Patent Grant
7840100
The field of the invention is optical communication and processing.
Light traveling in optical networks often suffers from chromatic dispersion, or polarization mode dispersion, which can lead to degradation of signals and unacceptably high error rates. In order for such networks to function, digital signals are converted into electrical signals, and back into fresh optical signals, at intervals short enough so that error rates are acceptably small. But such optical-electrical signal converters are expensive, and may dominate the cost of the network, and may slow it down. Methods of optically compensating for dispersion can help to reduce such costs, and/or increase speed, by allowing optical signals to propagate for greater distances before they become degraded.
Chromatic dispersion in optical networks, i.e. group delay which varies with frequency, is often compensated for by dispersion compensation fibers, as described, for example, in the article on “Dispersion Compensation” in the online Encyclopedia of Laser Physics and Technology, http://www.rp-photonics.com/dispersion_compensation.html, the disclosure of which is incorporated herein by reference. Such fibers generally produce dispersion compensation that is a fixed function of frequency, depending on the composition of the fiber, even if the requirements for dispersion compensation are changing in time, as occurs, for example, in reconfigurable optical networks. The dispersion compensation as a function of frequency generally provides a good match to the required dispersion compensation only over a limited bandwidth, because there are only a limited number of parameters, in the composition of the fiber, that can be used to adjust the dispersion as a function of frequency. This limits the bandwidth that can be used, and may limit the use of a given dispersion compensation fiber to only a single channel. Dispersion compensation fibers generally do not provide compensation for polarization mode dispersion. They may also be expensive and bulky, since a long length of fiber, as much as several kilometers, may be needed, and a given fiber can only compensate for a fixed degree of dispersion.
Gires-Tournois (GT) etalons are devices made of a transparent material with finite index of refraction n, with a fully reflecting back surface, and a partially reflecting front surface of reflectivity R, parallel to the back surface. Light entering the front surface reflects multiple times inside the etalon, and the interference of the beam components undergoing different numbers of reflections produces an output beam that has very little attenuation, but has a phase delay that is a nonlinear function of frequency, depending on the relationship of the wavelength of the light to the thickness d of the etalon. The GT etalon has a chromatic dispersion that is a periodic function of frequency, that depends on the index of refraction n, the thickness d, and the reflectivity R. An article at htpp://en.wikipedia.org/wiki/Gires-Tournois_etalon, downloaded on Feb. 14, 2007, the disclosure of which is incorporated by reference, provides more details of how GT etalons work, and describes some of their uses, for example in pulse compression, and in nonlinear Michelson interferometers.
An aspect of an embodiment of the invention concerns a dispersion compensation module in which a beam of light, after entering the dispersion compensation module, interacts with a dispersion compensating element, exits the element, and reflects back to the element, interacting with the element again.
Optionally the dispersion compensating element is a GT etalon, and the dispersion compensation produced by the module is greater than the dispersion compensation from a single interaction with a GT etalon, because the light interacts with it more than once. Optionally, the change in frequency over which the dispersion compensation repeats, due to the periodicity of the dispersion of the GT etalon as a function of frequency, is equal or nearly equal to the difference in frequency of adjacent channels used by an optical network in which the dispersion compensation module operates.
In some embodiments of the invention, the module comprises a block of a birefringent material, and the light beam, or at least a polarized component of it, passes through the block on a path from an input location on a first side of the block, to the dispersion compensating element on a second side of the block, opposite the first side. After interacting with the dispersion compensating element, the light travels back through the block, but with a changed polarization, for example rotated by 90 degrees, following a different path, oriented at a changed angle, to a directing element on the first side of the block, laterally displaced from the input location. The directing element, for example a mirror, or another GT etalon, directs the light back through the block to the second side, but with the polarization changed again, for example back to its value when the light crossed the block the first time, so the light again follows a different path, oriented at a changed angle, and the light returns to the dispersion compensating element laterally displaced from the location where the light previously interacted with the dispersion compensating element.
Optionally, the two sides of the block are substantially planar surfaces, and substantially parallel to each other. Optionally, the light propagates substantially perpendicular to the planar surfaces, just before it enters and just after it leaves the block, each time it passes across the block, even though the path inside the block is sometimes at an oblique angle to the surface, due to the birefringence.
Optionally, the polarization is changed by quarter wave plates, or equivalent elements, located on each side of the block, which the light passes through twice, each time it interacts with the dispersion compensating element or the directing element.
This process is optionally repeated a well-defined number of times, with the light bouncing back and forth between the two sides of the block, laterally displaced on each bounce, until it reaches an output port and leaves the dispersion compensation module. The number of bounces depends on the location of the output port, and on the difference in angle of orientation of the path in the birefringent material, for the different polarization states of the light.
In some embodiments of the invention, the dispersion compensation characteristics of the module are controlled in real time. For example, the dispersion compensation characteristics of the dispersion compensating element are controlled in real time, for example by controlling the temperature of the element. Alternatively or additionally, means are used to control the number of times the light interacts with the element, in real time. Optionally, feedback is used to optimize the dispersion compensation characteristics of the module in real time, for example by minimizing the bit error rate in an optical network, or by minimizing a more direct measure of dispersion.
