Field of the Invention
This invention relates generally to optical modulators, and more particularly to an optical modulator that uses optical predistortion to maximize linearity and also provides signals having substantially no chirp. The modulator enables low attenuation, low dispersion fiber optic transmission over fibers having a zero dispersion wavelength that is significantly different from the operating wavelength of the source laser.
Background of the Invention
Over the last several years, many fiber optic network providers have included 1550 nm links in their newer network buildouts, some of which have older fibers, creating networks in which 1550 nm source lasers transmit into conventional optical fibers optimized for 1310 nm transmission. One advantage of 1550 nm transmission is its superior attenuation performance. In conventional SiO.sub.2 -based single-mode optical fibers, for example, the attenuation of 1550 nm optical signals is about 0.2 dB/km, while the attenuation of 1310 nm signals in the same fiber is about 0.35 dB/km. The lower optical loss at 1550 nm allows for longer, unrepeatered link spacings.
Another advantage of 1550 nm transmission is the recent commercial availability of low cost, low noise rare earth ion doped optical fiber amplifiers. The development of these amplifiers has made it possible to reamplity and distribute signals without optical detection and regeneration, resulting in substantial cost savings. To date, the most widely used rare earth ion doped amplifiers make use of erbium doped fibers that operate in the range from about 1520-1565 nm. Because of these advantages, transmission at wavelengths of 1550 nm has become a practical and desirable option to traditional 1310 nm transmission.
Despite the advantages of 1550 nm operation, unacceptable levels of chromatic dispersion can occur when 1550 nm wavelengths are transmitted through conventional optical fibers. Dispersion is a linear effect present in an optical fiber whereby different frequencies of light travel with different group velocities through the fiber. Fiber optimized for 1310 nm wavelength transmission has a zero-dispersion wavelength at 1310 nm and dispersion of up to 17 ps/km/nm at 1550 nm. The higher dispersion at 1550 nm is a significant problem in longer links.
The effects of dispersion are particularly pronounced when present with any type of source chirp. The term "chirp" generally refers to the amount of frequency modulation or phase modulation of the optical signal induced by the intensity modulation process. In CATV networks, the combination of either type of chirp and fiber dispersion results in a significant dispersion penalty as the peaks of the AM signals are delayed relative to the valleys, resulting in unacceptable signal distortion. This distortion primarily manifests itself in a frequency-dependent contribution to system level composite second-order distortion ("CSO"). In a distributed feedback ("DFB") laser, applying the signal directly to the laser leads to a large amount of chirp. Accordingly, direct modulation of the 1550 nm source laser is generally not a viable option in most existing fiber optic links having fibers optimized for 1310 nm. Instead, relatively low chirp external modulation is typically used.
One commonly used external modulator is known as the Mach Zehnder modulator. When driven in the push-pull configuration, the Mach Zehnder produces an output signal exhibiting substantially zero chirp. Although beneficial for reducing the effects of dispersion in 1550 nm links, a drawback of the conventional Mach Zehnder modulator is that its output does not exhibit a high degree of linearity. In practice, second order distortion is eliminated by appropriately biasing the modulator. However, separate electrical predistorters are used to eliminate third order distortion. By using separate components for linearization, however, environmental stability, component matching and other considerations must be taken into account.
In response to the limitations in the linearity of the traditional Mach Zehnder, optically linearized modulators have been developed that include a pair of cascaded Mach Zehnder modulators. By appropriately varying the bias current and the amount of modulating voltage applied to each of the individual Mach Zehnders, both second and third order distortion components are canceled and a relatively high degree of linearity can be achieved, obviating the requirement of a separate predistorter. One drawback of these modulators for 1550 nm links is that they induce a substantial amount of modulator chirp, approximately 3.5 MHz/1% OMI at 500 MHz, where OMI refers to the optical modulation index. Although the magnitude of the modulator chirp is generally much smaller than that of laser chirp in a DFB laser, in conjunction with dispersion the modulator chirp caused by the cascaded Mach Zehnder detrimentally affects performance in any 1550 nm link having 1310 nm optimized fibers, and is unacceptable in longer, high performance 1550 nm links.
In view of the foregoing, those skilled in the art would prefer an optical modulator that exhibits the relatively high degree of linearity of the cascaded Mach Zehnder modulators, and also the substantially zero chirp exhibited by the traditional Mach Zehnder. Such a modulator would be particularly useful for enabling 1550 nm transmission over existing fibers having a zero-dispersion wavelength centered around 1310 nm.
