114 High-Speed Optical Modulators and Photonic Sideband Management Tetsuya Kawanishi National Institute of Information and Communications Technology 4-2-1 Nukui-Kita, Koganei, Tokyo, Japan Tel: 81-42-327-7490; Fax: 81-42-327-7938; E-mail: kawanish@nict.go.jp Abstract- Most important application of optical modulators is conversion of electric signals into lightwave signals. This is indispensable to highspeed optical links. However, recently, we need new functions such as optical label processing at nodes, signal timing control to prevent packet collision, etc. In this paper, we describe advanced optical modulators designed for the photonic sideband management techniques, such as resonant-type modulators for effective optical sideband generation, reciprocating optical modulators for photonic millimeter-wave generation, and optical frequency-shift-keying modulators for optical packet labeling. Index terms- microwave, optical modulation I. INTRODUCTION Optical intensity modulators, which are most commonly used in commercial systems, can be constructed by using electro-optic (EO) effect. In a material having the EO effect, the refractive index depends on the voltage applied on the material. Thus, we can obtain an optical phase modulator comprised of an optical waveguide and an electrode. The other types of modulators, such as intensity modulators, FSK modulators, can be constructed by optical phase modulators. The EO effect is so small that the length of the modulator should be a few centimeters to obtain effective modulation. When the frequency of the electric signal on the electrode is high, the wavelength of the electric signal would be less than the electrode length. In addition, there is phase delay due to lightwave propagation along the waveguide. Thus, we have to use a transmission line theory to design the electrode for high-frequency operation. There are two types of electrode for high-frequency operation: traveling-wave and resonant electrodes. The traveling-wave electrode gives broadband response from dc to 50 GHz. Thus, the modulator with the traveling-electrode is suitable for highspeed digital transmission system. On the other hand, the resonant electrode can be used for band operation, for example, 25-26GHz band. The modulator is useful for some special applications, such as, clock generation, sideband generation with a fixed frequency separation, and radio-onfiber systems. The modulator size is much smaller than conventional one, so that this is suitable for integrated modulators which have several functions. In this paper, we discuss the electrode structures for high-frequency operations, and the design scheme for resonant electrodes. We also describe novel optical modulators consisting of phase modulators: 1) reciprocating optical modulator and 2) frequencyshift-keying (FSK) modulator. II. RESONANT-TYPE MODULATOR Fig.1. Top view of an optical modulator with an double-stub structure.
115 Fig. 1 shows a resonant-type modulator structure comprising a pair of modulating electrodes and stubs [1]. The modulating electrodes are asymmetric coplanar waveguides (ACPWs). A feeding line, which is a coplanar waveguide (CPW), is connected at the center of the modulating electrode. Two stubs, which are also CPWs of the same length, are also connected at the junction of the modulating electrode and the feeding line, so that we call it a double-stub structure. The equivalent circuit is shown in Fig. 2, where the impedance of an arm of the electrode (Z 1 ) and that of a stub (Z 2 ) are expressed by between the field of the lightwave and the field induced by the electrode. The integral of the voltage on the electrode defined by Eq. (6), Φ, is termed normalized induced phase, and shows the effect of the resonant structure. The normalized induced phase for an optical modulator with a resonant structure can be larger than unity, while the normalized induced phase for a lossless perfectly velocity-matched traveling wave modulator equals unity. The half-wave voltage of the modulator Vπ is given by V in π/φ. Similarly, VπL, defined by V in πl/φ, is the voltage-length product which shows the degree of the modulation efficiency and the scale of the modulator. when we consider the case in which the modulating electrode is open-ended and the stub is short-ended. L i, γ i and Z 0i denote the lengths, the propagation coefficients and the