b) shows an example of the power envelope eye diagram for a 28 GBd NRZ-DQPSK signal with constant power at the decision point in the middle of the bit slot. All expected transitions are observed. A histogram is taken at the transition in a 1.6 ps time window. To investigate the influence of the amplitude effects for both SOA types, the power envelope of the 28 GBd NRZ-DQPSK signal is analyzed as a function of the channel input power. Assuming a Gaussian distribution, mean value and standard deviation are calculated.
Fig. 5 Amplitude distortions of the NRZ-DQPSK envelope in QD and bulk SOA. (a) Possible transitions in the constellation diagram. For a better long-term stability of the transmitter, only some transitions (“B”, “C”) are generated using amplitude modulation, instead of employing pure phase modulation (all states on the circle). (b) Example of the power envelope eye diagram for a 28 GBd NRZ-DQPSK signal, showing all possible transitions. A histogram is taken at the transition in a 1.6 ps time window. Assuming a Gaussian distribution, the mean value and standard deviation are investigated as a function of the device input power for both device types.
In DQPSK systems with direct detection, the signals are received using a delay interferometer (DI) followed by a balanced receiver. Fluctuations of the received power as well as deviations from ideal phase transitions can strongly degrade the signal.
As a subsequent comparison of the amplitude and phase characteristics of bulk and QD SOA will show, both SOA types are practically identical with respect to amplitude effects. However, QD SOA cause less phase errors due to their inherently lower alpha-factors, a fact that can significantly improve the input power dynamic range.
5.1 Amplitude fluctuations
Figure 6 shows the dependence of mean values (■, filled symbols) and standard deviations (□, open symbols) of bulk (blue) and QD SOA (black) as a function of the channel input power. In all cases, bulk and QD SOA behave identically. Measured at optimum receiver input power, the standard deviations are unaffected by input power levels above −10 dBm. No difference between bulk and QD SOA is observed with respect to the signal amplitude, so the observed dynamic range difference must be caused by phase effects.
Fig. 6 Amplitude transitions in bulk (blue) and QD SOA (black): Dependence of mean values (■, filled symbols) and standard deviations (□, open symbols) as a function of the channel input power. For all three possible transitions, bulk and QD SOA show an identical behavior. Measured at optimum receiver input power, the standard deviations are unaffected by input power levels above −10 dBm. No difference between bulk and QD SOA is observed with respect to the signal amplitude.
5.2 Phase fluctuations
Direct detection receivers for differential phase encoded signals are particularly susceptible to errors caused by deviations from the ideal phase transitions. For a power penalty less than 2 dB the phase error at the DI must be less than 10° [15,16]. Due to the fact that typical NRZ-DQPSK transmitters show a fast amplitude transition if a phase change occurs [17], SOA can induce errors by amplitude and phase fluctuations [18]. Since the gain saturation of both devices is similar (see Section 5.1) the observed IPDR difference must be attributed to phase induced errors. Typically, in SOA the phase recovery is slower than the amplitude recovery [5,19]. As a consequence amplitude transitions between bit slots influence the signal phase also at the decision point in the middle of the bit slot which introduces bit errors.
As an example, the bulk SOA 1 ch. power penalty for BER = 10−9 is depicted in Fig. 7(a) . The power penalty curves are marked according to the corresponding limits of the IPDR. For input power levels below −10 dBm, the DQPSK signal is limited by noise. For input power levels above −10 dBm, saturation of the SOA induces phase errors. The main difference between the samples arises for high input powers. Therefore, the phase limitations on the DQPSK signal performance is studied.
Fig. 7 Comparison of QD and bulk power penalty for all channels and BER. (a) Bulk SOA 1 ch. power penalty for BER = 10−9 as an example. The power penalty curves are marked according to the corresponding limits of the IPDR. For low input power levels, the DQPSK signal is degraded by noise (green). For high input powers, saturation of the SOA induces phase errors (red). (b) Power penalty for QD SOA vs. power penalty for bulk SOA when increasing the channel input power for BER = 10−9 (○: 1 ch., +: 2 ch) and BER = 10−3 (□: 1 ch., × : 2 ch.). The penalty is attributed to either noise or phase errors. All measurements shown in Fig. 3 displayed a similar ratio of the penalties. The largest difference between the samples arises for high input powers. (c) The power penalty for high input powers can be related to an effective phase error of the delay interferometer by the relation presented in [15]. The slope of a linear fit is 0.5, which actually corresponds to the ratio of the respective alpha-factors. The very good agreement between the calculated and measured ratio of the alpha-factors provides the explanation of the advantage of QD SOA in terms of IPDR: The smaller alpha-factor of QD SOA reduces the phase impairments.