Summary of the Invention
The present invention provides a novel optical modulator that uses optical predistortion to maximize linearity and also produces signals having substantially no modulator chirp. In a presently preferred embodiment, the optical modulator includes first and second optical interferometer stages, each stage also referred to as phase/intensity modulator. An input Y junction receives an optical signal from a source laser and splits the optical signal into a pair of waveguide paths (also referred to as interferometer arms) in the first interferometer. An electrode structure is provided that enables RF modulation through introduction of a first asymmetric electro-optical effect on the split signals. The asymmetric electro-optical effect is accomplished, for example, by varying the strength of the electric field applied by the electrodes across each of the arms, causing both intensity modulation and a residual phase shift in the optical signals. The signals are then mixed in unequal proportions in a first directional coupler between the first and second interferometers. In the second interferometer, the mixed signals are again separated into two additional interferometer arms having an associated electrode structure that enables further RF modulation through introduction of a second asymmetric electro-optical effect on the separated signals. The separated signals are then remixed in unequal portions in a second directional coupler which provides a pair of optical output signals.
Linearity is achieved in the output signals through optical predistortion by adjusting the biases and the relative amounts and phases of RF modulating voltage applied to the electrode structures such that any distortion components introduced by the first interferometer are canceled in the second interferometer. In addition, at least one of the output signals is made chirp-free by applying the asymmetric phase shifts in conjunction with the proportion of mixing in the directional couplers in such a way that any chirp introduced into the split optical signals by the first interferometers is substantially compensated for in the second interferometer and the directional couplers and is eliminated in at least one of the output signals. Signal modulation and linearity are not affected by the asymmetric phase shifts because there is no change in the net, or relative, phase shift between the split signals, only a change in the absolute phase shift applied to the individual signals. The modulated, chirp-free optical signal can then be introduced into an optical fiber having a zero-dispersion wavelength that is substantially different from the operating wavelength of the source transmitter without incurring a chirp-induced dispersion penalty.
Brief Description of the Drawings
These and other features and advantages of the present invention will become more apparent in light of the following detailed description read in conjunction with the accompanying drawings, in which:
FIG. 1 is a block diagram of a fiber optic system including an external optical modulator;
FIG. 2 is a schematic diagram of a conventional Mach Zehnder modulator;
FIG. 3 is a schematic diagram of an exemplary conventional optically linearized modulator consisting of a pair of cascaded Mach Zehnder modulators;
FIG. 4 is a presently preferred embodiment of the optical modulator according to the present invention in which the RF modulating electrodes are offset from the center of the waveguide paths to provide asymmetric electro-optic effects;
FIG. 5 is a block diagram illustrating the operation of the optical modulator according to the present invention
FIG. 6 is a view, partly in cross-section, of the optical modulator of FIG. 4;
FIG. 7 is a cross-sectional view of an X-cut modulator according to the present invention;
FIG. 8 is a cross-sectional view of a Z-cut modulator according to the present invention;
FIG. 9 is schematic view of another alternate embodiment of the optical modulator according to the present invention in which the effective length of the electrodes across one of the interferometer arms in each modulator stage is varied to provide the asymmetric electro-optic effects;
FIG. 10 is a schematic view of still another alternate embodiment of the present invention in which separate pairs of RF electrodes with independently adjustable inputs are provided on each interferometer arm to enable introduction of the asymmetric electro-optic effects;
FIG. 11 is a schematic view of still another alternate embodiment of the present invention in which a pair of RF electrodes is provided adjacent the modulator input to enable introduction of a phase shift to the input;
FIG. 12 is a schematic view of still another embodiment of the present invention in which separate pairs of RF electrodes are provided on each of the modulator outputs to enable phase correction on each of the outputs;
FIG. 13 is a block diagram of an externally modulated transmitter according to the present invention in which one of the optical output signals is used to provide feedback for optimizing modulator linearity;
FIGS. 14a-c are qualitative graphical depictions of residual phase modulation vs. intensity modulation for the conventional optically linearized modulator and the modulator according to the present invention; and
FIG. 15 is a qualitative graphical depiction of chirp vs. amount of asymmetry for exemplary embodiments of the present invention.
In FIGS. 4-10, similar reference numerals are used to designate similar parts. Similar reference numerals are also used to designate similar parts in FIGS. 3, 12 and 13.
