2678
OPTICS LETTERS / Vol. 31, No. 18 / September 15, 2006
Pulse-amplitude equalization using a polarization-maintaining laser resonator Young Min Jhon, Young Tae Byun, and Deok Ha Woo Photonics Research Center, Korea Institute of Science and Technology, 39-1 Hawolgok, Seongbuk, Seoul 136-791, Korea Received April 6, 2006; revised June 1, 2006; accepted June 19, 2006; posted June 28, 2006 (Doc. ID 69294); published August 25, 2006 For the first time to our knowledge, pulse-amplitude equalization of rational-harmonically mode-locked fiber ring laser pulses has been experimentally demonstrated using a polarization-maintaining laser resonator without any additional device. The pulse-amplitude distribution of the laser pulses was controlled by the modulator driving power, and stable pulse-amplitude-equalized pulses with repetition rates of 20, 30, and 40 GHz have been obtained in the linear region of the modulator. © 2006 Optical Society of America OCIS codes: 060.2320, 060.2410, 060.2420, 140.3510, 140.4050.
High-repetition-rate ultrashort pulse laser sources are essential for ultrahigh-speed optical communication and all-optical digital signal processing. The rational-harmonically mode-locked fiber ring laser (RHML-FRL) has been shown to be very useful for producing high-repetition-rate laser pulses while preserving all the merits of mode-locked fiber ring lasers, such as ultrashort pulse width, low chirp, broad tunability, and high power.1 In RHML-FRLs, laser pulses at repetition rates of integral multiples of the modulator driving frequency fm can be generated by slightly detuning fm from the harmonic mode-locking condition, that is, by satisfying fm = 共n ± 1 / p兲 fc (n , p are integers), laser pulses at pfm = 共np ± 1兲 fc can be generated, where fc is the fundamental cavity frequency. However, the pulse amplitude of RHMLFRLs suffers from inherent inequality, which makes the lasers inadequate for practical applications. Various studies attempting to equalize the pulse amplitudes have been reported, including methods such as optical feedback,2 even-order modulation with intracavity filtering,3 nonlinear polarization rotation,4 a semiconductor optical amplifier loop mirror,5 a nonlinear amplifying loop mirror,6 a nonlinear optical loop mirror,7 a semiconductor optical amplifier loop modulator,8 and a double pass with polarization control.9 These methods all require additional devices and precise control to achieve pulse-amplitude equalization (PAE). Recently, PAE was obtained in RHML-FRL by precisely controlling the transfer function of the modulator by use of a dual drive10 and by simply controlling the bias point and modulation index11,12 to obtain nonlinear modulation so that each laser pulse inside one modulation period 共1 / fm兲 experiences the same loss. In this Letter, we experimentally demonstrate PAE of RHML-FRL pulses using a polarizationmaintaining (PM) laser resonator by simply optimizing the modulator driving power in the linear region without any additional device or control. The subharmonic components were highly suppressed, down to 30 dB. The laser pulses need not experience the same loss in one modulation period but need only experience the same average loss for multiple (as many 0146-9592/06/182678-3/$15.00
times as the rational-harmonic order) round trips to yield PAE in the PM laser resonator. To achieve PAE in a RHML-FRL, each laser pulse in the same modulation period must experience the same average loss without any perturbations for multiple (as many times as the rational-harmonic order) round trips.1 In a regular single-mode fiber laser resonator, the polarization of the laser pulse can be changed by thermal or vibrational noise, which induces polarization-dependent loss, resulting in unequal pulses. However, in a PM laser resonator, the polarization instabilities are eliminated, and thus the PAE condition can be maintained. The optical transfer function of the modulator can be given by T共t兲 = 1/2共1 − ␣兲„1 + sin兵关b + M cos共mt兲兴其…, where ␣ is the insertion loss, b is the normalized bias, M is the modulation index, and m is the modulation frequency (b = Vb / V, M = Vm / V, Vb is the modulation bias, Vm is the modulation amplitude, and V is the half-wave voltage). In Refs. 10–12, b and M were precisely controlled in the nonlinear region of the modulator such that each laser pulse in one modulation period experienced the same loss; thus any polarization perturbation equally affected each laser pulse and thus the PAE condition could be maintained. However, the laser output showed poor stability because of the strict conditions for b and M, and even though PAE was accomplished, pulse width variation remained. In our study, the modulator is operated in the linear region (b = 0, M ⬍ 0), such that laser pulses experience the same average loss during multiple (as many times as the rational-harmonic order) round trips1 but not the same loss in one modulation period; thus the rational-harmonic mode-locking condition is much more easily maintained, delivering stable laser output. A schematic diagram of the experimental setup is shown in Fig. 1. The laser resonator is composed of all PM fibers and PM components to prevent any polarization instabilities and ensure stable laser operation.13 The PM fibers were spliced by a PM fu© 2006 Optical Society of America
September 15, 2006 / Vol. 31, No. 18 / OPTICS LETTERS
Fig. 1. PM laser resonator for PAE in RHML-FRL. WDM, wavelength-division multiplexer; M-Z, Mach–Zehnder; other abbreviations defined in text.
