Open Access
Add to BibSonomyAbstract
We demonstrate active shaping of the driving electrical pulses to a laser diode in order to compensate for the pulse shaping effects of gain saturation in an Yb doped fiber amplifier cascade and to allow the generation of user defined customized output pulse shapes. In particular we demonstrate the generation of square output pulses, which have the potential to significantly increase the maximum pulse energy extractable from an amplifier before the peak power reaches the threshold for SRS, and for high efficiency frequency conversion.
©2006 Optical Society of America
1. Introduction
High peak power pulsed lasers capable of delivering pulses with nanosecond durations represent useful sources for a growing number of advanced industrial applications, such as marking and material processing and are the subject of considerable research [1–4]. This area is currently dominated by Q-switched solid state lasers, operating at 1064 nm or its harmonics, or by excimer lasers. Fiber based MOPA systems incorporating a directly-modulated semiconductor diode as a seed laser and rare-earth doped fiber [5] appear particularly attractive in this context providing greater flexibility in terms of the critical pulse parameters, such as pulse duration and peak power than say a simple Q-switched laser system [6], as well as much lower cost than other systems. Nanosecond fiber amplifiers have produced pulses with peak powers of tens of kilowatts with single mode and much higher with multimode output [7–9]. To achieve the high pulse energies (mJ regime) and high peak powers (10kW regime) required using this scheme usually requires gains in excess of 40–50dB due to the relatively modest peak powers derivable from the diode seed. Given this level of gain and the relatively modest saturation energies associated with doped fibers, significant pulse reshaping due to depletion of the inversion over the timescale of the pulse can take place within the amplifier chain [10]. The resulting level of pulse reshaping will affect the output pulse duration and peak power in a non-linear manner. This has consequences for applications where pulse duration or shape is critical. It also lowers the limit on pulse energy before the onset of Stimulated Raman Scattering (SRS) which becomes significant above a critical peak power for a given amplifier chain length [11]. Effectively, SRS acts as a cap on peak power at the signal wavelength, by efficiently scattering the light to other wavelengths. If the pulse is significantly shortened due to reshaping, this critical peak power will be reached at lower pulse energy. This is significant for applications and processes that are wavelength dependent, such as frequency-conversion [12], where it is desirable to maintain the maximum power level (below the SRS threshold) for as much of the duration of the pulse as possible to maximize conversion efficiency.
Techniques which both overcome the impact of pulse amplifier gain saturation and allow control of the pulse shape at the system output are thus highly desirable. One possible approach is to actively control the shape of the input pulse so as to pre-compensate for the effects of the gain saturation induced pulse shaping in order to obtain a particular desired pulse form at the system output. In this paper we demonstrate the development of such a technique for use within a diode-seeded MOPA amplifier chain. Using active pulse shaping at the amplifier input we show that it is possible to generate a range of user defined pulse shapes at the system output in a very straightforward fashion. As examples we demonstrate the generation of high energy square optical pulses (ideal for frequency conversion processes, and for simply maximizing the output pulse energy for a given pulse duration and fixed peak power), as well as the generation of both triangular and two-step profiled pulses which are of potential interest for material processing applications.
2. Fiber MOPA system
Our experiments were conducted on a fully fiberised MOPA system as shown in Fig.1 and described as follows. A commercially packaged, fiber pigtailed diode laser operating at a wavelength of 1077nm (bandwidth 3nm) is used as a nanosecond pulsed source for a two stage Yb3+ doped, cladding pumped all-fiber amplifier chain [5]. Both the pre-amplifier and post-amplifier stages are fabricated in GT-Wave technology [13] and are pumped using broad striped diodes operating at a wavelength of 915nm. The power amplifier is a LMA fiber with M2<2. A passive delivery fiber of 5 meters length is spliced to the output of the power amplifier. The MOPA is configured to give >40dB gain, with up to 30dB in the first stage.