Detailed Description
Referring to FIG. 1, a fiber optic communications link includes an optical source 2, such as a distributed feedback ("DFB") laser, for producing a 1550 nm wavelength continuous wave ("CW") optical carrier signal. An external modulator 4 modulates the optical signal with an external time-varying RF signal, such as an amplitude modulated CATV signal. The modulated optical signal is introduced into conventional fiber optic cable 6 having a zero-dispersion wavelength centered about 1310 nm for transmission to a remote receiver 8. As is well-known, other components such as optical fiber amplifiers, isolators and repeaters (not shown) may also be included in the link.
Currently available external modulators include the single-stage balanced Mach Zehnder modulator and the optically linearized cascaded Mach Zehnder modulator. Referring to FIG. 2, the balanced Mach Zehnder modulator 20 is typically formed on a lithium niobate substrate (not shown) and includes an input 22 for receiving optical signal I.sub.IN. The signal is split at a Y-junction 24 into upper and lower parallel interferometer arms formed of spaced-apart waveguide paths 26, 28 and is recombined in a second Y-junction 30 which provides a modulated optical output signal I.sub.OUT. In a typical embodiment, an RF modulating electrode 34, such as a bias tee electrode, is positioned symmetrically between the upper and lower interferometer arms. A bias tee electrode refers to an electrode that is used to apply both an RF and a bias signal which are summed with a bias tee (not shown). Upper and lower ground electrodes 36, 38, which typically have the same effective length as one another, are each positioned parallel to the RF electrode opposite the respective arm, and are located at the same distance from the RF electrode.
In operation, an RF modulating voltage such as an amplitude modulated CATV signal, is applied to the RF electrode, causing time-varying oppositely directed electric fields to be applied to the interferometer arms. In accordance with the well-known electro-optic effect, the applied fields cause the index of refraction across the arms to vary with the changing amplitude of the modulating signal, thereby phase modulating the light in each of the arms. The phase shifts applied to the signals in each of the parallel arms are equal and opposite. As the light is then recombined at the second Y junction 30, the in-phase components of the signals are added and the out-of-phase components are subtracted, producing an intensity modulated output signal containing no residual phase modulation. Because there is only intensity modulation and no residual phase modulation, the single-stage balanced Mach Zehnder provides an output that is free of modulator chirp.
Although useful for providing a modulated output that does not contribute to dispersion-induced CSO, a drawback of the Mach Zehnder in wideband applications such as CATV transmission is that its output light power vs. modulating voltage transfer characteristic is nonlinear as a result of modulator nonlinearities and harmonic distortion. By appropriately adjusting the bias .alpha. to the RF input, cancellation of even order distortion can be accomplished. Reduction of third order distortion ("CTB") to acceptable levels, however, can only be obtained by the use of an electrical predistorter (not shown) that precisely compensates for the third order distortion generated by the modulator. A disadvantage of having to use separate external predistortion is that environmental instabilities, component matching and other considerations must be taken into account.
Referring to FIG. 3, the prior art cascaded Mach Zehnder modulator 48 relies on optical predistortion instead of external predistortion to linearize the modulator output. Because modulation and distortion correction both occur on the same lithium niobate substrate, the performance of the modulator is less susceptible to variation due to environmental instabilities. Exemplary cascaded Mach Zehnder modulators are disclosed, for example, in Skeie, U.S. Pat. No. 5,249,243, the contents of which are hereby incorporated by reference. One such modulator consists of two identical Mach Zehnder-type interferometers 50, 52 arranged in first and second stages. An optical input signal I.sub.IN is split in a Y-junction 54 into the first interferometer which includes substantially parallel upper and lower interferometer arms formed by spaced apart waveguide paths 58, 60 and an electrode structure consisting of an RF electrode 62, such as a bias tee electrode, and ground electrodes 64, 66. In some embodiments, due to manufacturing considerations involving acoustic resonances, for example, the upper and lower arms may be slightly tilted away from being parallel. Like the single balanced Mach Zehnder, the RF electrode is positioned symmetrically between the upper and lower arms and at substantially equal distances between the ground electrodes. The ground electrodes in some embodiments may be flared out slightly. The RF electrode 62 receives a first RF modulating voltage RF1 and associated bias .alpha.1 and induces oppositely directed electric fields across the first stage interferometer arms. The split signals from the first stage interferometer are then mixed in a first directional coupler 68, formed of closely spaced waveguide segments 70, 72 having coupling angle .gamma..sub.1, and separated into the second interferometer. The second stage Mach Zehnder also includes parallel upper and lower interferometer arms formed by spaced-apart waveguide paths 74, 76 and a symmetrically-positioned electrode structure consisting of an RF electrode 78 and ground electrodes 80, 82. The electrode structure receives a second RF modulating voltage RF2, which is proportional to RF1, and a bias .alpha.2 and induces oppositely directed electric fields across the second stage waveguide paths. A second directional coupler 86, formed of closely spaced waveguide segments 88, 90 having coupling angle .gamma..sub.2, is connected to the output of the second interferometer for coupling the parts of the signal and providing complementary modulated optical signals I.sub.OUT1 and I.sub.OUT2. For operation in the linearized configuration, the coupling angles, bias levels to each electrode and the splitting ratio between RF1 and RF2 are selected so that distortion introduced by the first interferometer is canceled in the second interferometer.
