Open Access
Add to BibSonomyAbstract
A high quality cavity-less pulse source, realized as a combination of linear pulse compression and self-phase-modulation (SPM) based regeneration is demonstrated and strictly characterized for the first time. The regenerated pulses, with 3.6 GHz repetition rate, are optimized through rigorous relative intensity-noise (RIN) measurement. Temporal intensity and chirp characterizations demonstrate that the pulses exhibit characteristic of low RIN, and are chirp- and pedestal-free. The cavity-less pulse source is further tested in a photonic-assisted analog-to-digital (ADC) configuration as the sampling source. A record result of more than 8 effective quantization bits at 202 MHz is demonstrated.
©2012 Optical Society of America
1. Introduction
Ultrafast pulse sources, mainly based on the mode locking technology, have a wide range of applications [1–5] such as optical sampling [2], optical high-speed communications [3], ultrafast spectroscopy [4], frequency metrology [5] etc. However, mode-locked laser pulse sources in practice suffer from inflexible, pre-set repetition-rates as well as a limited wavelength tuning range [6], owing to the inherent presence of the cavity. These properties considerably limit the highly desirable flexibility of the optical pulse sources in many practical applications, in contrast to a cavity-less scheme. A cavity-less optical pulse source is based on a single-pass structure, which makes the repetition-rate and operation wavelength significantly more flexible. Moreover, the timing jitter of a cavity-less source is solely dictated by the utilized radio-frequency (RF) synthesizer, as no optical cavity is present in its configuration. Cavity-less short-pulse sources can be realized by means of linear pulse compression [7], self-phase modulation (SPM)-based regeneration [8–10], and/or nonlinear pulse compression in a fiber optic parametric amplifier [11].
High quality and reliable short pulse sources have a great impact on the performance of many applications. Particularly for the photonic-assisted analog-to-digital conversion (ADC) applications, in which the pulse source is used for the sampling operation, the system performance is greatly limited by the amplitude noise and timing jitter of the optical sampling pulses [2, 12]. In this paper, a cavity less scheme with an exceptionally low relative intensity noise (RIN) and high pulse compression ratio is demonstrated. The pulse source is rigorously characterized as pedestal free and nearly chirp free. Moreover, the quality of the pulse source is further verified in a photonic-assisted ADC experiment, which reveals a record of effective number of bits (ENOB) exceeding 8 bits at 202 MHz. The results unmistakably demonstrate a high degree of pulse shape quality and stability for cavity-less sources, that is altogether uncompromised by the source flexibility.
2. SPM-based regenerated pulse
The experimental setup for the SPM-based pulse source is shown in Fig. 1 . The setup consists of two stages. The first stage is the linear compression stage, in which chirped pulses are compressed by a dispersive device, after cascaded phase and intensity modulation. On the other hand, the second stage is a SPM-based regenerator that employs Mamyshev’s principle of power-dependent spectrum broadening by SPM and offset filtering [8], leading to improved pulse quality characteristics.
Fig. 1 Experimental set up of pulse source. PM: phase modulator, MZM: Mach-Zehnder modulator, HP-EDFA: high-power EDFA, SMF: single mode fiber, HNLF: highly nonlinear fiber, FTF: flat-top filter, PS: phase shifter
In the first stage, a continuous wave (CW) laser at 1556.6 nm is used as a seeding source. The CW light is chirped by a phase modulator (PM), and then carved by a Mach-Zehnder modulator (MZM). The seeding RF 3.6 GHz harmonic is split into two paths: one of which drives amplitude carving by the MZM, whereas the other one imposes a strong chirp (after frequency quadrupling) by means of the PM. A phase shifter (PS) between the two paths is used to optimize the pulses’ chirp. Each of the modulators is followed by an erbium doped fiber amplifier (EDFA) compensating the inherent modulator insertion loss, and an optical bandpass filter limiting the amplified spontaneous emission (ASE) from the amplifiers. The chirped pulse train with 3.6 GHz repetition rate subsequently goes through single mode fiber (SMF) with anomalous dispersion leading to temporal compression of the underlying pulses. The pulse sources based on linear pulse compression technique have been studied previously. In [13] the pulses were optimized to be nearly pedestal free while possessing a narrow temporal width. In this paper, however, the additional purpose for the linear compression stage is to achieve a high peak power, which is further amplified by a high-power EDFA (HP-EDFA) (see Fig. 1), whose output power was maintained at 25.1 dBm in the experiments. The pulse peak power is designed such as to possess a large enough level which, subsequently, induces a significant SPM-spectral broadening in the second stage. We note that additional SPM broadening does occur in the subsequent span of SMF, whose main function is an additional (i.e. ultimate) pulse compression.
