4
IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 28, NO. 8, APRIL 15, 2016 919 Widely Continuous Tunable ps Pulse Train Generation Based on SOA Nonlinear Dynamics X. You, C. Gosset, and P. Gallion, Senior Member, IEEE Abstract—We propose a simple setup based on nonlinear dynamics in a semiconductor optical amplifier (SOA) and solitonic compression for the generation of stable and tunable repetition rate picosecond (ps) optical pulses. A quantum-well SOA is used to amplify electrical-optic-modulated rectangular optical signal, during which the nonlinear amplification process induces front-edge distortion. Due to the non-zero linewidth enhancement factor, the front-edge distortion can be isolated by the optical filtration of the chirped frequency component, which eventually becomes Gaussian-like pulse train of tens of ps FWHM pulsewidth. Following a stage of solitonic compression and a stage of pedestal suppression, we are able to obtain a train of 1-ps high-quality pulses for which the root mean square timing jitter is <2.5 ps. The repetition rate exhibits a wide continuous tunability from 70 to 790 MHz as well a wide wavelength tunability in the C-band. Index Terms— Optical pulse generation, optical propagation in nonlinear media, semiconductor optical amplifiers, optical pulse compression. I. I NTRODUCTION T RAINS of ultrashort optical pulses with duration from few tens of picosecond down to the femtosecond range are called for various applications, such as investigation of ultrafast processes (e.g., time-resolved spectroscopy) and fast optical data transmission signal characterization. For the latter which is more interested in telecommunication, the under sampling technique with a pulsed local oscillator (PLO) is a typical example [1], [2]. It makes use of a PLO to beat asynchronously with the signal under test. As a result, access to the intrinsic statistical properties such as constellation, eye diagram, bit error rate (BER) and signal noise ratio (SNR) is obtained [3]. For implementation, a train of brief optical pulse (typically in the 1 ps range or below) with repetition rate typically in the 100 MHz - 1 GHz range is required. As a review of pulse generation techniques, passive mode locked lasers are able to generate pulses of duration shorter than 100 fs [4], typically shorter than 400 fs [5]. The pulse repetition frequencies are from 1 GHz to 175 GHz [6]–[8], which are hardly tunable due to limitation of cavity. More recently Wilcox et al. [9] have reported repetition Manuscript received June 23, 2015; revised December 7, 2015; accepted January 9, 2016. Date of publication January 13, 2016; date of current version March 10, 2016. This work was supported by the French National Research Agency through the OCELOT Project. The authors are with the Laboratoire de Traitement et Communication de l’Information, Centre National de la Recherche Scientifique, Télécom ParisTech, École Nationale Supérieure des Télécommunications, Paris 75013, France (e-mail: [email protected]; christophe.gosset@ telecom-paristech.fr; [email protected]). Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LPT.2016.2517650 Fig. 1. Diagram of experimental setup. frequency tunable passively mode-locked VECSEL that emits sub-500-fs pulses over continuous tuning range from 2.78 to 7.87 GHz, but the pulse width increases up to 2.5 ps above 6 GHz. It is based on the mechanical tuning of the laser cavity length. However, the electro-optic modulation method provides an alternative way, such as the direct modulation of a laser diode or gain switching [10]–[12], the phase modulation following by a frequency dispersive medium called chirping compres- sion [13], [14] or phase modulation following by an optical filter [15]. There are also techniques using the electro- absorption modulator [16], which usually generates optical pulses of tens of ps pulse width and of repetition rate tens of GHz. The proposed method combines the large repetition rate tunability of generated electrical rectangular signal and the nonlinear amplification property of SOA. It provides a unique and simple way for generation of Gaussian like pulse train of tens of ps FWHM pulse width, which has already been reported previously [17]. While in this letter we explore in the compression of pulse width using solitonic compression [18], the suppression of pedestal due to higher order nonlinear effects in optical fiber [19]. Finally we are able to achieve ps pulse train of continuous repetition rate tunability from 70 MHz to 790 MHz. The RMS timing jitter measured using RF spectrum analysis [20] is less than 2.5 ps. The generated pulses have also the wavelength tunability in C band benefiting from a tunable continuous wave laser. II. EXPERIMENTAL SETUP As shown in Fig. 1, the experimental setup consists of 3 parts: part 1 for the preliminary pulse generation, 1041-1135 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 28, NO. 8, APRIL ... · pulse of Gaussian-like shape evolves like high order solitons, which always experience an initial pulse compression