The diode was driven at a repetition rate of 20 kHz using the current amplified output of an Arbitrary Waveform Generator (AWG). At maximum drive current the diode had a cw output power of ~330mW. The achievable average power of the power amplifier is greater than 10W, limited by pump power. The critical instantaneous power for significant SRS in this system was observed experimentally to be ~8 kW, limited by SRS in the pump delivery fiber.
A Matlab code controls the AWG, amplifier pumps and an oscilloscope which is used to acquire the output waveform. The AWG can generate arbitrary pulse shapes with a temporal resolution of 4ns with 16 bit amplitude resolution (ranging between 0.1 and 1V). The AWG signal is then fed into a voltage to current converter which drives the diode with an instantaneous current to the diode proportional (above some threshold level) to the applied instantaneous voltage. The bandwidth of the current driver is 100 MHz and that the maximum current that can be supplied by the driver is approximately ten-times the seed laser threshold. The electrical drive system thus provides a usable dynamic range for seed pulse shaping of around 10dB. Note that it is important to turn off the amplifiers when inputting a new pulse shape (i.e. between iterations of the optimization process described later) to avoid any potential for catastrophic self pulsation as can happen if the amplifier cascade is operated without an input signal.
Fig. 2. (a) square signal from AWG and corresponding optical output of diode, (b) corresponding optical output from MOPA at various pump power levels. The ringing on the pulse is due to non-ideal impedance matching of the current driver and laser and could be largely eliminated with improved design.
The output pulse shape is measured using a fast InGaAs photodetector and the temporal profile transferred to the computer. Figure 2 shows the pulse shaping that occurs in the MOPA to an input square pulse at maximum diode current at various levels of amplification. Similar reshaping of a square input pulse was observed and modeled in a similar system in reference [5]. An accurate numerical model describing the pulse evolution through the amplifier chain was developed. The model is based on the method of lines and takes into account the distribution of gain and signal along the fiber length. It can be used to simulate and predict the effect of fiber design and pumping conditions on the pulse shape, pulse energy and peak power.
3. Adaptive pulse shape control
In order to optimize the shape of the diode driving current pulse, the measured output pulse from the amplifier system is compared to a user-specified target pulse. The absolute values of the differences in instantaneous power between the target pulse and the normalized actual pulse shape are measured at 4ns intervals over the pulse duration and summed to produce a fitness parameter (F) for the pulse. The input pulse is then changed adaptively to minimize the fitness parameter. The input pulses are created from a small number of parametrically defined curves. It is important to minimize the number of parameters required to define the input pulses in order to minimize the number of iterations needed to obtain a satisfactory approximation to the desired output pulse form. For example, to generate the square pulses described herein, we found that it was possible to get good results using an exponentially ramped input pulse specified by the height of the starting point of the ramp (L) the degree of the rise (α) and the pulse width T (which in these experiments we kept fixed and did not seek to optimize), giving a shape of the form:
In order to get square output pulses in this instance then it was necessary to optimize just two parameters (L and α). The optimization method that we used was simulated annealing [14] according to which the system is moved towards the state which has lowest value of F through small iterative step changes in the parameters to be optimized.
Fig.3. (a) input electrical signal (dotted) and diode output pulses, (b) target pulse (dotted) and MOPA output (solid), comparison pulses (dashed) produced with square driving pulse.
The output pulse resulting from an optimized input pulse is shown in Fig. 3. The output pulse has an energy of 0.2mJ, with a peak power of 2.6 kW. Achieving the optimized pulse form required ~15 iterations. Figure 3(a) shows the input electrical signal (dotted) and output optical pulse from the diode. The pulse from the diode is significantly different from the input from the AWG, due to the response of the drive electronics and the diode. This difference is not necessarily important for the optimization process, except where the drive electronics introduces transients (see Fig. 2(a)) on a timescale shorter than the resolution of the AWG, and which therefore cannot be compensated by the active reshaping. Figure 3(b) shows the output pulses form the MOPA. The output of the MOPA is close to the targeted square pulse. Clearly, the ripple on the diode pulses, which is too fast to be fully compensated for by the AWG, reduces the quality of the output pulses to some extent; however relatively good square pulse quality is achieved. Note that even better pulse quality could in principle be achieved by incorporating a greater number of free parameters to define the input pulse shape, albeit at increased cost in terms of number of iterations to achieve an optimized output pulse form.