In the experiment, the compressed pulses with peak power of 31 W from stage 1 are launched into stage 2, i.e. the SPM-based regenerator, which contains a nonlinear device for spectrum broadening and a programmable flat-top filter (FTF) for regeneration. The pulses before FTF, captured with an optical sampling oscilloscope (OSO) [14], are shown in Fig. 2 as the blue curve and are characterized by a full width at half maximum (FWHM) of 2.5 ps. Finally, the output regenerated pulses, whose time and frequency domain correspond to the red solid curves in Fig. 2 and Fig. 3 , respectively, are pedestal free and possess an even narrower FWHM of 1.5 ps.
Fig. 2 The linear compressed pulse shape before (blue curve) and after (red curve) the regeneration offset flat-top filter (FTF), measured by optical sampling oscilloscope (OSO)
The nonlinear pulse compressor in the stage 2 comprises of a 259-m highly nonlinear fiber (HNLF), characterized by a zero-dispersion-wavelength of 1582 nm, and a dispersion slope of 0.0148 ps/km/nm2. The HNLF has a normal dispersion at the seeding wavelength of 1556.6nm, in order to ensure a smooth spectral broadening. Moreover, operation in the normal dispersion regime has an added benefit of suppressing the modulation instability as well as the associated noise generation [15]. The smoothly broadened spectrum as shown in Fig. 3 is generated with a 24.5-dBm input power into HNLF. The sliced spectrum after the FTF with different center wavelengths are also shown in Fig. 3.
The regenerated noise performance of the pulse source critically depends on the positioning and bandwidth of the FTF, due to the fact that the SPM spectral broadening is proportional to the pulse peak power [8]. A properly designed regenerator induces a uniform spectral broadening [16], while it simultaneously increases the signal-to-noise ratio by means of intensity to phase fluctuation conversion, as well as displacement of the deterministic waveform, away from the band of the noise-polluted seeding pulse. Indeed, a judicious filter offsetting results in the removal of the intensity noise associated with the pulses’ lower-power region. In addition, note that offsetting the filter wavelength away from the edge of the SPM broadened spectrum, works as to further reduce the amplitude fluctuations at the pulse peak. The energy transfer function in [16], in particular, corroborates the mentioned noise suppression in ‘space’ and ‘mark’ levels for optimized offset filtering. Besides, it should be emphasized that the peak power of the compressed pulses into the stage 2 has also been optimized [10]. In the experiment, this was implemented by means of adjusting the PM induced chirping, as well as by a meticulous choice of the SMF length, and monitoring the noise performance of the regenerated pulses.
3. Pulse source RIN characterization
The pulse source RIN was characterized at a constant optical power as follows. First, the intensity noise was obtained by subtracting the background level from the measurement of an electrical spectrum analyzer (ESA), and calibrated to the shot-noise level. Following that, the RIN was obtained as a ratio of the calibrated intensity noise and the average power of the regenerated pulses.
The noise performance of the pulse trains before and after the stage 2 (i.e. the SPM regenerator), as well as that of the CW seeding laser source are shown in Fig. 4 . The offset filter with a center wavelength of 1564.6 nm was found to regenerate pulses with best RIN performance, as shown in Fig. 4. In the optimal case (spectrum in the red solid curve in Fig. 3) the output power was 5 dBm.
Fig. 4 RIN measurement of CW laser source, pulse before and after regeneration at power level of 5 dBm
In order to compare the noise performance of the regenerated pulses with the shot noise limited level, the RIN measurement and comparison were evaluated at the power level of 5 dBm. The existence of an optimal FTF position (with respect to noise performance) was confirmed by the RIN measurement, and was located away from both the seeding wavelength and the edge wavelength of the SPM spectrum, as argued in [8]. For a reference, the original CW laser source was characterized by a shot noise limited performance at high frequencies (> 1 MHz) at the −157 dBc/Hz level. The successful pulse regeneration is demonstrated by the output pulses having an improved RIN performance, compared to the pulse train prior to regeneration: Indeed, the RIN level of regenerated pulses with 5 dBm average power level is characterized at −151 dBc/Hz at frequencies higher than 1 MHz. In effect, this corresponds to the RIN level improvement of more than 16 dB at low frequencies, after regeneration. In fact, a pulse source at 1543 nm with an even lower RIN has been demonstrated [17], based on this method.