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

  • IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 28, NO. 8, APRIL 15, 2016 919

    Widely Continuous Tunable ps Pulse TrainGeneration Based on SOA Nonlinear Dynamics

    X. You, C. Gosset, and P. Gallion, Senior Member, IEEE

    Abstract— We propose a simple setup based on nonlineardynamics in a semiconductor optical amplifier (SOA) andsolitonic compression for the generation of stable and tunablerepetition rate picosecond (ps) optical pulses. A quantum-wellSOA is used to amplify electrical-optic-modulated rectangularoptical signal, during which the nonlinear amplification processinduces front-edge distortion. Due to the non-zero linewidthenhancement factor, the front-edge distortion can be isolatedby the optical filtration of the chirped frequency component,which eventually becomes Gaussian-like pulse train of tens ofps FWHM pulsewidth. Following a stage of solitonic compressionand a stage of pedestal suppression, we are able to obtaina train of 1-ps high-quality pulses for which the root meansquare timing jitter is

  • 920 IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 28, NO. 8, APRIL 15, 2016

    Fig. 2. On the left: pulse shapes after the Mach-Zehnder modulator, afterthe SOA and after filter in part 1. On the right: the corresponding spectra.

    part 2 of solitonic compression and part 3 for pedestalsuppression. After each part, 10% of the power is used formonitoring. An optical switch is employed to facilitate theswitching between signals after each stage. The fibers beforethe optical switch are of length that can be ignored.

    In part 1, an NRZ waveform generator generates periodicrectangular waveform as driven signal for the modulator. Theintensity modulated laser beam is then injected in a quantumwell SOA (model JDS Uniphase CQF871/0) that inducesfront edge distortion due to its gain nonlinearities (Fig. 2 onthe left). As shown by the non-symmetrical spectrum afterSOA in Fig. 2, the distortion contribution has a frequencyshift as the linewidth enhancement factor is non-zero. Thenby taking benefit of an optical filter, we are able to achieve40 ps pulse train [17].

    Part 2 consists of an Erbium doped fiber amplifier (EDFA)and a given length of standard single mode fiber (SSMF).While the SSMF is anomalous dispersive, in this case, opticalpulse of Gaussian-like shape evolves like high order solitons,which always experience an initial pulse compression in anom-alous dispersive fibers [21]. Simulation results show that fordifferent fiber lengths of 12, 4.2 and 2 km the best compressionratios are respectively 4.92, 15.31 and 35.36 with proper initiallaunched powers. The launched power is always higher forshorter fiber length.

    In experiment, a 4.2 km SSMF is chosen as a tradeoffbetween initial power level and compression ratio. Thus therequired initial peak power is 541.8 mW that is equivalent to5.25 pJ per pulse, and it corresponds to 7.2 dBm at repetitionrate of 250 MHz. As a result, the full width at half maxi-mum (FWHM) is compressed to 4.49 ps under assumptionof hyperbolic secant pulse shape. The autocorrelation tracein Fig. 3 reveals an important issue, the pedestal, which canlargely reduce the pulse quality.

    Part 3 consists of an EDFA, a SSMF, a polarizationcontroller and an optical filter. The high output power ofEDFA induces nonlinearities inside the SSMF, which includethe higher order terms in nonlinear Schr’́odinger equationsuch as 3rd order dispersion, the intrapulse stimulated Ramanscattering and the self-steepening when pulse width becomecomparable to 1 ps [19].

    Fig. 3. Autocorrelation trace of pulses after the pulse compression in part 1.The launched power is 7.2 dBm at the repetition rate 250 MHz while the fiberlength is 4.2 km. The inset figure is the corresponding spectrum.

    Fig. 4. On the top: normalized autocorrelation trace and spectrum in part 3before the optical filter. On the bottom: normalized autocorrelation trace andspectrum in part 3 after the filter.

    This is further proved by the asymmetric spectrum in topright of Fig. 4, as only the higher order terms can bringasymmetric spectrum broadening when the input pulse issymmetric.

    Then by carefully tuning the optical filter, we obtain pulseautocorrelation trace as shown in Fig. 4 bottom left. The pulsethat we finally obtain benefits largely from the higher ordernonlinear effects during propagation in SSMF in part 3.