Although useful for providing optical predistortion, a drawback of the cascaded Mach Zehnder is that its modulated output signals exhibit both intensity modulation and residual phase modulation when the modulator is operated in the linearized configuration. The transfer matrix for each of the interferometers is given by: ##EQU1## where .phi. is the phase shift induced by the signal on the RF electrode and associated bias applied to the RF electrode. In some cases, as is well known, the bias may be applied through a separate bias electrode.
The transfer matrix for each of the directional couplers is given by: ##EQU2## where .gamma. is the coupling angle.
In operation, optical input signal I.sub.IN with amplitude "a" enters the modulator and is split 50:50 into signals a1 and a2 in the first modulator, each having a magnitude a/.sqroot.2. In the first modulator, the applied RF modulating voltage RF1 shifts the phase of the light equally and oppositely. The light is then interfered at the first directional coupler and modulated in the second interferometer to produce intermediate signals b1 and b2 before the second directional coupler. ##EQU3## where .phi..sub.1 =RF1+.alpha..sub.1, RF1 is the RF modulating signal and .alpha..sub.1 is the signal bias, and where .phi..sub.2 =RF2+.alpha..sub.2, RF2=kRF1 is the RF modulating signal, .alpha..sub.2 is the associated bias applied to the second RF electrode, and k is the splitting ratio RF2/RF1 between the first and second modulating voltages. In practice, the bias voltages on the modulator are set so that the values of .alpha.1 and .alpha.2 are nominally zero.
Assuming equal coupling angles .gamma..sub.1 =.gamma..sub.2 =.gamma., the outputs c.sub.i from the modulator are then:
Since the output intensity I.sub.OUT =.vertline.c.vertline..sup.2, the transfer function for the modulator is then: ##EQU4## For operation in the linearized configuration with optimal modulator sensitivity, it can be shown that .gamma..sub.1 =.gamma..sub.2 is about 63 degrees and k=0.5.
The optically linearized modulator functions essentially as two balanced Mach Zehnder modulators, where the distortion induced by the first modulator is canceled by the distortion of the second modulator. Viewed independently, each stage of the cascaded Mach Zehnder produces an output signal that is substantially chirp-free if .gamma..sub.1 =.gamma..sub.2 =45 degrees. When operated in the linearized configuration, however, the power transfer between the two arms is unequal, approximately 75:25, due to the non-45 degree coupling angles of the directional couplers. Consequently, unlike the balanced Mach Zehnder, the phase shifts introduced in the upper and lower interferometer arms of the cascaded Mach Zehnder do not cancel, resulting in output signals that include both intensity modulation and residual phase shift. The dashed line plot 94 in FIG. 14, obtained using commercially available MathCad.RTM. software, shows the residual phase shift in the output of the cascaded Mach Zehnder vs. intensity modulation, and demonstrates that the phase of the modulator is not constant, so the chirp, which is the derivative of the phase, is not zero. When transmitted through 1310 nm optimized fibers, this chirp contributes to unacceptable levels of dispersion-induced CSO distortion.
The present invention provides both the benefits of optical predistortion found in the cascaded Mach Zehnder modulator, and the substantially chirp-free output found in the single stage Mach Zehnder operated in the push-pull configuration. The present invention recognizes, among other things, that by modulating both the phase and intensity of the optical signal at each interferometer stage through the induction of asymmetric electro-optic effects, residual phase shift in at least one of the optical output signals can be significantly reduced and substantially eliminated while maintaining the benefits of optical predistortion.