In some embodiments of the invention, an input light beam, optionally unpolarized, is split into two polarization components before it enters the dispersion compensation module, and each polarization component, when it enters the module, interacts with a different dispersion compensation element, with different characteristics, thereby compensating for polarization mode dispersion in the input beam.
In some embodiments of the invention, the dispersion compensation module comprises a plurality of dispersion compensation elements, optionally with different characteristics, providing additional degrees of freedom in the dispersion compensation as a function of frequency.
An aspect of some embodiments of the invention comprises a device for lateral displacement of a light beam, or at least a polarized component of it, using a block of birefringent material with two sides, optionally substantially parallel planar surfaces. The light beam enters the block at a first location on the first side, and at least the component of the light beam crosses the block on a first path, to a first exit location on the second side. The light exits the block, and is directed back to the block by a directing element, such as a mirror, re-entering the block, optionally at the first exit location, but with changed polarization, for example rotated by 90 degrees. The polarization is changed by a polarization changing element, for example, by passing twice through a quarter wave plate, or equivalent element, between the second side of the block and the directing element. The light then crosses the block back to the first side, but following a different path, with a changed angle of orientation, due to the changed polarization, and reaches a second exit location on the first side, laterally displaced from the first location where the light entered the block on the first side.
Optionally, there is another directing element, such as a mirror, and optionally there is another polarization changing element, on the first side, and the light is again directed back to the block with its polarization changed, for example back to the polarization it had the first time it crossed the block. The process optionally repeats, with the light bouncing back and forth across the block, and becoming laterally displaced each time, until it exits the block for a last time, at an exit port.
Optionally, the light propagates substantially perpendicular to the planar surfaces, just before it enters and just after it leaves the block, each time it passes across the block, even though the path inside the block is sometimes at an oblique angle to the surface, due to the birefringence.
The device may displace the light by any of a plurality of different closely spaced distances, depending on the location of the exit port. Optionally, the direction of displacement of the light depends on its polarization when it enters the device. Optionally the light leaves through one of two exit ports, depending on its polarization when it enters the device. Optionally, there are a plurality of output ports, and one or both of the mirrors are movable, so that the choice of output port, and the displacement distance, is controllable in real time.
There is thus provided, in accordance with an exemplary embodiment of the invention, a method of laterally displacing at least a component of a light beam, the method comprising:
Optionally, directing the beam to enter the block comprises directing the beam to propagate substantially perpendicular to a surface of the block just before entering the block.
Optionally, directing the beam component to pass back through the block comprises directing the beam component to propagate substantially perpendicular to a first surface of the block just before entering the block, and to propagate substantially perpendicular to a second surface of the block which is substantially parallel to the first surface, just after passing through the block.
Optionally, directing the beam component to pass back through the block comprises directing the beam component back to substantially a same location where the beam component last exited from the block.
In an embodiment of the invention, the method comprises:
the second component of the beam thereby being laterally displaced from the first location when it exits the block a last time, in a direction opposite to the displacement of the first beam component.
There is further provided, according to an exemplary embodiment of the invention, a method of changing the dispersion of a light beam, the method comprising:
Optionally, the method includes directing the light beam to enter a block of birefringent material at a first entry location and to pass through the block on a first path to a first exit location, before causing the light beam to enter the dispersion compensating element the first time.
Optionally, the method includes directing the light beam to pass back through the block, in a changed polarization state, from the first exit location, following another path oriented at a changed angle due to its changed polarization state, to a location on the same side of the block as the first entry location, but laterally displaced.
In an embodiment of the invention, directing the light beam back to the dispersion compensating element each of the at least one times comprises:
There is further provided, according to an exemplary embodiment of the invention, a device for laterally displacing a beam, comprising:
wherein the device is configured so that when the beam enters the block from the input port, at least a component of the beam passes through the block to the first exit location, and the beam that re-enters the block at the first exit location follows a path oriented at a changed angle, due to its changed polarization, passing back through the block to the first side, reaching a second exit location displaced laterally from the first location.
Optionally, the device also includes:
Optionally, the first and second reflecting elements, and the first and second polarization changing elements, each extend laterally far enough along the block on their respective sides, so that the light passes back and forth through the block between successive exit locations on the first and second sides a plurality of times, being further laterally displaced each time.
There is further provided, according to an exemplary embodiment of the invention, a dispersion compensation device comprising:
Optionally, the dispersion compensating module comprises:
Optionally, the input module directs the light to the dispersion compensating element through the block of birefringent material from the second side to the first side, and the dispersion compensating module comprises a second polarization changing element situated between the first side of the block and the dispersion compensating element, which polarization changing element changes the polarization of the light interacting with the dispersion compensating element, so that after interacting the light follows a different path in the block, reaching the directing element instead of returning to the input module.
Optionally, the directing element comprises a mirror.
Optionally, the dispersion compensating element comprises at least one GT etalon.
Optionally, the directing element comprises a second dispersion compensating element.
Optionally, the second dispersion compensating element comprises at least one GT etalon.
Optionally, the at least one GT etalon comprises at least a first and a second GT etalon, with different characteristics, arranged so that the light interacts with the second GT etalon after interacting with the first GT etalon.
Optionally, the device comprises a control module which controls the degree of dispersion compensation, by controlling the temperature of at least one of the at least one GT etalons.