characteristic impedances of the modulating electrode ( i = 1 ), and of the stub ( i = 2 ). Total impedance of the double-stub structure is given by The voltage on the modulating electrode is given by Fig.2. An equivalent circuit of the double-stub structure. where Z f is the characteristic impedance of the feeding line. The induced phase at each optical waveguide is the sum of the Pockels effect with respect to the coordinate system moving along with the lightwave propagating in the optical waveguide. Thus, the difference of the induced phases of the two optical waveguides of the Mach-Zehnder structure can be expressed by where c is the speed of light. λ 0 and n 0 are the wavelength and refractive index of the lightwave, respectively. L ( =2L 1 ) is the total length of the modulating electrodes. Γ is the overlap integral Fig.3 Normalized induced phase of the fabricated modulator. To obtain compact and effective optical modulators for band-operation, we designed them
116 to have large normalized induced phases, by varying the lengths L 1 and L 2. We considered the electrode structure with the following parameters. The gap between the electrodes of ACPWs and CPWs is 27μm. The width of the modulating electrode is 5μm, while that of the stub is 50μm to reduce loss at the stubs. The characteristic coefficients of the modulating electrode (ACPW) are γ 1 = 25.92 + j735.2 and Z 01 = 70.1Ω. Those for the stub (CPW) are γ 2 = 14.29 + j863.5 and Z 02 = 28.7 Ω. By using Eq. (6), we obtained a combination of L 1 and L 2, which gave a maximum of Φ. When L 1 = 0.19 λ 1 and L 2 = 0.12 λ 2, Φ had a maximum of 2.44 at 10 GHz, where λ 1 and λ 2 are, respectively the wavelength on the modulating electrode and the stub. The normalized induced phase of the fabricated modulator for the 1.55-μm region, as shown in Fig. 3, had a peak of 2.65 at 10.59 GHz. The halfwave-voltage was 16.4 V at the peak, while the total length L was 3.25 mm. III. RECIPROCATING OPTICAL MODULATOR Reciprocating optical modulator (ROM), consisting of a pair of optical filters and an optical modulator [2], can generate high-order sideband components of electro-optic modulation, as shown in Fig. 4, where the fiber Bragg gratings (FBGs) were fixed in V-grooves on SiO 2 substrates and directly attached to the LiNbO 3 modulator chip. One of the optical filters is placed at the optical input port (input filter), and the other is at output port (output filter). By using an ROM, we can generate a lightwave modulated by an rf signal whose frequency is an integer multiple of an electric rf signal applied to the modulator. The phase-shifted FBG has a narrow pass band of 1.5 GHz in the reflection band. The output filter was a conventional FBG with a heater to let the center wavelength equal to that of the input filter. When the input wavelength was in the pass band of the input filter, both USB and LSB reciprocate between the input and output filters, so that high-order double sideband components can be effectively generated. The resonant electrode enhances modulation efficiency for forward and backward lightwaves, while a traveling-wave electrodes acts only on forward propagating lightwaves. The resonant frequency of the electrode was designed to be equal to inverse of delay in a reciprocation process, so that the successive modulation process can be in-phase. Fig. 5 shows the output spectrum of the fabricated ROM and that of the conventional optical phase modulation, where the input rf-power and frequency were, respectively, 19.8 dbm and 5 GHz. Optical power and frequency of the input lightwave were, respectively, 4.2 dbm and 192.956 THz. Sideband components lower than 5th-order were in the reflection band, so that 5th or 6th order harmonic components were generated effectively. In order to obtain very high-frequency components over 100 GHz, input and output filters should have wide reflection bands. For effective high order sideband generation, the filters should have steep reflection band edges. However, it is not easy to design optical fibers having wide reflection bands and very steep band edges. FBGs with apodization techniques are often used to suppress undesired sidelobes adjacent to the main reflection band. On the other hand, recently, we proposed ROM using uniform FBGs for input and output filters [3]. The uniform FBGs would have sidelobes, but there is a very narrow