We demonstrate tuneable optoelectronic bandpass filtering of an ASK modulated subcarrier data stream, using a two-section laser. Gain enhancement of 15dB is measured when the laser is locked to a 697MHz and 1100MHz carrier frequency.
© 2003 Optical Society of America
Self-pulsing lasers (SPL’s) have received much attention over the last thirty years . The mode of self-pulsation in this type of laser, as displayed in Fig. 1, is through an in-built saturable absorber in the optical cavity. Division between the two contacts is through an etched gap, the depth of which creates a known resistance between each section, thus allowing individual biasing of the two regions . The longer section can now be biased well into gain, and the shorter section can be unbiased or reverse-biased, the shorter section termed a saturable absorber. When there is a large enough difference in the carrier lifetimes between the two sections, self-pulsing may take place . It is well known that SPL’s can be effectively injection-locked by a superimposed RF signal , with a straightforward extension to digitally modulated signals , .
Over the last decade, subcarrier multiplexing (SCM) has become a popular mode of data transmission utilizing the high bandwidth of fibre , . However, to our knowledge, the use of a two-section SPL as an optoelectronic band-pass filter in this type of system has not been proposed and demonstrated before. In this paper, following on work on Sharp CD lasers , we show how injected RF signals in the 500MHz to 1.3GHz range can be effectively bandpass filtered by appropriate modification of the self-pulsation frequency. Through variation of the injected current to the gain region, or the reverse bias voltage on the absorber, the self-pulsate frequency and hence the gain-peak can be tuned. We show how this technique can therefore be used to bandpass filter the Amplitude Shift Keying (ASK) modulated signals.
The model used in this paper is based on works by Avrutin , Egan et al. , and the dimensionless rate equations were developed by Carr and Erneux . The rate equations used to describe the dynamics of the carrier densities and the photon density in the two regions, namely the gain and absorber regions are given in Eqs. (1–3) respectively.
The symbols used in the model are, jg,a , are the currents applied to the absorber and gain regions, e is the electron charge, d is the active region thickness, are the carrier lifetimes in both regions, ng,a , are the carrier densities in the gain and absorber regions, v is the group velocity of light in the cavity, B is the bimolecular coefficient, β is the spontaneous emission factor. Fs (t) is the Langevin noise term, α0 is the cavity loss and g, α are the gain and loss (absorption) in the gain and absorbtion regions, and q is taken as 2. S is the mean photon number in the cavity. The gain and absorption values are predicted using the linear gain approximation as first suggested by Dixon and Joyce .
ag,a are the differential gain and absorbtion parameters and noa,g are the transparency current densities of the respective regions. The dimensionless rate equations developed by Carr and Erneux are given by
Where I is the photon intensity, D1,2 are the carrier dynamics in both the gain and absorbing region respectively. A1,2 are the currents applied to both sections, η s is the spontaneous emission factor, a1,2 are the cross diffusion coefficients, these are made zero as cross diffusion out of the active region is not important in this case; bR1,2 is the radiative recombination rate and bA1,2 is the Auger recombination. Equations. (1–3) were then normalized and regrouped for agreement and substitution with Eqs. (6–8). Sa was the taken to be equal to Sg and the average S was used in Eqs. (1) and (2), thus making Sg and Sa unity. The following results were obtained after running the numerical model. Changing τ a in the model and keeping τ g constant, it could be observed that the current threshold increased, as can be seen from Fig. 1.
The results from running τ a ~ 185ps (1/Aa in Fig. 1) is almost in full agreement with our experimental findings for particular biasing conditions, i.e. Increasing the reverse bias to the absorber section, reduces the carrier lifetime in the absorber region due to an increased sweep out rate. This increases the probability of pulsations, as can be seen from Fig. 2. Therefore showing the best way to obtain pulsations is to both shorten the lifetime in the absorber region, by either reverse bias or ion implantation , and also to increase the ratio between the differential gain and absorbtion, thus minimising the absorber carrier density in between pulsations. Figure 3 shows the locking ability of the two-section laser. Here an external sine wave having a known frequency is superimposed onto the DC bias applied to the gain section, and with a slight change in the DC bias applied to the gain section, injection locking does take place. This locking procedure is a well-known characteristic of the laser , but the novelty is using this device to distinguish between desired and unwanted frequencies . The authors were only interested in pulsation and stable regions, but the model could also predict regions of bistability.
Also it can be seen from Fig. 4 that injection locking can be achieved even if the incoming frequency is different from the natural self-pulsation of the laser as long as the power of the incoming signal is sufficiently increased to compensate for the difference in frequency.
Figure 4 displays ΔFrequency (Ghz) against m, where m is the power needed for injection locking to take place, in arbitrary units, and ΔFrequency is the difference between the natural pulsations and the frequency of the incoming sine wave. This showed a linear dependence as described by Alders Law
The experimental set-up is shown in Fig. 5. A two-section laser with a GRINSCH structure and an emission wavelength of 813nm was used. A standard 200mA current source (ILX-3207; Ig in Fig. 1) was used to bias the gain-section of the laser. The absorber was reversed-biased using a high-accuracy voltage source (Data Precision 8200) with a voltage of -6.5V applied through a 2kΩ resistor.
Variation of this bias voltage can be used to change the lasing wavelength, for WDM applications. We measured the laser threshold to be 31mA at 300K. Two RF signal generators covering the ranges of 300MHz to 2000MHz were used. Injected power levels were investigated, and carrier signal breakthrough eliminated by verifying that the SPL detected output signal level did not vary with injected RF signal strength. The output from the laser was viewed electrically from the absorber bias-tee.