Optionally, the device comprises a heater, and a thermoelectric cooler, to control said temperature.
In an embodiment of the invention, the device includes a control module which controls at least two parameters of the dispersion compensating module.
Optionally, the first dispersion compensating module comprises at least two GT etalons, and the two parameters comprise dispersion characteristics respectively of the two GT etalons.
There is further provided, according to an exemplary embodiment of the invention, a multi-channel optical network with a substantially constant channel spacing, the network comprising a dispersion compensation device according to an embodiment of the invention, wherein a free spectral range of the at least one GT etalons is substantially equal to the channel spacing.
Optionally, the polarization changing element comprises a quarter wave plate equivalent through which the light passes on the way to the directing element, and through which the light passes again on the way back to the block from the directing element.
Optionally, the input module comprises a polarizing beam splitter.
Optionally, the input module is configured to convert an unpolarized light beam entering it into two parallel polarized light beams which enter the dispersion compensation module.
Optionally, the device includes an output beam directing element which directs the two parallel polarized light beams leaving the dispersion compensating module back through the input module, such that the polarizing beam splitter combines them into an unpolarized output light beam.
Optionally, the device includes an output module, configured to receive the two parallel polarized light beams leaving the dispersion compensating module, and to combine them into an unpolarized output beam.
Optionally, the dispersion compensating module is configured to change the dispersion of each of the two parallel light beams by a different amount, thereby providing polarization mode dispersion compensation to the output beam.
In an embodiment of the invention, the dispersion compensating module is configured to change the group delay of each of the two parallel light beams by a different amount, thereby providing polarization mode dispersion compensation to the output beam.
There is further provided, according to an exemplary embodiment of the invention, an optical network comprising a dispersion compensation device according to an embodiment of the invention, wherein at least one parameter of the dispersion compensating module is chosen to maximize one or more of Q-factor and eye-opening, or minimize bit error rate of the network, or both.
Optionally, the device includes a control module to control the degree of dispersion compensation provided by the dispersion compensating module, in real time.
In an embodiment of the invention, the dispersion compensation module comprises a movable mirror which covers the dispersion compensating element to an adjustable extent, the covering portion blocking the light from interacting with the dispersion compensating element, thereby controlling the degree of dispersion compensation.
Optionally, the device includes a control module which controls the movable mirror in real time.
There is further provided, according to an exemplary embodiment of the invention, an optical network comprising a dispersion compensation device according to an embodiment of the invention, wherein the control module is configured to use a feedback loop to control one or more of Q-factor, eye-opening, and bit error rate in the network, by controlling the degree of dispersion compensation provided by the dispersion compensation device.
There is further provided, according to an exemplary embodiment of the invention, an optical network comprising a dispersion compensation device according to an embodiment of the invention, wherein the control module monitors the wavelength of light in the network, and controls the degree of dispersion compensation depending on the wavelength.
The above and other features of the invention including various novel details of construction and combinations of parts, and other advantages, will now be more particularly described with reference to the accompanying drawings and pointed out in the claims. It will be understood that the particular method and device embodying the invention are shown by way of illustration and not as a limitation of the invention. The principles and features of this invention may be employed in various and numerous embodiments without departing from the scope of the invention.
In the accompanying drawings, reference characters refer to the same parts throughout the different views. The drawings are not necessarily to scale; emphasis has instead been placed upon illustrating the principles of the invention. Of the drawings:
In describing the drawings, language will sometimes be used to the effect that a light beam undergoes a first event, for example an interaction with or passage through an optical element, followed in time by a second event. It should be understood that such language means that the second event would follow the first event from the point of view of an observer traveling along the light beam in the direction that the light is propagating. But at any given time, both of the events may be taking place simultaneously, to different parts of the beam.
Device 100 may be incorporated into an existing transceiver or transponder used in an optical network, or it may be used as a stand-alone device, optionally in series with other plug-in modules in an optical network, for example an amplifier to compensate for beam attenuation An input light beam, propagating for example in an optical fiber 102, enters an input/output module 104. Light from the input light beam, optionally split into polarized beams by input/output module 104 as will be described below in
Beam splitter 306 splits the input light beam into two parallel linearly polarized beams 402 and 404. In the embodiment of the invention shown in
In some embodiments of the invention, the input beam from optical fiber 102 is polarized to begin with, and there is no polarizing beam splitter. In other embodiments of the invention, the input beam is unpolarized, there is no separate polarizing beam splitter, and dispersion compensating module 106 only allows one polarization component of the input beam to pass through. However, it is potentially advantageous to have polarized beams entering dispersion compensating module 106, so that all of the beam will pass through it.
Returning to the description of
It should be noted that, instead of using thin-film beam splitter 308, a mirror may be used to reflect output beam back toward polarizing beam splitter 306. In this case, the mirror is positioned out of the path of the input beam, so that the input beam is not deflected on its way from polarizing beam splitter 306 to dispersion compensating module 106.
In some embodiments of the invention, the direction of the light beam shown in
When the output beam reaches polarizing beam splitter 306, one of the two parallel beams optionally passes back through half-wave plate 406, or another half-wave plate, so that the two parallel beams again have two different linear polarizations, oriented 90 degrees apart. Polarizing beam splitter 306 then combines the two beams back into a single unpolarized beam, assuming that the input beam was originally unpolarized when it entered polarizing beam splitter 306. The unpolarized output beam then passes back through collimator 304, and is directed to optical fiber 108.