Referring to FIG. 4, in a presently preferred embodiment of the optical modulator according to the present invention, a Y-junction 100 splits an optical carrier signal I.sub.IN into a first stage interferometer, or phase/intensity modulator, 102 having parallel upper and lower interferometer arms formed of waveguide paths 104, 106 and an electrode structure 108 for applying a first asymmetric phase shift to the split signal. The terms "upper" and "lower" as used herein are to for distinguishing the interferometer arms; no physical placement or location of the arms is implied. The electrode structure consists of an RF electrode 110, such as a bias tee electrode, and upper and lower ground electrodes 112, 114 that are all substantially parallel to one another and to the waveguide paths. Unlike the balanced Mach Zehnder, the RF electrode is offset from the centerline 115 of the upper and lower interferometer arms and is positioned at unequal distances d1 and d2, respectively, from the upper and lower ground electrodes. The first interferometer is coupled to a second stage interferometer, or phase/intensity modulator, 116 that is substantially similar to the first stage interferometer through a directional coupler 117 having a coupling angle .gamma..sub.1. The directional coupler mixes the split signals in closely-spaced waveguide segments 118, 119 and separates the signals into the upper and lower interferometer arms 120, 122 of the second interferometer. Similar to the first interferometer electrode structure, second interferometer electrode structure 124 for applying a second asymmetric phase shift to the split signal, consists of an RF electrode 126 and upper and lower substantially parallel ground electrodes 128, 130. The second interferometer RF electrode is also offset from the centerline of the upper and lower interferometer arms and is positioned at unequal distances d1 and d2, respectively, from the upper and lower ground electrodes. The signals in the upper and lower arms of the second interferometer are coupled in a second directional coupler 132 having a coupling angle .gamma..sub.2 for mixing the signals in closely-spaced waveguide segments 133, 134 and providing first and second output signals I.sub.OUT1 and I.sub.OUT2.
The first stage electrode structure 108 is coupled to first RF input RF1 for receiving a first RF modulating voltage and associated bias .alpha.1. Similarly, the second stage electrode structure 124 is coupled to second RF input RF2 for receiving a second RF modulating voltage and associated bias .alpha.2, where RF2=k.multidot.RF1, and k, as in the cascaded Mach Zehnder, is the RF splitting ratio between the voltages applied in the first and second stages.
The asymmetric phase shifts are accomplished in the preferred embodiment of FIG. 4 by applying electric fields of differing strength across each of the interferometer arms, the differing electric fields being produced as a result of the unequal distances d1 and d2 between the RF electrode and the upper and lower ground electrodes. In the preferred embodiment, d2=(1+x)/(1-x).multidot.d1, where x is indicative of the amount of asymmetry. Accordingly, the applied fields in the upper arm and lower arms are also related by the ratio (1+x)/(1-x), and induce corresponding absolute asymmetric phase shifts in the respective optical signal parts. Application of the absolute phase shifts in this manner ensures that the relative or net phase shift between the upper and lower arms is constant for proper intensity modulation of the optical signal. Since the asymmetric phase shifts in each interferometer strongly modulates both the phase and intensity of the optical signal at each stage of the modulator, the individual interferometers are also referred to as phase/intensity modulators.
The block diagram of FIG. 5 illustrates the operation of the present invention. The optical signal I.sub.IN is split evenly in the Y-junction 100 between the upper and lower arms 104, 106 of the first phase/intensity modulator 102. The first stage electrode structure 108 applies a first asymmetric relative phase shift .phi.1 to the optical signal by inducing a positive phase shift (lead) in the part of the optical signal I.sub.a1 in the upper arm and a negative phase shift in the part of the optical signal I.sub.a2 in the lower arm, the magnitude of the lead in this example being greater than that of the lag. As a result of the asymmetric phase shift, the optical signals delivered to the first directional coupler 117 include both intensity modulation and residual phase modulation. At the first directional coupler, signals I.sub.s1 and I.sub.s2 are formed, each having a phase that is the vector sum of a fraction of the lead from the upper path optical signal I.sub.a1 and a fraction of the lag from the lower path signal I.sub.a2. In the second phase/intensity modulator 116, the electrode structure 124 applies a second asymmetric relative phase shift .phi.2 to the optical signal by inducing a lead in the part of the signal in the upper arm to form signal I.sub.b1, and a lag in the part of the signal in the lower arm to form signal I.sub.b2, the magnitude of the lead again being greater than that of the lag. In the second directional coupler 132, signals I.sub.OUT1, and I.sub.OUT2, are formed, each having a phase that is the vector sum of fraction of the lead from signal from the upper path optical signal I.sub.b1 and a fraction of the lag from the lower path signal I.sub.b2.