    Under assumption of hyperbolic secant pulse shape, theFWHM of filtered pulse is 1.1303 ps. The peak level topedestal level ratio of autocorrelation trace is 39. The timebandwidth product is calculated to be 0.4969, which is veryclose to the bandwidth-limited value of 0.315 for hyperbolicsecant shape pulses.

    In experiment, the launched power in part 3 is around835.7 pJ per pulse for 1.2 km SSMF. The optical filteringinduces a loss of energy about 12 dB, which is defined as theproportion of the output power of the filter to the input one.The remaining energy after the optical filter is about 21 pJper pulse. For pulses of smaller launched power, the spectrum

  • YOU et al.: WIDELY CONTINUOUS TUNABLE ps PULSE TRAIN GENERATION 921

    Fig. 5. Pulse profile simulation for 1200 m SSMF. The input pulse is ofFWHM 4.49 ps that correspond to the output of part 2. The input peak power is60 W or 491.5 pJ per pulse. For parameters of SSMF, the 3rd order dispersionslope at 1540 nm is 0.09 ps·nm−2·km−1, the value of intrapulse Ramanscattering coefficient is 3 fs, the dispersion parameter is 20.3 ps·nm−1·km−1,and the nonlinear coefficient is 0.002 W−1·m−1.

    broadening induced by nonlinear effects is not enough forseparating the pulse and the pedestal using optical filter.

    As the observation using autocorrelator loses the informa-tion of pulse shape in time domain, simulation with split-stepFourier method is performed to get a view of pulse shapeevolvement during the propagation. All the parameters aretaken as typical values of SSMF shown in the description ofFig. 5. By varying the initial peak power, we find that witha power of 60 W we can obtain relatively good simulationresults. As the parameters of SSMF are all taken as typicalvalues, it is difficult to obtain perfect fit between the simulationand the experiment.

    In Fig. 5 we observe the split of pulse during propagation.Due to the anomalous dispersion, the fact that the new largepeak propagates slower than the initial peak means a transferof energy to the low frequency side in the frequency domain.This result is consistent with Fig. 4, and this new large peakis experimentally what we observed after the filter.

    III. RESULTS AND DISCUSSION

    For application in optical sampling, the pulse width, thesignal-to-noise ratio (SNR) and the timing jitter are the keyparameters to characterize the quality of pulse train. The pulsesFWHM are derived from the autocorrelation trace under theassumption of a hyperbolic secant shape. The extinction ratioof the autocorrelation trace is taken to characterize the pulsequality instead of SNR.

    A. Timing Jitter Measurements

    The RF spectrum analysis for timing jitter measurementshas been extensively studied [22]–[25]. In experiment we areusing the method described in [22]. The main difficulty lies inthe determination of lower frequency boundary f1 and upperfrequency boundary fu for the integration.

    Fig. 6. Results of RMS timing jitter value as a function of the lowerfrequency f1. The upper frequency fu is set to be 118 kHz in this measurement.

    Fig. 7. Results of RMS timing jitter value as a function of the upperfrequency fu. The lower frequency f1 is set to be 2 kHz.

    Both timing jitters from 15th and 19th order harmonicsare calculated (Fig. 6 and 7). Ideally the two curves shouldcoincide.

    In Fig. 6, the 2 curves increase rapidly for lower frequencyvalue smaller than 2 kHz, which suggests that the spectra atthe range larger than 2 kHz is the major contribution to thetiming jitter [22].

    In Fig. 7, the 2 curves become almost flat above 118 kHz.It means that the integration after 118 kHz is only accumula-tion of background noise.

    We repeat the measurements for 6 times and we deter-mine the integration frequency range in a similar way. Theaverage value respectively for 15th and 19th harmonics are1.647 and 2.52 ps.

    Comparing with the RMS timing jitter estimation of 40 pspulses at the output of part 1 (Fig. 1) of about 2.5 ps,we conclude that the major contribution to the timing jittercomes from the electrical part of the system. This timing jitterestimation is a functionality providing by the Agilent DCA-Xoscilloscope with optical input for pulses larger than 30 ps.