passband between the main reflection band and a sidelobe. We fabricate an ROM with a pair of uniform FBGs. The designed deviation between the narrow passbands at the lower and upper edges of the main reflection band was 160 GHz. The uniform FBGs have narrow passbands near the edges of the main reflection bands, as shown in Fig. 6. The bandwidth of the main reflection band was slightly narrower than 160 GHz, where the passbands were at 1557.5 nm and 1556.2 nm. Designed frequency of modulating signal fed to the phase modulator was 40 GHz. When an input lightwave whose wavelength is close to 1557.5 nm is fed to the input filter, the fourth order sideband and input components can be taken out from the output filter, where the output signal is amplitude-modulated at 160 GHz. We used an
117 optical phase modulator with a traveling-wave electrode to achieve high-speed operation, where halfwave voltage of the modulator was 3.7 V at dc and 8.2 V at 40 GHz. the input lightwave and out of the main reflection band. The upper sideband components lower than the fourth order were reflected by the input and Fig.4 Reciprocating optical modulator with resonant electrode Fig.6 Reflectivities of FBGs as functions of input wavelengths. Fig.5 Output spectra modulated by ROM and by conventional phase modulation. We demonstrated generation of 160 GHz amplitude-modulated lightwave signals by using the ROM with the uniform FBGs, where the optical input power was 5.7 dbm. A pair of rf signals were fed to the ROM, to obtain bidirectional modulation. In ROM sideband generation process, sideband components in the reflection band reciprocate between the FBGs, and pass the modulator several times. Thus, bidirectional modulation is indispensable to obtain effective high-order sideband generation. The frequency of the rf-signals was 40.05 GHz, where the rf powers for forward and backward propagating lightwaves were, respectively, 21.4 dbm and 20.6 dbm. As shown in Fig. 6, the input lightwave was in the narrow passband near the longer wavelength band edge of the main reflection band. The fourth order upper sideband component was in the passband near the shorter wavelength band edge, where lower sideband components whose wavelengths were longer than Fig.7 Optical spectrum after the two FBGs. output filters, to reciprocate several times in the ROM.We also used two FBGs placed after the ROM output to suppress the undesired components. An optical spectrum after the FBGs was shown in Fig. 3, where the fourth order sideband component was effectively generated. We measured autocorrelation profiles of the optical signal, where an optical amplifier was used to compensate the loss in the ROM process. Fig. 8 shows an optical spectrum at the output port of the amplifier. The intensities of the two desired components (the input and the fourth order upper sideband) were equalized by using an
118 optical filter. As shown in Fig. 9, the measured autocorrelation profile was a sinusoidal-like curve of a 6.25 ps period, which corresponds to the 160 GHz amplitude-modulated lightwave consisting of the two spectral components. out affecting its intensity. Simultaneously modulated IM and FSK signals are independently demodulated by using an optical filter, so that the label information can be extracted without affecting the payload signal. For this purpose, fast and wideband external FSK modulation technique is required, which has never developed. In previous works, FSK signal was generated by direct modulation of electric current in a laser light source [4]. Thus, FSK bit rate is limited by the response of the laser. 160GHz optical clock signal generation was also obtained by using a reciprocating optical modulator with uniform FBGs. Fig.8 Optical spectrum at the amplifier output. Fig.10 Optical FSK modulator Fig.9 Autocorrelation profile. IV. OPTICAL FSK MODULATOR In optical packet systems, a combination of intensity modulation (IM) and frequency shift keying (FSK) is a promising technique for label switching. In this technique, payload signals are in IM format, while label information is written by FSK signal. The merit of this FSK labeling is that an FSK transmitter generates the label by FSK signal. The merit of this FSK labeling is that an FSK transmitter generates the label information on the optical carrier frequency with The FSK modulator consists of parallel four optical