As with the cascaded Mach Zehnder, modulator linearity is maximized by appropriately selecting the coupling angles of the directional couplers, adjusting the bias levels to the modulating electrodes and attenuating RF2 by a fixed and predetermined amount with respect to RF1. Assuming coupling angles of .gamma.1=.gamma.2=63 degrees for linear operation as in the conventional cascaded Mach-Zehnder, more than 50% of the light is coupled between the upper and lower waveguides of each coupler. Accordingly, in the upper output arm there is a greater mix of the lower input signal and a lesser mix of the upper input signal, while on the lower arm the reverse is true. By appropriately selecting the amount of asymmetry introduced at each stage in conjunction with the coupling angles, the chirp introduced by the first phase/intensity modulator can be reversed in the second phase/intensity modulator and in the directional couplers and substantially eliminated in one of the optical output signals. Intensity modulation and linearity of the optical output signals are not affected by the asymmetric phase shift because only the absolute phases of the parts of the signals transmitted through the interferometer arms are altered. The net, or relative, phase shifts .phi.1 and .phi.2 between the signals in the upper and lower arms, which causes the intensity modulation, is not affected by the asymmetry.
Following the nomenclature used in equations (1) through (5) for derivation of the transfer function for the cascaded Mach Zehnder, the intermediate signals b1' and b2' analogous to b1 and b2 in equation (3) are given by: ##EQU5## Inspection shows that:
Therefore, it follows that:
It can be seen that the output of the modulator, .vertline.c.sub.i '.vertline..sup.2 =I.sub.OUT ', is advanced in phase by a phasor which is proportional to x, but the intensity of the output is unchanged. Accordingly, for the optical modulator according to the present invention:
Since I.sub.OUT' =I.sub.OUT, this demonstrates that the asymmetry introduced by the modulator does not affect the intensity transfer function, and generates only chirp.
The choice of optimum x is dictated only by the free parameters of equation (6) which are the RF splitting ratio k and the coupling angle .gamma.. Since the optimum value of k depends on .gamma. for linear operation, the value of x is chosen in the preferred embodiment to cancel chirp. A numerical solution can be obtained for the optimum value of x as a function of .gamma. for the approximate range .phi..sub.1 .di-elect cons.(-0.3, 0.3) as shown in the following table, which suggests that a theoretical optimum value for the sum of k and x is approximately 0.88 for the exemplary embodiments indicated.
The parameters indicated in Table 1 are exemplary only, and may change depending on design specifications or desired results. Depending on the drive circuits and drive power, the optimum shifts. For example, at higher power, fifth order effects may have to be taken into account.
In practice, one first determines values for the splitting ratio and coupling angles which provide third order linearization. For example, for a given k, the only free parameter is the coupling angle .gamma., which is chosen to minimize the third order coefficient of .phi..sub.1. The x term can be ignored in this calculation since it does not affect third order linearization. Once .gamma. and k are determined, x is varied until the phase of the output does not change with input .phi..sub.1 (i.e., the residual phase in the output is constant).
FIGS. 14a-c show plots of residual phase modulation in the optical output signal I.sub.OUT1 vs. intensity modulation for the exemplary parameters indicated in the corresponding rows of Table 1. The graphs illustrate the substantially reduced residual phase modulation achieved through application of the asymmetric phase shifts of the present invention. In FIG. 14a, three separate values of x are plotted for the exemplary parameters k=0.5 and .gamma.=1.102. For values of x in the range from about 0.25 to 0.55, substantial improvement over the uncorrected case (x=0) is demonstrated. By substantial improvement, it is meant that there is approximately a 50% reduction in residual phase modulation in the output of the modulator relative to the uncorrected case. One of ordinary skill in the art, however, would readily appreciate that any realizable value of x greater than zero provides some improvement over the uncorrected case, but the closer x is to the optimum value, the greater the expected correction.
FIG. 15 is a plot of chirp vs. amount of asymmetry (x) for each of the exemplary sets of parameters indicated in Table 1. The chirp is defined as d.phi./dI, where .phi. is the phase of the output signal and I is the output intensity. Additional plots for different parameters can be obtained using commercially available MathCad.RTM. software.