In dispersion compensation device 200, there are no reflectors 308 and 310, or half-wave plates 312 and 314. Instead, the output beam emerging from dispersion compensating module 106 goes directly to another polarizing beam splitter and collimator which comprise output module 204, similar to polarizing beam splitter 306 and 304 which comprise input module 202, and the output beam then passes to optical fiber 108.
Between the two etalons is a block of birefringent material 506, with quarter wave plates 512 and 513 respectively located between block 516 and GT etalons 502 and 504. The GT etalons each have a highly reflecting surface 510 on the back, with reflectivity greater than 0.98 or greater than 0.995, for example, and a partially reflecting surface 508 on the front with reflectivity R. For example, R is less than 0.01, or between 0.01 and 0.05, or between 0.05 and 0.2, or between 0.2 and 0.9, or greater than 0.9. Such reflectances are obtained, for example, by coating the surfaces with appropriate thin layers. Both surfaces are optically flat and parallel to each other, separated by a distance d, with a transparent material in between, with an index of refraction n. The reflecting surfaces of each etalon are also parallel to the surface of block 506 that they face, and to the opposite surface of the block and the reflecting surfaces of the other etalon on the other side of the block. Etalon 504, however, has reflecting surfaces 508 and 510 missing from regions 516 and 528, at its two ends, with only the transparent material present there. Quarter-wave plate 513 also does not extend to the ends of block 506. Input beam 514 passes through region 516 of etalon 504, and then passes through birefringent block 506, missing quarter-wave plate 513. The polarization of input beam 514 is initially linear, and in a direction such that the beam follows a path 518, straight through birefringent block 506, normal to the surface of block 506.
The input beam then passes through a quarter-wave plate 512, becoming circularly polarized, interacts with etalon 502, and passes back through quarter-wave plate 512, becoming linearly polarized again but in a different direction. Although, for clarity, the light beam leaving etalon 502 is shown slightly displaced from the light beam entering etalon 502, and similarly for etalon 504, the light beam may enter and leave the etalon at the same location. When the light beam passes back through quarter-wave plate 512 the second time, its polarization will be rotated 90 degrees from the direction it had when traveling on path 518. The birefringent nature of the block 506 will thus cause the light to travel at a small angle to path 518, on a path 522, instead of retracing path 518. Path 522 reaches the other side of block 506 at a location 524, where quarter-wave plate 513 and the reflecting layers of etalon 504 are present.
The light beam then passes through quarter-wave plate 513, interacts with GT etalon 504, and passes back through quarter-wave plate 513. The two passes through quarter-wave plate 513 brings the polarization of the light beam back to the direction it had initially, when it followed path 518. The light then follows a path 526, parallel to path 518. The process repeats a plurality of times, as the light bounces back and forth between the two etalons, interacting with each one, and being displaced laterally in position, shown as a vertical displacement in
It should be understood that, in module 500, and in any of the devices described below which use quarter wave plates, one or more of the quarter wave plates may be replaced by any other optical element which causes linearly polarized light to rotate its direction of polarization by 90 degrees, if the light passes through the element twice. A general element satisfying this condition will be referred to herein as a “quarter wave plate equivalent.” For example, a Faraday rotator element can be used which rotates the direction of polarization of linearly polarized light passing through it by 45 degrees. Even when it is not explicitly stated that such a “quarter wave plate equivalent” can be used, it should be understood that it can be used instead of a quarter wave plate, wherever a quarter wave plate is mentioned.
In some embodiments of the invention, light beam 514 has a different linear polarization direction, such that paths 518 and 526 are at an oblique angle to the surface of block 506, rather than being normal to it, and path 522 is normal to the surface of block 506. In those embodiments of the invention, block 506 is oriented differently, if input light beam 514 still enters block 506 at the upper right, so that the obliquely propagating light beams go down and to the left, instead of down and to the right as they do in
It should be noted that, as long as the beam is displaced laterally by much more than its own Gaussian width from one interaction with one of the etalons, and the next interaction with the same etalon, then the different interactions will not interfere appreciably with each other, and the group delay in the beam, from each interaction with the etalon, can be calculated separately, using Eq. (3) below, without considering the other interactions.
Optionally, this criterion is used to determine the maximum number of times that the beam can interact with each etalon. The Gaussian width of the beam increases as the beam propagates back and forth across the block, for example at a spreading angle of λ/2a, where λ is the wavelength of the light, and a is the initial radius of the beam, for example at the input port to the dispersion compensating module. The width of the block is equal to the lateral displacement of the beam on each bounce, divided by the change in angle of propagation Δθ due to the birefringence of the block. And the distance of propagation of the beam is twice the width of the block, for each bounce. If we require the beam to be displaced laterally by at least 4a between bounces, then we find that the beam will double in width after (Δθ)a/2λ bounces. For example, if a=0.5 mm, Δθ=0.05 radians, and λ=1000 nm, then the beam can make 12 bounces before its width doubles. Beyond that number of bounces, portions of the beam from successive bounces may substantially overlap in the etalon, and may significantly interfere with each other. Such overlap can be avoided if the block is about 4a/Δθ=40 mm wide, and about 2(Δθ)a2/λ=25 mm long.