In order to confirm the significant effect of the filter position on the noise performance, the bandwidth of FTF was fixed at 1.32 THz and the center wavelength of FTF was tuned. Note that for the FTF center wavelength of 1570 nm (i.e. the edge of the SPM spectrum), the output power of the regenerated pulses dropped to 3 dBm. Consequently, the RIN evaluation was performed at the minimal obtainable level (i.e. 3 dBm), in order to ensure the consistency of the comparison and is shown in Fig. 5 . The center wavelength of the FTF with best noise performance was obtained at 1564.6 nm, which corresponds to the red solid curve in Fig. 3 (and the regenerated noise performance in Fig. 4). As argued above, the observed behavior is a direct consequence of the SPM-mediated intensity to phase conversion and overlap (or displacement) of the output pulse spectrum and the input spectral window that contains the noise from the previous stage. As the center wavelength of the FTF approaches the seeding wavelength 1556.6 nm, the noise performance drops quickly, as the output spectral window starts overlapping with the input one. On the other hand, as the FTF center wavelength approaches the edge of the generated SPM spectrum at 1570 nm, the amplitude fluctuations of the input pulses degrade the noise performance of the regenerated pulses by means of noise-induced stochastically fluctuating SPM broadening towards the end of the SPM generated spectral region [8].
Fig. 5 RIN and SNR performance of the regenerated pulses at different center wavelengths at the power level of 3 dBm
4. Temporal pulse characterization
The intensity and phase characterization of the regenerated pulse train is achieved by a “linear” FROG method [18, 19], as shown in Fig. 6 . In the linear FROG technique, a pulse is gated by itself rather than by nonlinearities. Subsequently, the gated part is frequency analyzed to generate a spectrogram, while the complete pulse information is retrieved from a numerical analysis of the obtained spectrogram. In our experimental realization shown in Fig. 6, the pulse train is attenuated and split into two paths: one of the paths is fed directly into a 40 GHz MZM, whereas the other one is passed through a programmable optical delay line, and is photo-detected. The thus-obtained RF signal is used to electrically drive the MZM, achieving the self-gating functionality.
Fig. 6 Experimental setup of linear FROG. VOA: variable optical attenuator, OSA: optical spectrum analyzer, PD: photodetector, PC: polarization controller
The intensity and phase information of the regenerated pulses in the time domain was retrieved from the FROG and is shown in Fig. 7(a) , whilst the frequency domain information is shown in Fig. 7(b). Note, however, that the chromatic dispersion associated with the fiber tail of the VOA, the coupler, the optical delay line, PC and the input fiber into the MZM (see Fig. 6) (around 10 m in total) ought to be subtracted from FROG-retrieved results (the non-calibrated curves in Fig. 7). In consequence, the comparisons of the characterizations before and after the chromatic dispersion chirp subtraction are also shown in Fig. 7(a), 7(b). Note that after calibration, the phase fluctuation at the central part of the pulse is less than 0.3 rad in the time domain, and less than 0.1 rad in the frequency domain, which clearly attests as to the transform limited characteristics of the output pulses. Consequently, the pulse FWHM after re-calibration amounts to 0.8 ps. The validity of this calibration is further verified with a pulse source operated at 1569 nm (FWHM = 1 ps) [20], which was also nearly chirp free (including the 10 m of standard SMF re-calibration). In addition, the pulse spectrum was independently measured by an OSA and compared with the FROG spectrum in Fig. 7(c), which further emphasizes the accuracy of the FROG result.
Fig. 7 Pulse characterization based on the linear FROG. Time domain (a) and frequency domain (b) comparisons before and after recalibration with respect to the SMF pigtails in the FROG setup, (c) comparison of the pulse spectra obtained by the OSA and the FROG.
