We report a long term stabilized 4th order rational harmonic mode-locked (RHML) erbium doped fiber laser operating at 40Gb/s. The cavity length drift induced laser instability is overcomed by constructing a modified regenerative type feedback loop to actively adjust the modulation frequency of the 10GHz driving signal. The 4th RHML fiber laser is tested to be highly stable against environmental perturbations.
©2005 Optical Society of America
A stable optical pulse source at high repetition rate is a prerequisite for the next generation ultrahigh speed optical time division multiplexing (OTDM) communication system . Among various kinds of available sources, actively mode-locked fiber laser  has been considered to be a very promising candidate owning to its relative ease in configuration as well as its superior compatibility with fiber based OTDM systems. Since the cavity length of the fiber laser is relatively long, in order to produce pulse train with very high repetition rate, fiber lasers are usually harmonic mode-locked (HML)  or RHML . It is well-known that for the case of HML, the repetition rate of the produced pulse train equals the RF driving modulation frequency, while for the case of RHML, the repetition rate of the produced pulse train is instead some integer multiple of the RF driving modulation frequency .
To commercialize the actively mode-locked fiber lasers, one important issue we need to address is the long-term stabilization of the actively mode-locked pulse train . Unlike the case of HML, for the case of nth order RHML, since the effective gating width of the modulation transmission window is reduced by a factor of n , the produced pulse train is much more sensitive to the cavity length drift caused by environment perturbations. As being illustrated by Fig. 1, numerical simulations show that for 4th order RHML driven by a 10GHz clock, a thermal cavity length drift as small as only 0.5ps is large enough to completely destroy the produced 40Gb/s pulse train. Experimentally, the observed time in which the stabilized mode-locking state for 40Gb/s operation is only a few minuets for the best case.
The most obvious way one can come up with to solve the thermal cavity length drift problem is to dynamically control the cavity length. Using this method, Shan et al  have demonstrated a stabilized operation of HML by constructing a phase-locked loop using a PZT drum as the actuators. About 35m long fiber was wound on the PZT drum in order to achieve a 5ps maximum cavity length change needed for the compensation of the thermal induced length drift. Similarly to this idea, another solution is to dynamically adjust the modulation frequency using the regenerative feedback techniques. This regenerative type feedback control method was recently demonstrated by Nakazawa et al  for the same case of HML in fiber lasers. In their experiment, a portion of the output optical pulse train was extracted and converted into the electrical domain. The modulation driving RF signal was then derived from the converted electrical signal by simply using a high Q band-pass filter or a voltage controlled oscillator (VCO).
The two methods mentioned above are quite successful for the case of HML. However, for the case of RHML, if we want to apply the same stabilization scheme, we cannot use the above two techniques directly. This conclusion can be easily understood if we recall that for RHML, the repetition rate of the produced pulse train does not equal the modulator driving frequency. As a result, the phase error signal needed for feedback control cannot be generated by simply employing a regular RF mixer or high Q band-pass filter. To extend the regenerative mode-locking technique from HML to RHML, a modification on the phase error detection is therefore required and has already been demonstrated . In this paper, we present an alternative simplified scheme to solve the phase error detection problem of 4th RHML, with the aim to generate a long term stabilized pulse train operating at 40Gb/s.
2. Principle of operation and experimental results
The schematics of the proposed long term stabilized 4th RHML fiber laser is shown in Fig. 2. Similar to previous design , the fiber laser cavity is consisted of a Fabry-Perot band-pass filer with a FWHM band width of 3nm, an isolator, a segment of erbium dope fiber pumped at 980nm, a polarization controller, a LiNbO3 intensity modulator, and an output 90/10 coupler. When the modulator driving frequency is carefully adjusted, the fiber laser is able to generate a temporally stable and supermode noise suppressed 4th RHML pulse train operating at 40Gb/s with a pulse width of 2.5ps, indicating that optical part of the laser design is successful. To achieve long term stability, a modified regenerative feedback control circuit is constructed and the detailed setup is shown in the dashed box of Fig. 2. The key part involved is a LiNbO3 intensity modulator placed outside the fiber laser cavity. The driving signal of this modulator is a superposition of a static DC bias phase VB from a voltage source and a 10GHz sinusoidal modulation phase VRF sin(ωt) from a VCO. The modulator intensity transfer function T can be written as
By carefully adjusting the bias phase VB and the modulation strength VRF so that cos(VB)=0 and J 1(VRF)=0, the 10GHz and 20GHz components of the modulator intensity transfer function can be eliminated. Neglecting higher order harmonics, the LiNbO3 modulator can then be viewed as a 30GHz gate, which serves as a optical-electrical mixer to convert the 40GHz frequency of the input 40Gb/s pulse train down to 10GHz. This down-converted 10GHz signal is then sent to the RF port of a regular mixer to beat with the 10GHz VCO reference signal. As a result, the phase error signal between the mode-locked 40Gb/s pulse train and the driving 10GHz VCO signal is obtained at the IF port. Note that to guarantee that the detected phase error is forced to and is maintained at zero, an integrator is inserted between the VCO tuning port and the mixer IF port. To facilitate the adjustment of 4th RHML, an adder with a switch and a tunable RF delay-line are also used.
