1. Introduction

Optical spectral processing is a very useful technique employed in many optical applications such as optical pulse shaping [1], dispersion compensation [2–4], equalization [5–7], and secure optical communications [8–16].

A photonic spectral processor (PSP) is commonly composed of a dispersive element to separate the light’s frequency components, which is followed by a modulator to encode the amplitude and phase of each wavelength [1–4]. The performance of the PSP is determined by its spectral resolution and addressability, which depend on the dispersive element's resolution and the phase modulator's technology, respectively [12].

Currently proposed PSPs utilize various dispersive elements such as reversed dispersion fibers [10], super-structured fiber Bragg gratings (SS-FBGs) [17,18] and waveguide grating routers (WGRs) [19,20]. The modulation has been performed by several devices such as a polymer thermos-optic lens [4], deformable mirror [21] and a spatial light modulator (SLM) [22, 23]. Recently, a high resolution PSP using a WGR with free spectral range (FSR) of 200GHz and 250 channels has been proposed [24]. However, fabricating such WGR requires both very large delay and high phase accuracy at its output [24]. Since these requirements are beyond the standard fabrication tolerances, the fabricated WGR requires an additional SLM to correct the phase errors [24, 25]. Another approach was to use ultra-violet pulsed laser that inscribes permanent optical path changes to the waveguide [26]. These WGR fabrication requirements resulted in a resolution limit of 0.8GHz.

In this paper we propose a novel PSP that overcomes the resolution limitation of the current devices, being able to obtain a spectral resolution of about 50MHz. In this new scheme the input signal passes through a spectral encoding block having high resolution features before it reaches the spatial dispersive element. The proposed encoding block consists of a set of parallel Fabry-Perot interferometers with different cavity lengths such that each interferometer has a slightly shifted output spectrum in a way that the encoding block covers the whole input signal's spectrum. In addition, our new scheme uses both dimensions of the SLM array rather than only one of them. A description of the suggested scheme and its principle of operation are presented in addition to experimental results that demonstrate spectral processing at high optical resolution of 577MHz, while using a dispersive element which has low spectral resolution of 6.5GHz. These results show the ability of the proposed method to significantly improve the PSP's performance compared to the current techniques. Based on the principles presented in this paper, further improvement of the system could achieve higher resolution of about 50MHz, depending on the Fabry Perot interferometers’ full width at half maximum (FWHM).

2. The proposed system

The suggested PSP scheme can be seen at Fig. 1(a).

 

Fig. 1 (a) Proposed system for high resolution spectral phase encoder. (b) Detailed structure of Fabry-Perot encoding block with different output wavelengths per each of its rows.

Download Full Size | PDF

The PSP consists of a Fabry-Perot Interferometer (FPI) encoding block, optical diffraction grating, LCoS SLM, and lenses used to collimate and magnify the field that incidents on the SLM. The FPI block shown in Fig. 1(b) consists of a set of parallel Fabry-Perot interferometers, each having a different cavity length, and thus a different set of output wavelengths. The output beam of each interferometer goes to a different vertical location on the diffraction grating. A Fourier lens is then used to collimate the grating's output on the SLM. Thus, the output of each FPI reaches a different row of the SLM, where different output wavelengths of each FPI are focused on different columns of the SLM, so both dimensions of the SLM are used. Finally, the SLM modulates the spectral amplitude and phase of the signal.

The resolution limit of the PSP is dictated by the FWHM of the FPIs rather than the diffraction grating's resolution. This is obtained by using FPIs with FSRs higher than the resolution limit of the grating. Thus, the diffraction grating can resolve different wavelengths of the same FPI to different columns of the SLM. In addition, the different lengths of the FPIs' cavities should be designed such that the spectrum of the FPI array will span the entire input signal’s spectrum. Since the PSP's resolution is limited by the FPIs' FWHM rather than the diffraction grating's resolution, we obtain a considerable resolution enhancement, as the diffraction grating's resolution is about a few GHz while the FPI FWHM can be three orders of magnitude lower.

Figure 2 shows an example for spanning a bandwidth of 20GHz by an array of FPIs. The different lines in the graph represent the FPIs' output intensity versus wavelength, where each color corresponds to a different FPI. In our example we chose the FSR of each interferometer in the array to be equal to 5GHz, while the FWHM was about 700MHz. Thus, placing all the interferometers in parallel ensures a complete coverage of the bandwidth.

 

Fig. 2 Spectral division of a 20GHz bandwidth by FPIs array. Different colors correspond to different FPIs.

Download Full Size | PDF

3. Experimental results

In order to demonstrate the ability of the proposed PSP to achieve high spectral resolution, the experimental setup shown at Fig. 3 was built.