One might contemplate replacing block 506 in module 500 by empty space, and directing input beam 514 into this space at an angle slightly oblique to the horizontal, so that the beam continues to travel slightly downward as it bounces back and forth between etalons 502 and 504. However, using birefringent block 506, and directing input beam 514 so that it enters block 506 perpendicular to its surface, has the potential advantage that the light will enter the etalons perpendicular to their reflecting surfaces, which allows the etalons to function as designed, with their full group delay as a function of frequency. The light will also exit block 506 perpendicular to its surface, which may make it easier to direct the light into an optical fiber, or another optical element, than if the light were to exit device 500 at an oblique and possibly variable angle. The interface between an optical fiber carrying the input beam, and block 506, may also be simpler to design and more reliable, if the light enters block 506 perpendicular to its surface, rather than entering device 500 at an oblique angle.
The index of refraction n of one or both of the GT etalons may be changed by changing its temperature. The distance d may also change, due to thermal expansion. Due to both of these effects, changing the temperature of GT etalons 502 and/or 504 may be used to control the dispersion, as function of frequency, that they produce in the light beam each time it interacts with them. In some embodiments of the invention, GT etalons 502 and/or 504 are heated and/or cooled by thermoelectric devices 532, or by any other means of heating and/or cooling known to the art, in order to control the dispersion compensation. Optionally, this may be done in real time, as will be described in
It should be understood that the dispersion compensating modules shown below in
At least one of the mirrors, for example mirror 652 in the case illustrated in
By moving the moveable mirror, for example mirror 652, relative to the other mirror, for example mirror 653, the path length traveled by beam 402 is made different from the path length traveled by beam 404, thereby compensating for polarization mode dispersion of the original input beam. As may be seen in
It should be noted that the devices shown in
As used herein, “dispersion” is a general term that includes both chromatic dispersion, which refers to the group delay depending on frequency, and polarization mode dispersion, which refers to the group delay depending polarization mode. As used herein, “dispersion compensating element” includes elements such as the set of mirrors 652 and 653, which change the relative group delay of beams 402 and 404, resulting in a change in the polarization mode dispersion of the output beam emerging from the dispersion compensating device, after one of the beams has passed through half-wave plate 406, or a different half-wave plate, and beams 402 and 404 have been recombined into a single beam by polarizing beam splitter 306, as shown in
In module 800, instead of single GT etalons 502 and 504, one on each side of block 506, there are a plurality of GT etalons, for example two GT etalons 804 and 808, on the left side of block 506, and a plurality of GT etalons, for example two GT etalons 802 and 806, on the right side. Alternatively, there are different numbers of GT etalons on each side, for example one GT etalon on one side, and two on the other side. By using a plurality of GT etalons with different characteristics, and possibly with changeable characteristics that can be controlled independently, module 800 may enable better matching of dispersion compensation as a function of frequency, to the dispersion as a function of frequency found in the input beam. As a result, the residual dispersion of the output beam (group delay ripple) may be lower, and/or the dispersion compensation may be effective over a greater frequency range, allowing the optical network to function at greater channel bandwidth.
Module 800 also optionally includes a moveable mirror 810, which can cover an adjustable portion of GT etalons 804 and 808, ranging for example from 0% to 100%. For the covered portion of the GT etalons, light reflects from mirror 810, instead of interacting with the GT etalon. The position of mirror 810 affects the number of times the light interacts with a GT etalon, and consequently affects the degree of dispersion compensation in output beam 812. Mirror 810 is optionally moved by a motor, for example a motor with the characteristics described for the moving mirror in
Optionally, the moveable mirror is on the other side of block 506, covering a portion of GT etalons 802 and 806, or there are moveable mirrors, optionally separately controllable, on both the left and right sides of block 506. Additionally or alternatively, there are optionally separately controllable mirrors covering a portion of GT etalon 804 from above, and/or covering a portion of GT etalon 802 from above, providing additional degrees of freedom for controlling the dispersion compensation of module 800.
A GT etalon acts like an all-pass filter, with a phase shift Φ given by
for light of frequency ν, where c is the speed of light, and
The group delay as a function of frequency is given by
where τ0=4πnd/c is the time for light to travel twice the width d of the GT etalon.
assuming that ντ0>>1. As will be explained, this periodicity of the group delay has the potential advantage that the GT etalon can be used to compensate for dispersion in more than one channel at once.
For an optical network with many channels, it may be important to compensate for dispersion within the bandwidth of each channel, to avoid distorting the signal and generating bit errors, but it may not matter if different channels have different overall group delays, particularly if only chromatic dispersion is being compensated, not polarization mode dispersion. In such a case, the GT etalons are optionally designed so that free spectral range 906, given by
is equal to the channel spacing. Then, if the dispersion compensating module effectively compensates for dispersion within the bandwidth of one channel, it may also effectively compensate for dispersion within the bandwidth of each of the other channels, assuming that all of the channels together span a relatively small range of frequencies, and assuming that the dispersion being compensated in the optical network is a slowly varying function of frequency.