Referring to FIG. 6, the modulator is constructed by conventional processes on a lithium niobate substrate 140. The interferometer arms 104, 106, 120, 122 in the first and second interferometers are formed by diffusing titanium into the substrate. RF electrodes 110, 136 and ground electrodes 112, 114, 128, 130 are formed, for example, by depositing a first thin adhesion layer of chrome and a second relatively thicker layer of gold or aluminum on top of the substrate. In this example, each of the electrodes has a U-shape, in which the legs are adapted for connection to RF input pads and ground pads on a printed circuit board (not shown). The bodies of the electrodes are positioned substantially parallel to one another and to the interferometer arms. It should be noted that the interferometer arms and electrode structures disclosed herein are merely exemplary and may be configured in a variety of different ways to implement the teachings of the present invention. For example, another way to fabricate these modulators is by diffusing hydrogen into the substrate. A variant of this process is referred to as annealed proton exchange, currently practiced by United Technologies Corporation, Hartford, Conn.
FIG. 7 is a cross-sectional view of an exemplary X-cut phase/intensity modulator according to the present invention. In this embodiment, the electrodes are deposited adjacent to and on the sides of the interferometer arms 104, 106. The distance d1 between one end of the RF electrode 110a and ground electrode 112a and the distance d2 between the other end of the RF electrode 110a and ground electrode 114a are varied to enable introduction of unequal magnitude electric fields generally directed sideways along the Z-axis. FIG. 8 is a cross-sectional view of one interferometer of a Z-cut chirp-free modulator according to the present invention. In this embodiment, the electrodes are deposited on top of the interferometer arms 104, 106. The distance d3 between one end of the RF electrode 110b and ground electrode 112b and the distance d4 between the other end of the RF electrode 110b and ground electrode 114b are varied to enable introduction of unequal magnitude electric fields generally directed upwardly and downwardly along the Z-axis.
The preferred embodiment of the present invention shown in FIGS. 4-9 in which unequal electrode spacings are used to accomplish the asymmetric electro-optic effects is advantageous because it can be manufactured through minor variation of conventional techniques for producing the exemplary cascaded Mach Zehnder modulator described herein. Furthermore, because the same mask is used to fabricate the first and second electrode structures, acoustic interactions and frequency response are substantially equal in both electrode structures.
Notwithstanding the apparent advantages of the preferred embodiment, the asymmetric relative electro-optic effects can be accomplished in numerous other ways besides through unequally spaced-apart electrodes. Generally speaking, the absolute phase shift induced in a given interferometer arm is: ##EQU6## where n.sub.o is the nominal optical index of refraction; r.sub.eff is the effective electro-optic coefficient in the direction of E; E is the net electric field applied across the arm; L is the length over which the field is applied; and .lambda. is the optical wavelength. In practice, the phase shift between the signals in the two interferometer arms can be varied by adjusting any of the parameters in equation (10) except the optical wavelength. The most practical parameters to adjust are the electrode length L or the electric field E. In the preferred embodiment of FIG. 4, the asymmetric phase shifts are induced by adjusting the relative electric fields applied to the upper and lower arms.
Referring to FIG. 9, in another embodiment of the present invention, the effective length of the electrodes across the lower interferometer arms 106, 122 in each respective phase/intensity modulator stage is varied relative to the electrodes in the upper arms to accomplish the asymmetric electro-optic effects. In this embodiment, the electric field E remains the same in each arm. In the exemplary embodiment shown, the RF electrode 149 for the first phase/intensity modulator stage 146 has an effective length at its upper end 150 that is greater than the effective length at its lower end 154. Upper and lower first stage ground electrodes 152, 156 have effective lengths that generally correspond to the lengths of the corresponding ends of the RF electrode. Similarly, the RF electrode 151 for the second phase/intensity modulator stage 148 has an effective length at its upper end 158 that is greater than the effective length at its lower end 162. Upper and lower first stage ground electrodes 160, 164 have effective lengths that generally correspond to the lengths of the corresponding ends of the RF electrode. Although shown in a generally trapezoidal shape, the shape of the electrodes should be taken into account for impedance matching and to minimize the effects of fields in other than the desired direction.
By varying the effective lengths for the respective electrodes associated with the upper and lower interferometer arms by the factor (1+x)/(1-x), where x does not equal zero, asymmetric electro-optic effects can be applied to the signals in each of the phase/intensity modulators for implementing the present invention. This embodiment has the advantage of having distances between the RF electrodes and the ground electrodes that are nominally the same.