This was verified optically using a 3GHz bandwidth Si photodiode (Hamamatsu S7911). After amplification, the detected signal was passed to a spectrum analyser so that the injected RF and self-pulsation signals could be observed.
Two carrier frequencies at 697MHz and 1100MHz were generated using the RF sources, and together with the self-pulsating frequency of the laser were observed as shown in Fig. 6a. The input power levels of the carrier signals were kept equal at -48dBm (into 50Ω). With simple adjustment of the SPL electrical bias, selective locking onto either the 697MHz or 1100MHz signals was achieved, as displayed in Figs. 6b and 6c. In the case of the 697MHz signal, 15.1dB power gain was achieved, with the higher-frequency (unlocked) signal experiencing negligible enhancement. This preliminary result shows for the first time the excellent functionality, in the novel application as a tuneable filter.
It can be seen that synchronisation to the higher frequency shows increased noise, albeit still at a low level. This is believed to be due to the laser reaching its self-pulsation range. We then carried out tests with an ASK modulated data stream using an Anritsu ME522A pattern generator, which was used to produce a 27-1 PRBS data stream at 60Mb/s, which ASK modulated RF carriers at both at 850MHz and 1100MHz. The modulated data was introduced into the gain region of the two-section laser via a standard bias-tee. This data was amplified by at least 15dB when the self-pulsations of the laser synchronised with either of the modulated carriers. The resultant modulated carrier and data are shown in Fig. 7, with the information side bands clearly visible. This signal was then demodulated using a readily available double-balanced mixer (MCL ZFM-150) as synchronous demodulator, with the output displayed on a sampling oscilloscope (Tektronics TDS-3032). Finally, the experiment was repeated with 160Mb/s data, and the clean eye-diagrams for 60Mb/s and 160Mb/s shown in Figs. 8a and 8b respectively.
A new tuneable optoelectronic band-pass filter has been proposed and demonstrated based on a standard two-section SPL. A model using the well-known rate equations allowed us to look at the pulsation frequency range and the injection process. Frequency ranges depend on the SPL characteristics and could be extended into the Bluetooth (2.4GHz) and HIPERLAN2 (5GHz) ranges.
Sandra E. M. Dudley is grateful for many fruitful conversations about laser modeling with Professor Mike J. Adams, Electronic Systems Engineering Department at the University of Essex.
1. G. J. Lasher, “Analysis of a proposed bistable injection laser,” Solid State Electron. 7, 707–716 (1964). [CrossRef]
2. N. G. Basov, “0-1-Dynamics of injection lasers,” IEEE J. Quantum Electron. QE-4, 855–867 (1968). [CrossRef]
3. M. Ueno and R. Lang, “Conditions for self-sustained pulsations and bistability in semiconductor lasers,” J. Applied Physics 58, 1689–1692 (1985). [CrossRef]
4. P. E. Barnsley, G. E. Wickens, H. J. Wickes, and D. M. Spirit, “A 4×5 Gb/s transmission system with all-optical clock recovery,” IEEE Photon. Technol. Lett. 4, 83–86 (1992). [CrossRef]
5. J. B. Georges and K. Y. Lau, “800Mb/s microwave FSK using self-pulsating compact-disk laser diodes,” IEEE Photon. Technol. Lett. 4, 662–665 (1992) [CrossRef]
6. R. S. Bates and S. D. Walker, “450Mb/s BPSK and 1Gb/s QPSK throughput subcarrier multiple access networks using 790nm self-pulsating laser transmitter network for computer applications,” Electron. Lett. 27, 1014–1016 (1991). [CrossRef]
7. S. F. Su, A.R. Bugos, V. Lanzisera, and R. Olshansky, “Demonstration of a multiple access WDM network with subcarrier multiplexed control channels,” IEEE Photon. Technol. Lett. 6, 461–463 (1994). [CrossRef]
8. S. D. Walker, M. Li, A. C. Boucouvalas, D. G. Cunningham, and A. N. Coles, “Design Techniques for Subcarrier Multiplexed Broadcast Optical Networks,” IEEE J. Sel. Areas Commun. 8, 1276–1284 (1990) [CrossRef]
9. Sandra. E. M. Dudley, Terence J. Quinlan, and Stuart. D. Walker, “A new optoelectronic tunable bandpass filter based on a self-pulsating laser and photodiode combination,” IEEE Wireless Design Conference, London, (2002).
10. E. A. Avrutin, “Analysis of spontaneous emission and noise in self-pulsing laser diodes,” IEE Proceedings-J. 140, 16–20 (1993).
11. Egan A. et al, “Theoretical investigations of electro-optical synchronisation of self-pulsating laser diodes,” Proc.-Optoelectronics 143, 31–36 (1996). [CrossRef]
12. T. W. Carr and T. Erneux, “Dimensionless rate equations and simple conditions for self-pulsing laser diodes,” IEEE J. Quantum Electron. 37, 1171–1177 (2001). [CrossRef]
13. R. Dixon and W. B. Joyce, “Possible model for sustained oscillations in Al, GaAs double heterostructure lasers,” IEEE J. Quantum Electron. QE-15, 470–474 (1979). [CrossRef]
14. M. Kuznetsov, “Pulsations of semiconductor lasers with a proton bombarded segment: Well-developed pulsations,” IEEE J. Quantum Electron. QE-21, 587–592 (1985). [CrossRef]