For a given GT etalon, the group delay as a function of frequency can vary, depending on R, as shown in
can still be satisfied with this small relative change in nd.
Choosing R, n, and d for each GT etalon, and choosing the lengths of the different etalons, and the degree of birefringence of the birefringent block, which controls the number of times N the light beam interacts with each etalon, provides a number of degrees of freedom for the design of the dispersion compensating module. These degrees of freedom may be useful for providing a better match to any of a wide variety of group delays as fixed functions of frequency across the bandwidth of one channel. And, as noted above, this may automatically provide a good match to the group delay of the optical network as a function of frequency (up to the addition of a constant) for each of the other channels, if the channel spacing is close to
The more degrees of freedom there are, the closer the match can be between the group delay as a function of frequency for the dispersion compensating module, and the group delay as a function of frequency that is needed to compensate for the dispersion of the channels used by the optical network. A better match between these two functions may reduce the residual dispersion (group delay ripple) in the optical network, or increase the channel bandwidth over which a given residual dispersion can be achieved, or both.
The above remarks concern compensating for a fixed group delay as a function of frequency for each channel, by the choice of fixed design parameters for the dispersion compensating module. In addition, the dispersion compensating module can compensate for group delay as a function of frequency that varies in time, by using control parameters that can be changed in real time. As described above, these control parameters may include the temperature of one or more GT etalons, controlled for example by thermoelectric devices as shown in
A dispersion compensation module such as module 500 in
To illustrate the order of magnitude of dispersion compensation that a module such as module 500 or module 800 might produce, consider light having wavelength approximately λ=1000 nm, in the near infrared, and etalons of thickness d=1 mm, and index of refraction n=1.5. Then τ0=4πnd/c=0.06 nanoseconds, and ν=c/λ=3×1014 Hz, so ντ0=1.8×104. Then the channel spacing is 17 GHz, and there are, for example, about 1800 channels within a frequency range equal to 10% of ν. With R=0.3, a moderately small value chosen to make GD(ν) fairly sinusoidal, the maximum slope in GD(ν) is approximately 3τ02 for a single interaction of light with an etalon. If the light interacts 10 times with each of two etalons in the dispersion compensating module, then the total dispersion compensation can be as great as 60τ02. An optical fiber of length L, designed to carry light of frequency ν, might be expected to have a chromatic dispersion of a fraction of L/cν, for example 0.2 L/cν. Setting 0.2 L/cν equal to 60τ02 and solving for L, we find that the dispersion compensating module can compensate for the chromatic dispersion of an optical fiber of length L=100 km. Although this estimate may be overly optimistic, a dispersion compensating module similar to module 500 or module 800 may be able to compensate for chromatic dispersion in an optical fiber more than 5 km long, or more than 10 km long, or more than 20 km long. If the width of each channel is about 10% of the spacing between channels, i.e. 1.7 GHz, and dispersion is compensated exactly at the center of each channel, then we might expect the residual dispersion near the edge of each channel, due to the d3(GD)/dν3 term for the etalons, to be on the order of 1% of the uncompensated chromatic dispersion. The parameters listed above are merely representative, and in an actual optical network may be greater or smaller by up to a factor of 1.2, or by up to a factor of 1.5, or by up to a factor of 2, or by up to a factor of 5, or by up to a factor of 10, or by up to a factor of 20, or by more than a factor of 20.
If a dispersion compensation module has two GT etalons, for example GT etalons 502 and 504 in
Here, the subscripts 1 and 2 refer to the two etalons, which may have different parameters R, n, and d, and hence different values of σ and τ. The number of interactions N of the light beam with each etalon may also differ for the two etalons, for example if there are sliding mirrors as in
if ντ0>>1.
Using etalons with relatively high reflection coefficient R, for example with R>0.5, has the potential advantage that, for a given N, ν and τ, the first derivative of the group delay (the chromatic dispersion compensation) will generally be greater than for lower R. For example, if the channels are centered on the steep parts of the sharp peaks of curve 902, then the first derivative of the group delay can be made arbitrarily high by making R close enough to 1, for example by using R greater than 0.7, or greater than 0.8, or greater than 0.9. However, with relatively high R, the first derivative of the group delay will generally be a steeper function of τ, and may be more difficult to control precisely, by changing τ, than if a lower value of R is used. Also, higher order derivatives of group delay with respect to frequency may be greater, if R is greater, possibly leading to greater group delay ripple, although it is possible to reduce such group delay ripple by using channels of narrower bandwidth compared to the spacing between channels. On the other hand, if R is very small, much less than 0.5, then the first derivative of group delay will be smaller, and a larger module, with a greater number of bounces N of the light beam, will be needed to obtain the same chromatic dispersion compensation. Optionally, a moderately low value of R is chosen as a compromise between these extremes, for example R=0.1, 0.2, or 0.3. Such a choice of R may produce a relatively high maximum chromatic dispersion compensation, which is still a relatively smooth function of τ, with relatively low values of the higher derivatives of group delay.
