The temporal traces of pulse trains on different wavelengths at the input and output side of the SOA are displayed in Fig. 2 with an input signal corresponding to a Gaussian pulse train whose repetition rate is 1.25 GHz and full width at half maximum (FWHM) is 70 ps. The FWHMs of the output lights on signal and probe wavelengths are 88.1, 85.7, and 85.5 ps, respectively. The input powers of the pump and two probe lights are 1, 0.6, and 0.6 mW.
Fig.2 Temporal trace of lights at different wavelengths when the input signal is a Gaussian pulse train. (a) Input signal; (b) output signal; (c) output probe at lp1; (d) output probe at lp2 |
Full size|PPT slide
The temporal traces of pulse trains on different wavelengths at the input and output side of the SOA are displayed in Fig. 3 with an input signal of a polarity-reversed Gaussian pulse train whose repetition rate is 1.25 GHz and FWHM is 70 ps. The FWHM of the output signal and probes are 55.3, 51.9, and 53.6 ps, respectively.
Fig.3 Temporal trace of lights at different wavelengths when the input signal is a polarity-reversed Gaussian pulse train. (a) Input signal; (b) output signal; (c) output probe at lp1; (d) output probe at lp2 |
Full size|PPT slide
The waveforms of the generated UWB signals are shown in Fig. 4, while the time delay of ODL1 is controlled at different amounts. Throughout the operation process, the time delay caused by ODL2 is controlled as twice as the value of the time delay caused by ODL1. In Figs. 4 (a)–4(c), a Gaussian pulse train is used as the input signal. After time delays are introduced, each Gaussian pulse in the output signal pulse train is adjusted in the middle of two polarity-reversed Gaussian pulses in the output probe pulse trains. Our results reveal a UWB pulse train in which each pulse exhibits a negative doublet shape in the final output. In Figs. 4(d)–4(f), a polarity-reversed Gaussian pulse train is used as the input signal, that is, each polarity-reversed Gaussian pulse in the output signal pulse train is adjusted in the middle of two Gaussian pulses in the output probe pulse trains. As such, the UWB signal is produced in a positive doublet shape at the final output.
Fig.4 Waveforms of the obtained UWB signal in doublet shape with time delay introduced by ODL1 tuned at different values. (a)–(c) 40, 80, and 120 ps (UWB signals in negative doublet shape); (d)–(f) 40, 80, and 120 ps (UWB signals in positive doublet shape) |
Full size|PPT slide
The amplification and wavelength conversion in SOA are not perfect. Nonlinear effects, such as XGM and cross-phase modulation (XPM), likely cause a slight distortion in pulse shape. Consequently, Gaussian and polarity-reversed Gaussian pulses are not absolutely symmetric. The overlap becomes heavy when the time delay is small. Thus, the left and right peaks may vary (Fig. 4 (d)).
In Figs. 2 and 3, the pulse widths of the generated probe pulses are almost equal. However, the output signal pulse is asymmetric and thus results in different left and right FWHMs when the time delay is small, that is, the overlap of pulses is heavy. However, the overlap becomes slight when the time delay is large, and the same left and right FWHMs appear.
The dependence of the generated UWB doublet pulse widths on time delay caused by ODL1 are plotted in Fig. 5. The FWHMs of the left, center, and right parts of the generated UWB signal in a negative doublet shape can vary from 44.2 to 80.8 ps, 57.8 to 85.7 ps, and 42.8 to 80.7 ps, respectively (see Fig. 5(a)). For the generated UWB signal in the positive doublet shape, the corresponding parameters can change from 38.8 to 51.4 ps, 35.1 to 55.3 ps, and 36.6 to 51.4 ps, respectively (see Fig. 5(b)).
Fig.5 Pulse widths depending on time delay of ODL1. (a) UWB signals in negative doublet shape; (b) UWB signals in positive doublet shape |
Full size|PPT slide
The FWHM of the output signal in the Gaussian shape is 88.1 ps (44.05 ps for a half pulse width) because Gaussian and polarity-reversed Gaussian pulses on the output signal and probes are not absolutely symmetric. The peaks of the probe pulses remarkably affect the center FWHM of the final output signal when the time delay is small (40 to 60 ps). As such, the value abnormal changes.
The RF spectra of the generated signals and their envelopes are shown in Fig. 6 to verify the successful generation of the UWB doublet signal. Figure 6 illustrates a repetition frequency of 1.25 GHz of the output UWB doublet signal.
Fig.6 RF spectra of the generated signals and their envelopes in doublet shape with time delay introduced by ODL1 tuned at different values. (a)–(c) 40, 80, and 120 ps (UWB signals in negative doublet shape); (d)–(f) 40, 80, 120 ps (UWB signals in positive doublet shape) Fc: central frequency; BW10 dB: bandwidth at 10 dB |
Full size|PPT slide
Figure 7 denotes the RF spectrum variation with time delay changes caused by ODL1. In the experiment, the central frequency, bandwidth at 10 dB, and fractional bandwidth of the generated UWB signal in the negative doublet shape can be varied from 6.22 to 3.75 GHz, 7.68 to 4.76 GHz, and 123.5% to 126.9%, respectively. For the UWB signal in a positive doublet shape, these parameters can be changed from 8.82 to 3.92 GHz, 11.77 to 4.89 GHz, and 133.4% to 124.7%, respectively. We note that the RF spectra of the realized signals fit the UWB definition of FCC quite well.
Fig.7 Central frequency, 10 dB bandwidth, and fractional bandwidth depending on the time delay of ODL1. (a) and (b) UWB signal in negative doublet shape; (c) and (d) UWB signal in positive doublet shape. (a) and (c) central frequency and 10 dB bandwidth; (b) and (d) fractional bandwidth |
Full size|PPT slide
The pulses of the output signal and probes from SOA are narrower when the input signal is in polarity-reversed Gaussian shape. Under this condition, the left and right peaks possibly influence the final shape and consequently exhibit more remarkable changes in the pulse width of the positive doublet than in the negative doublet. As such, the spectrum parameters are also altered.
The central frequency and 10 dB bandwidth of the negative doublet almost isometrically change because of the mild overlap and thus slight alter the fractional bandwidth. However, the 10 dB bandwidth of the positive doublet decreases more rapidly than the central frequency does when time delay increases. As a result, this phenomenon decreases the fractional bandwidth.
Analyzing the simulation results shown in Figs. 4, 5, 6, and 7, we discover that the left, central, and right FWHMs of the generated UWB signal in doublet shape increase when the time delay caused by ODL1 increase. Conversely, the center FWHM of the UWB signal in the negative doublet shape slightly decreases when the time delay is small, but the central frequency and bandwidth at 10 dB decrease. In this manner, the tunability in signal pulse width and RF spectrum is obtained for the generated UWB signal in doublet shape by adjusting the time delay caused by ODL1 and ODL2. Moreover, switchability in the polarity of pulse can be achieved by switching the input signal between Gaussian pulse train and polarity-reversed Gaussian pulse train.
The proposed all-optical switchable UWB doublet signal generation scheme can be conveniently applied to the pulse polarity modulation (PPM) scheme for all-optical UWB signal transmission and the PPM is regarded as an essential modulation method for the application of UWB signals [
12]. The optical tunability characteristic of the proposed UWB doublet signal generation scheme provides the facilities for signal pulse width and RF band control in a practical and flexible application system.