phase modulators as shown in Fig.10. When we apply a pair of rf-signals, which are of the same frequency fm and have a 90-degree phase difference, to the electrodes RFA and RFB, frequency shifted lightwave can be generated at the output port of the modulator. A sub Mach- Zehnder structure of path 1 and 3 should be in null-bias point (lightwave signals in the paths have 180-degree phase difference), where the dcbias can be controlled by RFA The other sub Mach-Zehnder structure of path 2 and 4 are also set to be in null-bias point by using RFB. To eliminate upper sideband (USB) or lower sideband (LSB), the lightwave signal in each path also should have 90-degree phase difference each other. When the phase difference induced by RFC is 90 degrees, we can get carrier-suppressed single sideband modulation comprising one of the sideband components (USB or LSB). Thus, the optical frequency of output lightwave can be switched by changing the induced phase at RFC. The amplitudes of USB and LSB are, respectively, described by [1+exp(iφ FSK )]/2 and [-1-
119 exp(-iφ FSK )]/2, where φ FSK is the induced phase difference at RFC, and φ FSK = 0 degrees corresponds to an optimal condition for USB generation. Thus, by feeding an NRZ signal, whose zero and mark levels correspond to φ FSK = 0, 180 degrees, to RFC, we can generate an optical FSK signal, without any parasitic intensity modulation. The bandwidth for the FSK signal should be smaller than the rf-frequency fm, when the FSK signal is demodulated by optical filters. The setup for FSK/IM modulation is shown in Fig. 11. We experimentally demonstrated simultaneous transmission of IM (9.95Gbps) and FSK (1Gbps), which are NRZ PRBS (2 31-1) signals. The frequency deviation of FSK (2fm) was 25 GHz. The FSK signal was demodulated by using an optical filter which can discriminate between USB and LSB. The extinction ratio of the IM modulation was 3.9 db. We measured bit-error-rate (BER) performance and eye-diagrams of back-to-back, as shown in Fig. 12. Residual IM in the FSK modulation was so small that the IM output which can be generated just by feeding the FSK/IM signal to a high-speed photo detector has a clearly opened eye-diagram. We obtained error-free transmissions for FSK and IM. Fig.11 Experimental setup for optical FSK/IM signal generation IM transmission IV. CONCLUSION For high-speed optical modulation in millimeteror micro-wave bands, we investigated the resonant-type optical modulator to reduce the half-wave voltage in desired frequency. The enhancement due to the resonance was 2.65 at 10.59 GHz. We also demonstrated 50GHz millimeter-wave generation from 5GHz microwave input, by using a reciprocating optical modulator. Optical FSK modulation technique was also described, where optical FSK labeling on IM signals was demonstrated. ACKNOWLEDGMENT We would like to express our appreciation to Dr. M. Izutsu, Dr. M. Tsuchiya, Dr. T. Sakamoto and Dr. S. Shinada of National Institute of Information and Communications Technology, for their useful discussion. We also wish to acknowledge the support of Mr. J. Ichikawa, Mr. S. Oikawa, Mr. K. Higuma and Mr. T. Fujita of Sumitomo Osaka Cement, and Mr. K. Yoshiara of Mitsubishi Electric. This study was partially supported by Industrial Technology Research Grant Program in 2004 from New Energy and Industrial Technology Development Organization of Japan, and by Grant-in-Aid for Young Scientists (A) 17686032 from the Ministry of Education, Culture, Sports, Science and Technology, Japan. REFERENCES Fig.12 Eye diagrams and BER curves for FSK and [1] T. Kawanishi, S. Oikawa, K. Higuma, Y. Matsuo and M. Izutsu, LiNbO 3 resonant-type optical modulator with double-stub structure, Electron. Lett., Vol. 37, pp. 1244-1246, Sept. 2001 [2] T. Kawanishi, S. Shinada, T. Sakamoto, S. Oikawa, K. Yoshiara and M. Izutsu, Reciprocating optical modulator with a resonant modulating electrode, Electron. Lett., Vol. 41, pp. 69-70, March 2005 [3] T. Kawanishi, S. Shinada, T. Sakamoto, M. Izutsu, S. Oikawa, Y. Shimakura and K. Yoshiara, 160GHz Reciprocating Optical Modulator Using
120 Uniform Fiber Bragg Gratings, OFC 2006 JThB28, Ahaheim, USA, 5-10 March 2006. [4] T. Kawanishi, K. Higuma, T. Fujita, J. Ichikawa, T. Sakamoto, S. Shinada and M. Izutsu, Highspeed optical FSK modulator for optical packet labeling, IEEE/OSA J. Lightwave Technol., Vol. 23, pp. 87-94, Jan. 2005