If the current is continuously tuned, it is possible to observe different waveforms through the
complete working cycle of the SOA-MZI. Once we perform such process, we select the two stages
that best fit the conditions for generating a doublet. When the electrical current of SOA2 is tuned to
270 mA, the input pulse is large enough to saturate the interferometric structure (SOA-MZI) and
therefore a doublet pulse is generated. In this context, it is also feasible to switch the polarity of the
obtained doublet pulse by modifying the value of I2 and setting it at 176 mA. Fig. 5(a) and (b) plot the
waveform and corresponding spectrum for the measurement performed at I2 ¼ 270 mA. Fig. 5(c)
and (d) reveal these same parameters but with a current value of 176 mA. By comparing the
characteristics of the obtained doublet pulses, we can determinate that the amplitude of the UWB
doublet in Fig. 5(a) is much higher than the one shown in Fig. 5(c). The reason is due to the fact that
the conversion efficiency is lower for a current of 176 mA than for 270 mA as shown in Fig. 2. In
addition, Fig. 5 includes the FCC mask in the electrical spectrum of each doublet pulses. When the
obtained spectrum is analyzed in terms of the FCC mask, we observe that RF power could be
reduced at levels where FCC requirements are accomplished. For practical applications in order to
solve limitations in terms of maximum emission by means of a UWB antenna, an additional filtering
stage could be considered.
Comparing with previous works using SOAS [11]–[13], we obtain a better performance in terms of
noise characteristics as the MZI structure improves the extinction ratio of the conversion process
[15]. Recently, a photonic approach based on an optical carrier phase-shifting method by cascading
PolMs has been presented [7]. The generated pulse complies with the FCC mask but reconfiguration
or scalability is more complex compared to our proposal as it involves the manipulation of
polarization states. Other reported techniques which are based on PM to IM conversion [8] and
birefringence tailored time delays using SOAs [9] suffer from scalability problems. Alternate
schemes for the generation of higher order derivatives exist leading to a better power efficiency [10],
but at the expense of considerable system complexity. In contrast, our approach permits to obtain a high performance having the potential to easily achieve higher order pulses by increasing the
number of optical carriers. In principle, the use of EOMs and SOAs could increase the complexity in
the experimental assembly. However, the last advances on photonic integrated circuits (PIC) lead
to the introduction of EOMs, SOAs and even AWGs on a single chip showing enhanced features
and robustness as well as reduction of size, weight, cost and power consumption [16]. Very recently
an IR-UWB pulse shaping through a PIC frequency discriminator has been demonstrated [17].
Therefore, progress and availability of PIC technologies increase the viability of the propose
scheme for future improvements.