Figure 7(b) shows the penalty for the QD SOA vs. the power penalty for the bulk SOA when increasing the channel input power levels for BER = 10−9 (○: 1 ch., +: 2 ch) and BER = 10−3 (□: 1 ch., ×: 2 ch.). The penalty is attributed to either noise or phase errors. All measured data which have been displayed in Fig. 3 resulted in a similar penalty ratio.
The influence of small phase errors is equivalent to the effect of deviations from the optimum operating point of the delay interferometer. For demonstrating this influence on the power penalty, we calculate the equivalent phase misalignment of a DI that would lead to the actually measured power penalty [15]. The absolute value of this phase error contains large uncertainties due to the fact that the phase error probability density is unknown. However, bulk and QD SOA are operated under identical conditions, and the calculated phase errors of the bulk SOA can be therefore compared to the phase errors of the QD SOA. As the biggest phase error will determine the bit error ratio, the calculated values give an estimate of the worst-case phase error.
Figure 7(c) compares the calculated equivalent phase errors for the bulk SOA to the equivalent phase errors for the QD SOA. A linear fit of the data shows a slope of 0.5. Assuming that the observed power penalty can be completely attributed to phase errors, this slope gives the ratio of the alpha-factors of both devices. It is in good agreement with the results extracted from the independently measured P2P phase changes using the FREAG technique, shown in Fig. 1(c). The fast amplitude transients in NRZ-DQPSK signals induce amplitude fluctuations in the SOA. These fluctuations induce carrier density fluctuations, which in turn cause refractive index variations and such create phase errors. Due to the fact that the alpha-factor in QD SOA is low, the amplitude to phase conversion is reduced compared to bulk SOA, so less phase errors are introduced. This general advantage of QD SOA also applies for other differential phase encoded formats like NRZ/RZ-DPSK or RZ-DQPSK.
6. Conclusion
The input power dynamic range (IPDR) for a 28 GBd NRZ-DQPSK signal amplified in a 1.5 µm QD SOA is improved more than 10 dB compared to a low confinement bulk SOA, which was especially designed for linear amplification. This enhancement found in QD SOA is attributed to the smaller alpha-factor, a fact which reduces phase impairments of differentially phase encoded signal. The IPDR for a QD SOA is about 20 dB at a bit error ratio of BER = 10−9 and exceeds 32 dB for a bit error ratio of BER = 10−3.
Acknowledgments
This work was supported by the German Research Foundation (DFG), the DFG Center for Functional Nanostructures of the University of Karlsruhe (CFN), and the Karlsruhe School of Optics & Photonics (KSOP).
References and links
1. D. R. Zimmerman and L. H. Spiekman, “Amplifiers for the masses: EDFA, EDWA, and SOA amplets for metro and access applications,” J. Lightwave Technol. 22(1), 63–70 (2004). [CrossRef]
2. M. Sauer, and J. Hurley, “Experimental 43 Gb/s NRZ and DPSK performance comparison for systems with up to 8 concatenated SOAs,” in Lasers and Electro-Optics, 2006 and 2006 Quantum Electronics and Laser Science Conference. CLEO/QELS 2006. Conference on, (2006), p. CThY2.
3. E. Ciaramella, A. D’Errico, and V. Donzella, “Using semiconductor-optical amplifiers with constant Envelope WDM Signals,” IEEE J. Quantum Electron. 44(5), 403–409 (2008). [CrossRef]
4. J. D. Downie, and J. Hurley, “Effects of dispersion on SOA nonlinear impairments with DPSK signals,” in Proc. of LEOS 2008, (2008), p. WX3.