2679
of 100 ps coincides well with the modulation frequency of 9.996500 GHz. The fundamental cavity frequency fc of the laser obtained from the supermodebeating (SMB) spectrum was measured to be 4.49 MHz, corresponding to an optical cavity length of 66.8 m, where the laser cavity length was slightly detuned from the exact mode-locking condition to enhance SMB. The SMB power was suppressed more than 55 dB at the exact mode-locking condition. Mode locking was obtained with a modulation index of 0.31–0.41 at fm = 9.996500 GHz, which is the 2167th harmonic of fc. The pulse width was measured to be 4.0 ps with a linewidth of 0.68 nm, giving a time–bandwidth product of 0.34, implying a transform-limited sech2 pulse as shown in Fig. 3. Detuning fm by 2.307 MHz 共fc / 2兲, 1.538 MHz 共fc / 3兲, and 1.153 MHz 共fc / 4兲, we obtained the second, third, and fourth RHML-FRL pulses at repetition rates of 20, 30, and 40 GHz, respectively. The SMB noise was suppressed below 45 dB within the mode-locking range of ±7 kHz, but the amplitudes of the laser pulses were not equal and showed a periodic behavior (repeated pulse distributions with modulation period). The oscilloscope traces of the fourth RHML-FRL pulses are shown in Fig. 4. Adjusting the modulator driving power, we could control the inequality of the laser pulses and obtain PAE by optimizing the driving power. We can find that tuning the modulator
Fig. 2. (Color online) (a) Oscilloscope trace and (b) SMB spectrum of the 10 GHz mode-locked laser pulses.
sion splicer (Ericsson FSU 925 PM-A/E) with accuracy of less than ±0.1° between the optical axes and loss of less than 0.5 dB. The 10 m PM erbium-doped fiber (EDF) was pumped by four 980 nm laser diodes with a total power of 400 mW. We added 24 m PM dispersionshifted fiber (DSF) to optimize dispersion. The laser was tunable from 1530 to 1560 nm by a tunable filter with a 3 dB bandwidth of 3 nm. A polarizationsensitive optical isolator was used to ensure unidirectional and single polarization laser operation. An optical delay line (Santec ODL-600, 0 – 200 ps, 0.1 ps) was introduced into the laser cavity to control the laser cavity length. The laser was modulated by a 10 GHz Mach–Zehnder intensity modulator (Sumitomo Osaka Cement T.MZH-1.5-10, V = 9.0 V at 10 GHz) driven by a signal generator (Anritsu Wiltron Synthesized CW Generator 68037B, 2 – 20 GHz, 20 dBm) through a RF amplifier (Miteq AMF-5B097102-33P, 30 dB, 33 dBm). The output of the laser was observed with a 50 GHz sampling oscilloscope (Tektronix CSA 803 with sampling head SD-32), a RF spectrum analyzer (Anritsu MS2688C) through a 45 GHz photodetector (NewFocus 1014), and an optical spectrum analyzer (Ando AQ6317B, 0.01 nm). The pulse width of the laser pulse was measured by an autocorrelator (Femtochrome FR-103XL). Figure 2 shows the oscilloscope trace and the RF spectrum of the laser output operating in the conventional mode-locking condition. The pulse separation
Fig. 3. (a) Optical spectrum and (b) autocorrelation trace of the 10 GHz mode-locked laser pulses.
Fig. 4. Oscilloscope traces of the fourth 共40 GHz兲 RHML laser pulses versus modulator driving power (modulation index M): (a) 19 dBm 共M = 0.31兲, (b) 20 dBm 共M = 0.35兲, (c) 21 dBm 共M = 0.40兲, (d) 22 dBm (M = 0.44, equalized), (e) 23 dBm 共M = 0.50兲, (f) 24 dBm 共M = 0.56兲. 共50 ps/ div.兲
2680
OPTICS LETTERS / Vol. 31, No. 18 / September 15, 2006
Fig. 5. RF spectra of fourth 共40 GHz兲 RHML laser pulses (a) without and (b) with PAE.