For comparison, in Fig. 3(b) we show two output pulses obtained by driving the diode with a square pulse, with different levels of amplification in the final amplifier stage. One comparison pulse has approximately the same pulse energy as the shaped pulse, but the peak power (4.3 kW) is 64% greater. The other comparison pulse has approximately the same peak power as the reshaped pulse, but the pulse energy (0.13 mJ) is only 67% that of the shaped pulse. The pulse duration (from switch on to switch off) of the shaped input pulse is not identical to the pulses produced with square input pulses. This is due to the response time of the electronics of the diode drive circuit, which respond within ~1 ns to a full height step input, but which have up to a 20ns delay for smaller step inputs. This could easily have been accommodated for by inclusion of an additional optimization parameter (the drive pulse length). It should be noted that these nanosecond responses are not a function of the diode or its packaging. It can be seen that reshaping of the pulse allows considerable improvement in pulse quality. The limitation on achieving square pulses at higher energy appears to be the restricted (~10dB) usable dynamic range that is obtainable by direct modulation of the diode. (As the pulse energy increases, the reshaping that needs to be compensated requires greater and greater changes in amplitude across the seed pulse shape but if the laser is operated too close to threshold then the increased noise associated with this starts to impact the output pulse shape quality and stability).
As a demonstration of the versatility of the system, two other customized output pulse shapes were produced: a triangular pulse and a two-step pulse with durations of 100ns and 200ns, respectively. In order to make modification to the program simple, and to limit the number of iterations required, we chose to restrict the number of pulse optimization parameters to just two for both pulse forms. The other parameters used to specify the input pulses were fixed with values chosen based on experience with simpler pulse forms. Plots on the left of Fig. 4 illustrate the relatively simple input pulse shapes used.
For the triangular pulse case, the input consisted of three sections: an exponential rising edge, a flat top and a linearly decreasing tail. The first section of the pulse is shaped as defined by Eq. (1), but in this case the initial height L is fixed (at an AWG voltage of 0.1V, just above the threshold level), and the variable parameters are the exponential factor α and the duration T. The second section is constant at maximum voltage. The total duration of the first and second sections is fixed, so that increasing T decreases the duration of section two. The third section is a fixed ramp down to a voltage of ~0.2V. Thus, only 2 independent variables, α and T are varied during the optimization process.
Fig. 4. Applied signal (dotted), diode output (solid green), target pulses (dashed) and MOPA output (solid blue) for triangular and 2-step pulses
The two-step pulse could be considered as two consecutive square pulses. This seed pulse in this instance was specified by 8 parameters. Based on experience with shaping square optical pulses, the leading rising edge was specified by a straight line starting from 0.3V finishing at 1V over a duration of about 50ns, with no variable parameters. The second part of the pulse had the shape given by Eq. (1). The two parameters which were optimized by the adaptive process were the starting height L and the exponential factor α of this second rising feature, just as when the target shape was a single square pulse. The optimized input and corresponding output pulses are shown on the right of Fig. 4. The quality of the pulses is highly impressive given the limited number of free parameters.