Besides the FROG measurement, the pulse shape was also measured by an OSO and an autocorrelator for comparison, The latter measurement results are shown as black and magenta curves in Fig. 8 , respectively. As seen in Fig. 8, both of those characterizations imply wider pulse widths compared with that of the FROG. However it ought to be emphasized that the resolution of the OSO was 0.5 THz (~0.8 ps), as compared with the 0.2 ps resolution for FROG. Furthermore, the autocorrelation of pulses are wider than the pulse itself. As far as the autocorrelator measurement, a broadening factor should be considered in calculating the pulse width. For instance, assuming Gaussian pulse shape for the regenerated pulses, a factor of should be included in the characterization, which implies a FWHM of 1.7 ps for the regenerated pulses measured by the autocorrelator.
5. Regenerated pulses in linear photonic-assisted sampling
In order to further validate the pulse integrity, the constructed cavity-less source was used as the sampling gate in a photonic-assisted ADC. High precision ADC requires exceedingly high quality sampling pulses, in terms of the timing jitter, as well as the amplitude noise. In particular, the pulse properties become increasingly important for wider sampling and/or ADC bandwidths. Compared with electronic gates, optical pulses provide low timing jitter at high sampling bandwidth [2, 12]. The timing jitter of the constructed pulse source was measured by an ESA and was calculated by integrating the measured phase noise from 50 kHz all the way to 1.8 GHz (i.e. the Nyquist bandwidth). The measured timing jitter of this regenerated pulse source was found to be less than 40 fs. It must be emphasized, however, that the obtained phase noise is comparable with the background noise of ESA above 1 MHz, clearly implying that the obtained result of 40 fs represents a significant overestimate of the actual source timing jitter characteristic.
The experimental setup for linear sampling is shown in Fig. 9 . The pulse train with 3.6 GHz repetition rate is launched into a MZM as a sampling gate. The sampled pulses after the MZM were low-pass filtered and amplified, only to be detected by a 3.6-GSample/s electronic ADC with a 12-bit resolution and 8.6-bit ENOB property.
The ENOB for the pulse source was characterized at 8.06 bits, and the FFT spectrum of the sampled data is shown in Fig. 10 . The signal digitization process was aided by a look-up-table based algorithm [21], mitigating the nonlinear distortions originating from the nonlinear response of the MZM as well as the electrical amplifier before the electrical ADC, in order to focus solely on the sampling pulse source properties. For a comparison, the back-to-back ENOB characteristics of the RF source as well as the CW light were also measured, and were found to correspond to 8.6 bits and 8.25 bits, respectively, further corroborating the outstanding characteristics of the constructed cavity-less source.
6. Conclusion
We have presented a rigorous and thorough characterization of a SPM-based regenerated cavity-less source, characterized by a low RIN, nearly chirp free features, and less than 40 fs timing jitter stability. We have provided an optimization procedure as to the HNLF parameters and the filter positioning, leading to the optimal source characteristics, after regeneration. The pulse source was rigorously characterized in both time and frequency domains, while the low RIN performance was achieved by optimizing the filter offset in the Mamyshev’s regenerator. As a final demonstration of the obtained pulse train quality, the source was used for a photonic-assisted AD conversion which resulted in an experimental demonstration of higher than 8.0 bits of ENOB performance, thus, fully validating the source applicability to high speed, high precision ADC applications.
References and links
1. H. A. Haus, “Mode-locking of lasers,” IEEE J. Sel. Top. Quantum Electron. 6(6), 1173–1185 (2000). [CrossRef]
2. P. W. Juodawlkis, J. C. Twichell, G. E. Betts, J. J. Hargreaves, R. D. Younger, J. L. Wasserman, F. J. O'Donnell, K. G. Ray, and R. C. Williamson, “Optically sampled analog-to-digital converter,” IEEE Trans. Microw. Theory Tech. 49(10), 1840–1853 (2001). [CrossRef]
3. J. A. Salehi, A. M. Weiner, and J. P. Heritage, “Coherent ultrashort light pulse code-division multiple access communication systems,” J. Lightwave Technol. 8(3), 478–491 (1990). [CrossRef]
4. J. Shah, “Ultrafast luminescence spectroscopy using sum frequency generation,” IEEE J. Quantum Electron. 24(2), 276–288 (1988). [CrossRef]
5. H. R. Telle, G. Steinmeyer, A. E. Dunlop, J. Stenger, D. H. Sutter, and U. Keller, “Carrier-envelope offset phase control: a novel concept for absolute optical frequency measurement and ultrashort pulse generation,” App. Phys,” Appl. Phys. B. 69(4), 327–332 (1999). [CrossRef]
6. S. A. Diddams, “The evolving optical frequency comb,” J. Opt. Soc. Am. B 27(11), B51–B62 (2010). [CrossRef]
7. T. Kobayashi, H. Yao, K. Amano, Y. Fukushima, A. Morimoto, and T. Sueta, “Optical pulse compression using high-frequency electrooptic phase modulation,” IEEE J. Quantum Electron. 24(2), 382–387 (1988). [CrossRef]
8. P. V. Mamyshev, “All-optical data regeneration based on self phase modulation effect,” 24th European Conference on Optical Communication (ECOC)1998.