To illustrate how the group delay, and its first derivative, are controlled by controlling τ1 and τ2, we will consider the case where R1=R2=R<<1, and only keep the lowest order terms in R. But it should be understood that a similar analysis could be done taking into account some or all higher order terms in R, and/or values of R which differ for the two etalons. In this case, from Eq. (4),
We wish to choose a τ1 and τ2 for which d(GD)/dν will just compensate for chromatic dispersion, for a frequency ν0 at the center of a channel in the middle of the frequency range of all channels that are being used, and we want d(GD)/dν to have nearly the same value at the center frequency of all channels, over a wide a range of channels. This assumes that the range of frequency of all the channels being used is small compared to ν0, and that the chromatic dispersion is nearly the same for all the channels in this range. A different procedure for choosing τ1 and τ2 is optionally used if those assumptions are not true. We also want d2(GD)/dν2 to compensate for the rate of change of chromatic dispersion with frequency, at ν0 and at the center frequencies of the other channels to good approximation, to keep group delay ripple low. Then, from Eqs. (6), τ1 and τ2 optionally satisfy
at ν=ν0, where (CD)ν0 is the chromatic dispersion at ν0. From Eq. (7), τ1 and τ2 optionally also satisfy
at ν=ν0. In Eqs. (8) and (9), we have assumed, as noted above, that τ1 and τ2 are both nearly equal to τ0, with the differences between τ0, τ1 and τ2 being important only in the sine and cosine functions. If Eqs. (8) and (9) are satisfied at ν=ν0, then they are also satisfied to good approximation at the center frequencies of other nearby channels, ν=ν0±2πk/τ0, for k=1, 2, 3, . . . , because the sine and cosine functions in Eqs. (8) and (9) are periodic in ν with periods nearly equal to the channel spacing 2π/τ0. In order to estimate how far from ν=ν0 Eqs. (8) and (9) are satisfied to good approximation at the channel frequencies, sin(ντ1) may be written as
sin(ντ1)=sin [(ν−ν0)τ0+(ν−ν0)(τ1−τ0)+ν0τ1]=sin [(ν−ν0)τ0+ν0τ1] cos [(ν−ν0)(τ1−τ0]−cos [(ν−ν0)τ0+ν0τ1] sin [(ν−ν0)(τ1−τ0)] (10)
As long as (ν−ν0)(τ1−τ0)<<1, sin(ντ1)≈sin [(ν−ν0)τ0+ν0τ1)], which is periodic in ν at the channel spacing 2π/τ0. Similarly, cos(ντ1) is well approximated by a function that is periodic at the channel spacing as long as (ν−ν0)(τ1−τ0)<<1, and sin(ντ2) and cos(ντ2) are well approximated by functions that are periodic at the channel spacing as long as (ν−ν0)(τ2−τ0)<<1. So Eqs. (8) and (9) are satisfied to good approximation at any channel frequency that is well within both 1/(τ1−τ0) and 1/(τ2−τ0) of the channel frequency ν0. Optionally, values of τ1 and τ2, satisfying Eqs. (8) and (9) at ν=ν0, are chosen to be as close as possible, or nearly as close as possible, to τ0. Such a choice has the potential advantage that the chromatic dispersion is well compensated, and the group delay ripple is kept relatively low, not only for the channel at frequency ν0, but for about as broad a range as possible of other channel frequencies around ν0.
To illustrate how to find values of τ1 and τ2 that satisfy Eqs. (8) and (9) at ν=ν0, while being as close as possible to τ0, we consider, for simplicity, the case where N1=N2. It will be understood, to one of skill in the art, how to generalize this procedure to the case where N1 differs from N2. Then, from Eq. (8) at ν=ν0,
and from Eq. (9) at ν=ν0,
It is noted that the rate of change of chromatic dispersion with frequency, d(CD)/dν, is often on the order of the chromatic dispersion CD divided by ν0. In this case, the right hand side of Eq. (12) will be smaller than the right hand side of Eq. (11) by a factor on the order of ν0τ0, which is much greater than 1 if the channel spacing is small compared to ν0. In this case, to good approximation the right hand side of Eq. (12) may be set equal to zero. The parameters τ1 and τ2 then satisfy Eq. (12) when
τ1+τ2≈(2j+1)πν0−1 (13)
for some integer j=0, ±1, ±2, . . . . In this case, sin(ν0τ1)=sin(ν0τ2), and Eq. (11) is satisfied when
for some integer k=0, ±1, ±2, . . . . If the right hand side of Eq. (12) is not approximated as zero, then Eqs. (13), (14a) and (14b) are replaced by different expressions for τ1 and τ2, which depend on d(CD)/dν as well as on CD.
Optionally, in order to make τ1 and τ2 about as close as possible to τ0, j is chosen to be an integer close to ν0τ0/π, and k is chosen to be an integer close to ν0τ0/2π. Then (τ1−τ0) and (τ2−τ0) will both be less than or comparable to 1/ν0, and the module will provide nearly the same first and second derivatives of group delay for all channels, provided their range is small compared to ν0.