    B. The Tunability on Repetition Rate

    In part 2 of experimental diagram (Fig. 1), an EDFA beforethe SSMF is used to amplify optical pulses to the required

  • 922 IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 28, NO. 8, APRIL 15, 2016

    peak power, which also means that the average output powerwill increase as the repetition rate increases. The output powerrange of EDFA appears to be the limiting factor for the upperrange of repetition rate tunability.

    For example, a 40 ps 250 MHz repetition rate pulsetrain at the output of part 1 is amplified to 7.5 dBm asrequired for solitonic compression in 4.2 km SSMF. TheEDFA in our experiment has an output power range between2 and 12.5 dBm. This means that the repetition rate can onlybe tuned between 70 MHz and 790 MHz.

    Then in the part 3, the average power required is experi-mentally determined to be 23.2 dBm for 1.2 km SSMF withpulse repetition rate at 500 MHz. The EDFA before 2nd stagehas an output power range from 12 dBm to 26.75 dBm,which means that the limit of repetition rate is between40 MHz and 1.1 GHz.

    Experiments are performed at different repetition rates of100, 250 and 500 MHz. The FWHM pulse widths are respec-tively 0.986, 0.895 and 1.136 ps under hyperbolic secant shapeassumption.

    As for the solitonic compression, higher compression ratioalways means higher peak power. At the same time higherrepetition rate also requires higher average power. Thereforewe need to make a tradeoff between the tunability and thepulse width due to the limited output range at the EDFA.

    IV. CONCLUSION

    We have performed the ps optical pulse train generationexperiment based on SOA nonlinear dynamics and solitoniccompression. The result shows that this technique is verypromising for versatile source for ultrashort pulse generation.The tunability range and the pulse quality are very suitablefor optical sampling. The wavelength can cover the full rangeof telecommunication band. The major contribution to theobserved timing jitter comes from electrical part meaningthat a better jitter performance by improving jitter noise ofelectrical devices in the future can be expected.

    REFERENCES

    [1] I. Shake, W. Takara, S. Kawanishi, and Y. Yamabayashi, “Optical signalquality monitoring method based on optical sampling,” Electron. Lett.,vol. 34, no. 22, pp. 2152–2154, Oct. 1998.

    [2] C. Dorrer, “Monitoring of optical signals from constellation diagramsmeasured with linear optical sampling,” J. Lightw. Technol., vol. 24,no. 1, pp. 313–321, Jan. 2006.

    [3] P. Gallion, X. You, C. Gosset, and F. Grillot, “Bandwidth and dynamicrange of a pulsed local oscillator coherent optical detection. Applicationto linear optical sampling,” Proc. SPIE, Next-Generation Opt. Commun.,Compon., Sub-Syst., Syst. III, vol. 9009, p. 90090O, Dec. 2013.

    [4] A. H. Quarterman et al., “A passively mode-locked external-cavitysemiconductor laser emitting 60-fs pulses,” Nature Photon., vol. 3,no. 12, pp. 729–731, Nov. 2009.

    [5] P. Klopp et al., “Mode-locked InGaAs-AlGaAs disk laser generatingsub-200-fs pulses, pulse picking and amplification by a tapered diodeamplifier,” Opt. Exp., vol. 17, no. 13, pp. 10820–10834, Jun. 2009.

    [6] S. Hoogland, A. Garnache, I. Sagnes, J. S. Roberts, and A. C. Tropper,“10-GHz train of sub-500-fs optical soliton-like pulses from a surface-emitting semiconductor laser,” IEEE Photon. Technol. Lett., vol. 17,no. 2, pp. 267–269, Feb. 2005.

    [7] D. Lorenser et al., “50-GHz passively mode-locked surface-emittingsemiconductor laser with 100-mW average output power,” IEEEJ. Quantum Electron., vol. 42, no. 8, pp. 838–847, Aug. 2006.

    [8] K. G. Wilcox et al., “175 GHz, 400-fs-pulse harmonically mode-locked surface emitting semiconductor laser,” Opt. Exp., vol. 20, no. 7,pp. 7040–7045, Mar. 2012.

    [9] K. G. Wilcox, A. H. Quarterman, H. E. Beere, D. A. Ritchie, andA. C. Tropper, “Repetition-frequency-tunable mode-locked surface emit-ting semiconductor laser between 2.78 and 7.87 GHz,” Opt. Exp.,vol. 19, no. 23, pp. 23453–23459, Nov. 2011.