Referring to FIG. 10, in another alternate embodiment of the present invention, separate pairs of RF electrodes with independently adjustable inputs are provided on each arm of the modulator to enable introduction of the asymmetric electro-optic effects. In this embodiment, a first RF input RF1 is connected to a first electrode set for the upper interferometer arm 104. The electrode set includes an RF electrode 196 and a parallel spaced-apart ground electrode 198 that together straddle the arm. Similarly, a second RF input RF1' is connected to a separate electrode set formed of RF electrode 200 and ground electrode 202 for the lower interferometer arm. By applying unequal voltages to RF1 and RF1', an asymmetric electro-optic effect can be accomplished in the first stage phase/intensity modulator 192. In the second phase/intensity modulator 194, an individual RF input RF2 is connected to another electrode set formed of RF electrode 204 and ground electrode 206 for the upper interferometer arm, and a separate input RF2' is connected to an electrode set formed of RF electrode 208 and ground electrode 210 for the lower interferometer arm. As in the first stage phase/intensity modulator, by applying unequal voltages to RF2 and RF2', an asymmetric electro-optic effect can be accomplished. This embodiment has the advantage of providing a means for adjusting the asymmetric phase shifts in each arm in order to compensate for fabrication errors in each stage and component mismatches, between the coupling angles of first and second directional couplers, for example. It has the further advantage of allowing fine tuning of the modulator to completely eliminate any chirp in the output signal merely by adjusting the RF inputs to each electrode set.
In some situations, it may be useful to unbalance the asymmetric electro-optic effects applied to the split signal in each of the stages. In such a case, the value of x is varied slightly in each stage. As will be appreciated by one of ordinary skill in the art, the particular values of x applied in each modulator stage should not effect the net chirp. In an embodiment having unbalanced stages the average value for x can be taken into account. Accordingly, in any of the embodiments shown, the amount of asymmetry in the first stage can be viewed as x, and the amount of asymmetry in the second stage can be viewed as y. The average of x and y can be used as a rough approximation in place of x in the modulator transfer function for determining the amount of correction.
FIGS. 11 and 12 illustrate still other embodiments of the present invention. The invention realizes, among other things, that the output of the modulator can be multiplied by a phasor as shown in equation (9) to compensate for the inherent residual phase modulation introduced by the modulator. As shown in FIG. 11, it follows that one can use a phase modulator 210 to apply a phase shift to the input to the modulator to reduce residual phase modulation in at least one of the outputs. Similarly, as shown in FIG. 12, the phase shifts can be applied by phase modulators 215, 217 applied to the upper and lower outputs of the modulators, respectively to achieve the desired correction in each of the outputs. These embodiments have the added advantage of being relatively easy to manufacture, since they require only additions to existing cascaded modulators. However, they also have the disadvantage of requiring a longer lithium niobate substrate on which to place the additional modulator or modulators.
Referring to FIG. 13 the optical modulator according to the present invention can also be used in a feedback control system 240 in which one or more of the optical output signals 252, 255 is used to provide feedback for optimizing modulator linearity. In this system, a CW laser 242 provides an optical signal I.sub.IN to an external modulator 244 designed according to the teachings of the present invention. An RF modulating voltage carrying information for modulating the optical signal is split in a splitter 246 including power driver 248 and power splitter 250 to provide first and second modulating signals RF1 and RF2 to the modulator. In the embodiment of FIG. 10, additional splitters (not shown) can be employed to further split RF1 and RF2 for application to the upper and lower electrode sets. The modulator provides a pair of optical output signals I.sub.OUT1 and I.sub.OUT2.I.sub.OUT1, which is free of any residual phase modulation, can be introduced into a 1310 nm optimized optical fiber 252 for transmission without chirp-induced dispersion penalty to a remote optical receiver (not shown). Both I.sub.OUT1 on cable 252 and I.sub.OUT2 on cable 255, are provided as feedback to a monitor 254 which monitors the signal second and third order distortion components. Information gathered from the feedback signal is then used in a CSO/CTB processor 256 to adjust the splitting ratio k between RF1 and RF2 and the DC bias levels applied to each stage of the modulator for optimizing linearity. In practice, based on distortion in the output signal the bias is adjusted, to minimize second and higher even order effects, and the splitting ratio is adjusted to minimize third order effects and higher odd order effects, if present.
Although various embodiments of the present invention have been described herein, those skilled in the art will realize that numerous modifications can be made without departing from the spirit and scope of the present invention. For example, the Y-junction used to initially split the signal can be replaced by a directional coupler. Therefore, the scope of the present invention should not be limited but should instead be read in light of the following claims.