No abstract

The nonlinearity difference in the two passbands of a distributed-feedback semiconductor laser amplifier was studied experimentally. A theoretical explanation was given by using the transmission matrix approach. The difference of nonlinearity in the two passbands was found to be enhanced greatly by the mechanism of asymmetric facet reflection.Active optical bistability in Febry-Perot (FP) and distributed-feedback (DFB) semiconductor laser amplifiers has the advantages that the power required for switching is approximately three orders of magnitude smaller than that of the passive bistability and, at the same time, the optical gain of the active bistability can provide a sufficient optical power to drive the subsequent optical gates in the system. 1 ' 2 An additional advantage offered by the DFB laser amplifier bistable element is that its filter characteristic can dramatically reduce spontaneous emission 3 in comparison with the FP laser amplifier, and a higher contrast of optical logical gate can be achieved. 4 The two passbands of a DFB laser amplifier, which are determined by the Bragg diffraction, also make it possible for it to be used in systems with two different wavelengths.In this Letter we extend the average field approximation reported in Refs. 1 and 2 to give a unified treatment of optical bistability for FP and DFB laser amplifiers and for a DFB laser amplifier with finite facet reflections. This particular structure is important in determining the single-longitudinal-mode behavior of DFB lasers and the filter characteristic of DFB laser amplifiers. Here we show, experimentally and theoretically, that this additional structure in a DFB laser amplifier will result in a large nonlinearity difference with respect to its two passbands.The experimental setup is as follows. Two identical DFB buried-heterostructure laser diodes were used as a probe and an amplifier; both of them had one facet antireflection coated and the other facet cleaved. The lasing wavelengths were approximately 1554 nm, and the probe was biased at 2.5 times its threshold current. Two diffraction-limited lenses with a N.A. of 0.6 were used for the beam coupling. An optical isolator inserted between the probe and the amplifier provided 40 dB of isolation, and a half-wavelength plate was used to match the polarization of the two lasers. A monochromator and a FP scanning interferometer were used for rough and fine measurement of the spectrum. Frequency matching and adjusting were accomplished by controlling the laser heat-sink temperature, and the change of the coupled optical power caused by the change of temperature was controlled to less than 5%. Figure 1 gives the measured output near the two frequencies corresponding to the lasing band [ Fig. 1(a)] and the nonlasing band [ Fig. 1(b)] of the amplifier biased at 95% of its threshold. The probe power coupled into the amplifier was approximately 0.9 MW. The short-wavelength band (which would lase if biased above threshold) shows a bistability loop, whereas the long-wavelength band shows...

No abstract

No abstract

In this scheme, each reflection coefficient is quantised independently. The analysis steps are as follows. The best reflection coefficient is obtained by searching the codebook using eqn. 7, then from eqn. 5 the prediction errors are updated using the quantised reflection coefficient. These steps are repeated until all reflection coefficients have been fixed. Scheme 2: two step search: In this scheme, two reflection coefficients are quantised in a single step. By evaluating eqn. 5 for two consecutive steps, we have = m a d + k,?,+,Xl + k;+J-2k:+lkn+2B; + 4k,+tk,+2a7 (8)-2k,+ 1(1+ kf+i)bT-2k,+ 2 B' ; The minimum of a;;+' is achieved by searching the quantisation tables for k,,, and k,,' in a single step. Once the best k,,, and km+2 are found, they are kept constant for the remaining analysis steps. This scheme requires more computation than scheme 1 as all combinations of two consecutive tables have to be searched.