4. Conclusion
We have demonstrated an effective, simple and versatile method for controlling pulse shape in diode-seeded fiber amplifier cascades that should be of significant value in enhancing the use of fiber based MOPA systems for a range of industrial and scientific applications. The primary restriction on the range of user-definable pulse shapes that can be achieved is limited to some extent by the usable dynamic range of laser output and bandwidth of the drive electronics. However, useful complex pulse forms can be obtained by optimization of a surprisingly small number of drive parameters. It is to be appreciated though that the technique can be readily extended to incorporate a greater number of degrees of freedom, albeit at increased cost in terms of the required optimization cycles. Note that the use of an external method of shaping the pulses between the diode and the amplifiers would allow us to operate the diode well above threshold for the duration of the pulse whilst allowing much higher dynamic ranges for pulse shape control – albeit at increased cost. Further extension of this work in both respects is to be anticipated in due course.
References and Links
1. M. von Allmen and A. Blatter, Laser-Beam Interactions with Materials (Springer, 1995). [CrossRef]
2. A. Misra, A. Mitra, and R.K. Thareja, “Diagnostics of laser ablated plasmas using fast photograph,” Appl. Phys. Lett. 74, 929–931, (1999). [CrossRef]
3. B. Toftmann, J. Schou, T.N. Hansen, and J.G. Lunney, “Angular Distribition of Electron Temperature and Density in a Laser-Ablation Plume,” Phys. Rev. Let. 84, 3998–4001, (2000). [CrossRef]
4. T. Chen and R.B. Darling, “Parametric studies on pulsed near ultraviolet frequency tripled Nd:YAG laser micromachining of sapphire and silicon,” J. Mater. Process. Technol. 169, 214–218 (2005). [CrossRef]
5. F. Ghiringhelli, K. Vysniauskas, L. M. B. Hickey, M. N. Zervas, A. Malinowski, and D. J. Richardson, in Proceedings of CLEO Europe, Munich ICM, International Congress Centre, Munich, Germany12–17 June 2005.
6. A. Piper, A. Malinowski, K. Furusawa, and D.J. Richardson, “High-power high-brightness mJ Q-switched ytterbium-doped fibre laser,” Electron. Lett. 40, 928–929, (2004). [CrossRef]
7. J. Limpert, S. Hofer, A. Liem, H. Zellmer, A. Tunnermann, S. Knoke, and H. Voelckel, “100-W average-power, high-energy nanosecond fiber amplifier,” Appl. Phys. B 75, 477–479 (2002). [CrossRef]
8. M.Y. Cheng, Y.C. Chang, A. Galvanauskas, P. Mamidipudi, R. Changkakoti, and P. Gatchell, “High-energy and high-peak-power nanosecond pulse generation with beam quality control in 200-mu m core highly multimode Yb-doped fiber amplifiers,” Opt. Lett. 30, 358–360, (2005). [CrossRef] [PubMed]
9. C. Codemard, C. Farrell, P. Dupriez, V. Philippov, J.K. Sahu, and J. Nilsson, “Millijoule, high-peak power, narrow-linewidth, sub-hundred nanosecond pulsed fibre Master-Oscillator Power-Amplifier at 1.55 mu m,” C. R. Phys. 7, 170–176, (2006). [CrossRef]
10. L. M. Franz and J. S. Nodvik, “Theory of Pulse Propagation in a Laser Amplifier,” J. Appl. Phys. 34, 2346–2349 (1963). [CrossRef]
11. R.G. Smith “Optical power handling capacity of low loss optical fibers as determined by stimulated Raman and Brillouin scattering,” Appl. Opt. 11, 2489–2494, (1972). [CrossRef] [PubMed]
12. A.P. Liu, M.A. Norsen, and R.D. Mead, “60-W green output by frequency doubling of a polarized Yb-doped fiber laser,” Opt. Lett. 30, 67–69, (2005). [CrossRef] [PubMed]
13. A. B. Grudinin, J. Nilsson, and P. W. Turner, in Proceedings of CLEO Europe, Nice, France. 10–15th Sept 2000.
14. S. Kirkpatrick, C. D. Gelatt, Jr., and M. P. Vecchi, “Optimization by Simulated Annealing,” Science 220, 671–680, (1983). [CrossRef] [PubMed]