5. T. Vallaitis, C. Koos, R. Bonk, W. Freude, M. Laemmlin, C. Meuer, D. Bimberg, and J. Leuthold, “Slow and fast dynamics of gain and phase in a quantum dot semiconductor optical amplifier,” Opt. Express 16(1), 170–178 (2008). [CrossRef] [PubMed]
6. R. Brenot, F. Lelarge, O. Legouezigou, F. Pommereau, F. Poingt, L. Legouezigou, E. Derouin, O. Drisse, B. Rousseau, F. Martin, and G. H. Duan, “Quantum dots semiconductor optical amplifier with a-3dB bandwidth of up to 120 nm in semi-cooled operation,” in Proc. Optical Fiber Communication Conference (OFC'08), (San Diego, CA, USA, 2008), p. OTuC1.
7. T. Akiyama, M. Sugawara, and Y. Arakawa, “Quantum-dot semiconductor optical amplifiers,” Proc. IEEE 95(9), 1757–1766 (2007). [CrossRef]
8. R. Bonk, C. Meuer, T. Vallaitis, S. Sygletos, P. Vorreau, S. Ben-Ezra, S. Tsadka, A. R. Kovsh, I. L. Krestnikov, M. Laemmlin, D. Bimberg, W. Freude, and J. Leuthold, “Single and multiple channel operation dynamics of linear quantum-dot semiconductor optical amplifier,” in Proc. European Conference on Optical Communication,2008. ECOC 2008, (Brussels, Belgium, 2008), p. Th.1.C.2.
9. H. A. Haus, “The noise figure of optical amplifiers,” IEEE Photon. Technol. Lett. 10(11), 1602–1604 (1998). [CrossRef]
10. T. Briant, P. Grangier, R. Tualle-Brouri, A. Bellemain, R. Brenot, and B. Thedrez, “Accurate determination of the noise figure of polarization-dependent optical amplifiers: theory and experiment,” J. Lightwave Technol. 24(3), 1499–1503 (2006). [CrossRef]
11. F. Lelarge, B. Dagens, J. Renaudier, R. Brenot, A. Accard, F. van Dijk, D. Make, O. L. Gouezigou, J.-G. Provost, F. Poingt, J. Landreau, O. Drisse, E. Derouin, B. Rousseau, F. Pommereau, and G.-H. Duan, “Recent advances on InAs/InP quantum dash based semiconductor lasers and optical amplifiers operating at 1.55µm,” IEEE J. Sel. Top. Quantum Electron. 13(1), 111–124 (2007). [CrossRef]
12. C. Dorrer and I. Kang, “Real-time implementation of linear spectrograms for the characterization of high bit-rate optical pulse trains,” IEEE Photon. Technol. Lett. 16(3), 858–860 (2004). [CrossRef]
13. F. Ginovart, J. C. Simon, and I. Valiente, “Gain recovery dynamics in semiconductor optical amplifier,” Opt. Commun. 199(1-4), 111–115 (2001). [CrossRef]
14. A. A. M. Saleh and I. M. I. Habbab, “Effects of semiconductor-optical-amplifier nonlinearity on the performance of high-speed intensity-modulation lightwave systems,” IEEE Trans. Commun. 38(6), 839–846 (1990). [CrossRef]
15. K.-P. Ho, “The effect of interferometer phase error on direct-detection DPSK and DQPSK signals,” IEEE Photon. Technol. Lett. 16(1), 308–310 (2004). [CrossRef]
16. H. Kim and P. J. Winzer, “Robustness to laser frequency offset in direct-detection DPSK and DQPSK systems,” J. Lightwave Technol. 21(9), 1887–1891 (2003). [CrossRef]
17. P. J. Winzer and R.-J. Essiambre, “Advanced optical modulation formats,” Proc. IEEE 94(5), 952–985 (2006). [CrossRef]
18. X. Wei and L. Zhang, “Analysis of the phase noise in saturated SOAs for DPSK applications,” IEEE J. Quantum Electron. 41(4), 554–561 (2005). [CrossRef]
19. J. Wang, A. Maitra, C. G. Poulton, W. Freude, and J. Leuthold, “Temporal dynamics of the alpha factor in semiconductor optical amplifiers,” J. Lightwave Technol. 25(3), 891–900 (2007). [CrossRef]