9. H. Hu, H. C. H. Mulvad, C. Peucheret, M. Galili, A. Clausen, P. Jeppesen, and L. K. Oxenløwe, “10 GHz pulse source for 640 Gbit/s OTDM based on phase modulator and self-phase modulation,” Opt. Express 19(26), B343–B349 (2011). [CrossRef] [PubMed]
10. P. Johannisson and M. Karlsson, “Characterization of a self-phase-modulation-based all-optical regeneration system,” IEEE Photon. Technol. Lett. 17(12), 2667–2669 (2005). [CrossRef]
11. A. O. J. Wiberg, C. S. Brès, B. P.-P. Kuo, J. X. Zhao, N. Alic, and S. Radic, “Pedestal free pulse source for high data rate optical time-division multiplexing based on fiber-optical parametric processes,” IEEE J. Quantum Electron. 45(11), 1325–1330 (2009). [CrossRef]
12. G. C. Valley, “Photonic analog-to-digital converters,” Opt. Express 15(5), 1955–1982 (2007). [CrossRef] [PubMed]
13. A. O. J. Wiberg, Z. Tong, L. Liu, J. L. Ponsetto, V. Ataie, E. Myslivets, N. Alic, and S. Radic, “Demonstration of 40 GHz analog-to-digital conversion using copy-and-sample-all parametric processing,” Optical Fiber Communications Conference2012 Technical Digest, paper OW3C.2.
14. M. Westlund, H. Sunnerud, M. Karlsson, and P. A. Andrekson, “Software-synchronized all-optical sampling for fiber communication systems,” J. Lightwave Technol. 23(3), 1088–1099 (2005). [CrossRef]
15. E. Myslivets, B. P. P. Kuo, N. Alic, and S. Radic, “Generation of wideband frequency combs by continuous-wave seeding of multistage mixers with synthesized dispersion,” Opt. Express 20(3), 3331–3344 (2012). [CrossRef] [PubMed]
16. M. Matsumoto, “Fiber-based all-optical signal regeneration,” IEEE J. Sel. Top. Quantum Electron. 18(2), 738–752 (2012). [CrossRef]
17. Z. Tong, L. Liu, A. O. J. Wiberg, N. Alic, and S. Radic, “Self phase modulation based low-noise, cavity-less short pulse source for photonic-assisted ADC conversion,” 38th European Conference on Optical Communication (ECOC)2012, paper Th.1.F.5.
18. K. T. Vu, A. Malinowski, M. A. F. Roelens, M. Iben, P. Petropoulos, and D. J. Richardson, “Full characterization of low-power picosecond pulses from a gain-switched diode laser using electrooptic modulation-based linear FROG,” IEEE Photon. Technol. Lett. 20(7), 505–507 (2008). [CrossRef]
19. C. Dorrer and I. Kang, “Simultaneous temporal characterization of telecommunication optical pulses and modulators by use of spectrograms,” Opt. Lett. 27(15), 1315–1317 (2002). [CrossRef] [PubMed]
20. L. Liu, Z. Tong, A. O. J. Wiberg, N. Alic, and S. Radic, “Generation and characterization of self-phase modulation based cavity-less pulse source,” Frontiers in Optics/Laser Science XXVIII (Fio/Ls)2012, paper FW5D.2.
21. V. Ataie, E. Myslivets, A. O. J. Wiberg, B. P.-P. Kuo, N. Alic, and S. Radic, “Linearization of copy-and-sample photonics analog to digital converter based on a multidimensional look-up table,” Optical Fiber Communications Conference2012 Technical Digest, paper OM2B. 6.