 

Fig. 3 (a) A scheme of the experimental setup. (b) Image of the experimental setup. EDFA - Erbium doped fiber amplifier, P.C - polarization controller, F.P - Fabry-Perot, B.S-beam splitter, SLM - spatial light modulator.

Download Full Size | PDF

A tunable laser source was used in order to scan wavelengths from 1532.4nm to 1532.5nm in steps of 0.001nm (about 125 MHz) in scanning rate of 0.65sec per wavelength. The input laser beam was collimated and split into two beams, where each beam was filtered by a different Fabry-Perot interferometer. The two interferometers had FSRs of 9.1GHz and 9.2GHz with FWHM of 55MHz and 52MHz, respectively. The cavity length of each FPI was controlled by an input voltage driving a piezo electric transconductor to determine the output spectrum of each FPI.

The output beams of the two interferometers propagate to the reflective diffraction grating which has a period length of $d=1.67\mu m$ and width of $B=50mm$, hence its spectral resolution is [27]:

We first set the FPIs cavity lengths by scanning the above mentioned wavelength range and measuring the intensities at the two detectors versus time without any phase modulation at the SLM. We repeated this measurement for different cavity lengths of the two interferometers until their output wavelengths were as close as possible to each other. The measured results can be seen in Fig. 4.

 

Fig. 4 Light intensity as function of optical frequency for two different Fabry-Perot resonators with wavelength sweep in their inputs.

Download Full Size | PDF

Figure 4 shows the measured intensity at the cavities outputs, where the x axis was converted from time to optical frequency as follows.

Since each 0.65sec the input laser wavelength increased by 0.001nm (−125MHz), the conversion relation between time differences $\Delta T$to frequency differences $\Delta \nu$ is given by:

After finding the desired cavities lengths we performed two measurements where in each measurement we attenuated the power of a different FPI's output beam by the SLM. The attenuation was performed by loading a phase blazed grating on the area of the SLM where the beam to be attenuated reaches [28]. The blazed grating shifted the output angle of the reflected beams according to the blazing angle, so the spatial phase modulation would change the measured intensities at the detectors. The measured intensities versus time for these two measurements can be seen in Figs. 5(a) and 5(b). These measurements show that it is possible to modulate each wavelength separately at high spectral resolution of 577MHz instead of the 6.45GHz resolution limit of the diffractive grating. The theoretical resolution limit of our setup is the sum of the tunable laser's linewidth, which is 500MHz (we used the HP 8168F tunable laser) and the FWHM of the FPIs, which is about 50MHz. Hence, the theoretical resolution limit is 550MHz, which is compatible with the measured 577MHz resolution limit. The setup's 52MHz resolution, dictated by the FPIs' FWHM, can be measured by using a narrow linewidth input laser.

 

Fig. 5 Light intensity vs. optical frequency for two FPIs' outputs that are 577MHz apart where the input laser's frequency is swept: (a) modulation of only one wavelength. (b) modulation of only the second wavelength.

Download Full Size | PDF

Note that usage of spectral grating based on amplitude transmission Fabry-Perot is not very energetically efficient. However, one of the ways to solve it is to use spectral phase grating. Such an etalon is called Gires-Tournois which is obtained if one of the mirrors in the Fabry-Perot resonator has 100% reflectivity while the reflectivity of the second mirror controls the Finesse of the resonator.

Another important issue is the usage of the proposed configuration for optics communication. Our encoding high Finesse Fabry-Perot resonator needs to be charged with photons which takes time and thus it affects the highest temporal frequency that can be sent through the device. The temporal response will be limited by temporal constants equal to the product between the Finesse and the time it takes the photons to travel forth and back between the two mirrors of the resonator (this time equals to twice the distance between the mirrors divided by the speed of light in the etalon). It is true that our current optical realization included physically large etalon, but a compact monolithic realization is also possible for the future and in this case the distance between the two mirrors can be very small which shall reduce the effect of the above-mentioned limitation. For instance, let us assume that the distance between the mirrors is 10μm and the refractive index of the medium is n = 1.5 then for Finesse of even 1000, the temporal limitation will be 50 pico-seconds (20 GHz).

4. Conclusion

We propose a novel PSP which overcomes the current PSPs' spectral resolution limitations by using a Fabry-Perot interferometer array before the dispersive element. An experimental proof of concept was demonstrated, achieving a 577MHz resolution limit by using two Fabry-Perot interferometers with different cavity lengths, followed by a diffraction grating and an SLM. Further improvement of the experiment's setup can yield a resolution limit of about 50MHz.

Please note that the innovation of our approach is in the system design that we have and not in the specific experimental setup that we used to show its proof of concept. We do intend in the near future to construct an experimental monolithic device that will follow the theory of the current manuscript which will be experimentally characterized. Ref [29]. is a former work done in our group which addresses the experimental construction of a Fabry-Perot resonator having space varying free spectral ranges as we aim to have for the current super resolving concept.