Fig. 6. (a) Optical spectrum and (b) autocorrelation trace of fourth 共40 GHz兲 RHML laser pulses.
driving power toward its optimum value [Fig. 4(d), 22 dBm, M = 0.44], the pulses with larger energies give their energies to pulses with smaller energies and finally reach PAE. Over the optimized value, the smaller pulses continue to grow to be as large as the former larger pulses, and vice versa. Thus, the shape of the pulse train shows symmetric behavior below and over the optimization point but with the positions of the larger pulses are changed. This procedure can be understood as finding an initial condition where each pulse in the modulation period has the same amplitude. When the modulator driving power is reduced from its optimum value, the cavity loss is decreased, resulting in an increase of the net gain, which increases the amplitudes of the laser pulses in the starting part of the pulse distribution in the modulation period, and vice versa. Similar characteristics were also observed for second and third rational-harmonic mode locking, where PAE were obtained using modulation indices M of 0.35 and 0.37, respectively. The RF spectra of the laser pulses with and without PAE are shown in Fig. 5. As shown in Fig. 5(a), lower harmonic components such as 10, 20, and 30 GHz components are also present with the 40 GHz component when the pulses are unequal. However, after PAE is obtained, the lower harmonic components are suppressed by 30 dB as in Fig. 5(b). The optical spectrum of the equalized pulses is shown in Fig. 6(a). We can find that the dominant modes are separated by 0.32 nm 共40 GHz兲, while the intermediate modes are well suppressed. The autocorrelation trace is shown in Fig. 6(b). The temporal spacing between the adjacent pulses is measured to be 25 ps, corresponding to a 40 GHz repetition rate. The pulse widths were measured to be 4.8 ps with a time–
bandwidth product of 0.39, indicating the presence of some phase modulation but still a nearly transformlimited sech2 pulse shape. The stability of the laser pulses were investigated by measuring the peak-topeak amplitude noise, which remained within 4% for 3 h of operation without active feedback control of laser cavity length,14 while 1% and 3% were obtained for the second and third rational-harmonic mode locking, respectively. To investigate the effect of polarization instabilities, we introduced an intracavity in-line polarization controller by applying nonuniform stress to the fiber and observed the laser output. As the polarization maintenance was broken, PAE pulses were no longer observed and the PAE condition could not be found through the entire tuning range of modulator driving power, including modulation frequency tuning. In conclusion, we have experimentally demonstrated pulse-amplitude equalization of a rationalharmonic mode-locked fiber ring laser pulses using a polarization-maintaining laser resonator by simply optimizing the modulator driving power in the linear region without any additional device. PAE pulses with repetition rates of 20, 30, and 40 GHz have been obtained by detuning the modulator driving frequency and optimizing the modulator driving power. We found from the experimental results that the pulse-amplitude inequalities observed so far in RHML-FRLs can be attributed mainly to polarization instabilities. The authors thank Yoo Seung Lee for experimental help and useful discussions. Y. Jhon’s e-mail address is
[email protected]. References 1. Z. Ahmed and N. Onodera, Electron. Lett. 32, 455 (1996). 2. C. G. Lee, Y. J. Kim, H. K. Choi, and C.-S. Park, Opt. Commun. 209, 417 (2002). 3. K. K. Gupta, N. Onodera, and M. Hyodo, Electron. Lett. 37, 948 (2001). 4. Z. Li, C. Lou, K.-T. Chan, Y. Li, and Y. Gao, IEEE J. Quantum Electron. 37, 33 (2001). 5. H. J. Lee, K. Kim, and H. G. Kim, Opt. Commun. 160, 51 (1999). 6. M.-Y. Jeon, H. K. Lee, J. T. Ahn, D. S. Lim, H. Y. Kim, K. H. Kim, and E.-H. Lee, Electron. Lett. 34, 182 (1998). 7. M.-Y. Jeon, H. K. Lee, J. T. Ahn, K. H. Kim, D. S. Lim, and E.-H. Lee, Opt. Lett. 23, 855 (1998). 8. W. W. Tang and C. Shu, IEEE Photon. Technol. Lett. 15, 21 (2003). 9. Y. Shiquan, L. Zhaohui, Z. Chunliu, D. Xiaoya, Y. Shuzhong, K. Guiyun, and Z. Qida, IEEE Photon. Technol. Lett. 15, 389 (2003). 10. Y. J. Kim, C. G. Lee, Y. Y. Chun, and C.-S. Park, Opt. Express 12, 907 (2004). 11. G. Zhu, H. Chen, and N. Dutta, J. Appl. Phys. 90, 2143 (2001). 12. X. Feng, Y. Lin, S. Yuan, G. Kai, W. Zhong, and X. Dong, IEEE Photon. Technol. Lett. 16, 1813 (2004). 13. B. K. Kim, J. C. Lee, Y. M. Jhon, M.-W. Kim, S. K. Kim, S. S. Choi, and M. S. Oh, Opt. Lett. 24, 391 (1999). 14. Y. M. Jhon, C. W. Jang, J. H. Kim, Y. T. Byun, and S. H. Kim, in IEEE LEOS 2002 (IEEE, 2002), paper TuBB4.