The operation of the modified regenerative fiber laser is as follows. First, the integrator output is disconnected from one of the input port of the adder so that the loop is open. Then we carefully adjust the DC bias voltage at the other input port of the adder and therefore changes the free oscillating frequency of the VCO. At some VCO frequency which satisfies the requirement of 4th RHML, a temporally stable 40Gb/s pulse train is generated. At that point, we then adjust the bias and the modulation strength of the external cavity LiNbO3 modulator so that a 30GHz modulation gate can be formed. The RF delay-line is also tuned so that the output of the RF mixer is roughly zero. With all these steps being done, we then close the loop by reconnecting the output of the integrator to one of the input port of the adder. Once the loop is closed, it is then expected that the feedback mechanism will work against the thermal drift and keep the laser long term stabilized.
In Fig. 3, we show the auto-correlator trace and the eye-diagram of the regeneratively RHML 40Gb/s pulse train. The eye diagram looks clean and the pulse width is measured to be 2.5ps (FWHM). To test how effective is the long term stabilization, the amplitude of the auto-correlation signal is measured as a function of time. The result is shown in Fig. 4 from which it can be seen that a stable operation as long as 8 hours has been achieved with very little amplitude fluctuations. Also shown in the inset of Fig. 4, we compare the auto-correlation trace of the pulse train before the test and after the test. These two auto-correlation traces look very similar. The small difference for the auto-correlation between the start point and the end point of the test is believed to be caused by the polarization evolution inside the cavity since none of the components used in the fiber laser cavity is polarization maintaining.
To further exam the thermal stability of the regenerative feedback controlled RHML, part of the fiber cavity is intentionally heated while the output of the pulse train is monitored using the auto-correlator. A comparison between the response from the regular mode-locked laser and the regeneratively mode-locked lasers is shown in Fig. 5. For the case of regular RHML, the pulse train is completely destroyed in 30 seconds as soon as the heater is turned on. However, for the case of regeneratively feedback controlled RHML, the pulse train is well preserved as long as the thermal equilibrium is re-achieved in about 15 minutes, indicating a significant improvement on the long term stability against environmental perturbations.
We have demonstrated a long term stabilized 4th order rational harmonic mode-locked fiber laser operating at 40Gb/s. With the help of the regenerative type feedback control technique based on two-step optical-electrical down-conversion, dramatic enhancement on the mode-locking thermal stability has been achieved.
References and Links
1. S. Kawanishi, “Ultra-high speed optical time-division-multiplexed transmission technology based on optical signal processing,” IEEE J. Quantum Electron. 34, 2064–2079, (1998) [CrossRef]
2. M. Horowitz, C.R. Menyuk, T.F. Carruthers, and I.N. Dulling III, “Theoretical and experimental study of harmonic mode-locked fiber lasers for optical communication systems,” J. Lightwave Technol. 18, 1565–1574, (2000) [CrossRef]
3. J. Li, P.A. Andrekson, and B. Bakhshi, “Direct generation of subpicosecond chirp-free pulses at 10GHz from a nonpolarization actively mode-locked fiber ring laser,” IEEE Photon. Technol. Lett. 12, 1150–1152, (2000) [CrossRef]
4. E. Yoshida and M. Nakazawa, “80 similar to 200GHz erbium doped fiber laser using a rational harmonic mode-locking technique,” Electron. Lett. 32, 1370–1372, (1996) [CrossRef]
5. G. Zhu, H. Chen, and N.K. Dutta, “Time domain analysis of a rational harmonic mode-locked ring fiber laser,” J. Appl. Phys. 90, 2143–2147, (2001) [CrossRef]
6. L.E. Nelson, D.J. Jones, K. Tamura, and E.P. Ippen, “Ultrashort pulse fiber ring lasers,” Appl. Phys. B 65, 277–294, (1997) [CrossRef]
7. X. Shan, D. Cleland, and A. Ellis, “Stabilizing erbium fiber soliton laser with pulse phase locking,” Electron. Lett. 28, 182–184, (1992) [CrossRef]
8. M. Nakazawa, E. Yoshida, and Y. Kimura, “Ultrastable harmonically and regeneratively mode-locked polarization maintaining erbium fiber ring laser,” Electron. Lett. 30, 1603–1605, (1994) [CrossRef]
9. G. Zhu, Q. Wang, H. Chen, H. Dong, and N.K. Dutta, “High-quality optical pulse train generation at 80Gb/s using a modified regenerative type mode-locked fiber laser,” IEEE J. Quantum Electron. 40, 721–725, (2004) [CrossRef]