    [10] J. T. Ong et al., “Subpicosecond soliton compression of gainswitched diode laser pulses using an erbium-doped fiber amplifier,”IEEE J. Quantum Electron., vol. 29, no. 6, pp. 1701–1707,Jun. 1993.

    [11] K. A. Ahmed, K. C. Chan, and H.-F. Liu, “Femtosecond pulse gener-ation from semiconductor lasers using the soliton-effect compressiontechnique,” IEEE J. Sel. Topics Quantum Electron., vol. 1, no. 2,pp. 592–600, Jun. 1995.

    [12] H. Hu et al., “Pulse source based on directly modulated laser and phasemodulator,” Opt. Exp., vol. 15, no. 14, pp. 8931–8937, Jul. 2007.

    [13] T. Kobayashi, H. Yao, K. Amano, Y. Fukushima, A. Morimoto, andT. Sueta, “Optical pulse compression using high-frequency electroop-tic phase modulation,” IEEE J. Quantum Electron., vol. 24, no. 2,pp. 382–387, Feb. 1988.

    [14] H. Hu et al., “10 GHz pulse source for 640 Gbit/s OTDM based onphase modulator and self-phase modulation,” Opt. Exp., vol. 19, no. 26,pp. B343–B349, Nov. 2011.

    [15] Q. Wang, L. Huo, Y. Xing, C. Lou, and B. Zhou, “OpticalGaussian pulse generator using phase modulator based spectral slicingand compression,” in Proc. Lasers Electro-Opt. (CLEO), Jun. 2013,pp. 1–2, 9–14.

    [16] K. Taira and K. Kikuchi, “Subpicosecond pulse generation usingan electroabsorption modulator and a double-stage pulse compres-sor,” IEEE Photon. Technol. Lett., vol. 15, no. 9, pp. 1288–1290,Sep. 2003.

    [17] X. You, G. G. Perez-Sanchez, and C. Gosset, “Continuously-tunable,from MHz to GHz range, pulse train generation and its dynamicproperties,” Proc. SPIE, Nonlinear Freq. Generat. Convers., Mater.,Devices, Appl. XIII, vol. 8964, p. 89641P, Feb. 2014.

    [18] G. Steinmeyer, D. H. Sutter, L. Gallmann, N. Matuschek, and U. Keller,“Frontiers in ultrashort pulse generation: Pushing the limits in linearand nonlinear optics,” Science, vol. 286, no. 5444, pp. 1507–1512,Nov. 1999.

    [19] G. P. Agrawal, “Effect of intrapulse stimulated Raman scattering onsoliton-effect pulse compression in optical fibers,” Opt. Lett., vol. 15,no. 4, pp. 224–226, Feb. 1990.

    [20] D. von der Linde, “Characterization of the noise in continuously oper-ating mode-locked lasers,” Appl. Phys. B, vol. 39, no. 4, pp. 201–217,1986.

    [21] L. F. Mollenauer, R. H. Stolen, J. P. Gordon, and W. J. Tomlinson,“Extreme picosecond pulse narrowing by means of soliton effect insingle-mode optical fibers,” Opt. Lett., vol. 8, no. 5, pp. 289–291,May 1983.

    [22] Y. Matsuo, N. Nishizawa, M. Mori, and T. Goto, “Measurement oftiming jitter in wavelength tunable femtosecond soliton pulses,” Opt.Rev., vol. 7, no. 4, pp. 317–322, Mar. 2000.

    [23] E. Yoshida and M. Nakazawa, “Measurement of the timing jitter andpulse energy fluctuation of a PLL regeneratively mode-locked fiberlaser,” IEEE Photon. Technol. Lett., vol. 11, no. 5, pp. 548–550,May 1999.

    [24] J. P. Tourrenc et al., “Experimental investigation of the timing jitter inself-pulsating quantum-dash lasers operating at 1.55 µm,” Opt. Exp.,vol. 16, no. 22, pp. 17706–17713, Oct. 2008.

    [25] A. Finch, X. Zhu, P. N. Kean, and W. Sibbett, “Noise characterizationof mode-locked color-center laser sources,” IEEE J. Quantum Electron.,vol. 26, no. 6, pp. 1115–1123, Jun. 1990.

    /ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 600 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages false /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 400 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

    /Description >>> setdistillerparams> setpagedevice