Optionally, the number of bounces N in the etalons is chosen, given the values of R and τ0, chosen as described above, so that the argument of the arcsin function in Eqs. (14a) and (14b) is less than 1, for the maximum chromatic dispersion (CD)ν0 that the module will ever need to compensate. If N is adjustable, for example as in module 800 of
The first derivatives of curves 1002 and 1004, producing the chromatic dispersion compensation, have the same magnitude and the same sign, corresponding to sin(ν0τ1)=sin(ν0τ2), as noted above after Eq. (13). For fixed N and fixed R, the sign and magnitude of the chromatic dispersion compensation may be changed by changing τ1 and τ2 so as to change the phase difference between curves 1002 and 1004, while keeping the average phase of curves 1002 and 1004 fixed, at a value that satisfies Eq. (12). In some embodiments of the invention, τ1 and τ2, and the phases of curves 1002 and 1004, are kept fixed, at values that satisfy Eq. (12), and the magnitude of the chromatic dispersion compensation is changed by changing N, using a moveable mirror as shown in
It should be noted that, because GD(ν) is periodic in ν at or nearly at the channel spacing 906, the first and second derivatives of GD(ν) are nearly the same for the central frequency of every channel, over a range of many channels, with the channel spacing being small compared to the frequencies of the channels. The range of frequencies ν shown in
In some embodiments of the invention, instead of setting the second derivative of the group delay to zero, or to another value, in order to specify the values of the control parameters of the etalons, the group delay itself is set to a desired value. This is done, for example, in order to use the same dispersion compensation module to compensate for both chromatic dispersion and polarization mode dispersion, with a different group delay used for the two polarization modes. Alternatively, a separate module, such as module 650 in
In some embodiments of the invention, the group delay, the first derivative of the group delay, and the second derivative of the group delay are all set at desired values, using three independent control parameters, chosen, for example, among τ1, τ2, N1, and N2, or combinations of them. Such a method may work best if R is moderately large for at least one of the etalons, and different for the two etalons, so that the shape of GD(ν), as well as its amplitude and phase, can be changed by adjusting N1 and/or N2. There may also be more than two etalons, as in
Changing the shape of GD(ν) by adjusting N1 and/or N2 has other potential advantages, even without using the same module to compensate for both chromatic dispersion and polarization mode dispersion. For example, it may then be possible to independently control the first derivative, second derivative, and third derivative of the group delay, or even higher derivatives, thereby reducing the group delay ripple, in a chromatic dispersion compensator. But using moderately small R, and having GD(ν) nearly a sinusoidal function, has the potential advantage that the feedback loops used to control the module, as described for example in
It should be noted that the τ's and N's may have different characteristics as control parameters, due to the fact that N is an integer, while τ may be controlled continuously, for example by controlling the temperature. For example, N may not allow as fine a degree of control as τ, but may allow more robust control in the presence of noise.
At 1104, the needed dispersion compensation, i.e. group delay as a function of frequency within each channel, is determined. Optionally, this is done by a direct measurement of residual dispersion in one or more of the channels. Alternatively or additionally, the needed dispersion compensation is determined empirically, by temporarily changing the dispersion compensation in different ways, and seeing what produces the best quality of the transmitted signal. For example, the first derivative of the group delay as a function of frequency within the channel is first increased, and then decreased, and the quality of the transmitted signal is compared in the two cases. Optionally, other changes in dispersion compensation are also made, for example the second derivative in group delay as a function of frequency is held at two or more different values, for each of two or more different values of the first derivative, and an optimal value of the first and second derivatives is found, which maximizes the signal quality. Other such methods will be apparent to those of skill in the art of optical communications.
At 1106, a calculation is made of the change in control parameters, for example the temperature of one or more etalons, or the position of one or more moveable mirrors, needed to achieve the optimal dispersion compensation found in 11104.
In some embodiments of the invention, the optimal change in dispersion compensation is not calculated at all, but the optimal change in the control parameters is directly found empirically, for example by finding the set of control parameters which maximizes the measure of signal quality.
At 1108, the control parameters are changed by the amount calculated to be optimal. Control then returns to 1102, and the measure of signal quality is measured again.
In some embodiments of the invention, the control algorithm for the feedback loop is more complicated than the one illustrated in
Device 1200 has a first planar mirror 1212, adjacent and parallel to surface 1206, and a second planar mirror 1214 adjacent and parallel to surface 1208. Between mirror 1212 and surface 1206, there is a first quarter wave plate 1216, and between mirror 1214 and surface 1208, there is a second quarter wave plate 1218. Mirror 1214 and quarter wave plate 1218 do not block light from entering block 1204 at input port 1210, because, for example, each of mirror 1214 and quarter wave plate 1218 does not extend as far as input port 1210, or has an opening through which light passes at input port 1210.
When light beam 1202 first reaches surface 1206, at a location 1220, it passes through quarter wave plate 1216, reflects from mirror 1212, passes back along nearly the same path through quarter wave plate 1216, and back into block 1204, at location 1220. Location 1220 is a potential exit point, because, if mirror 1212 were removed from location 1220, then the light would exit from block 1204, and would not re-enter block 1204. Because the light beam has passed twice through quarter wave plate 1216, when it re-enters 1204 it is linearly polarized in a direction perpendicular to the direction of polarization that it had when passing through input port 1210. Due to the birefringence of block 1204, the light beam, now labeled 1222, is refracted by block 1204, propagating at an oblique angle to surface 1206. Consequently, light beam 1222 reaches surface 1208 at a location 1224, rather than retracing the path of light beam 1202 to the location of input port 1210.
Location 1224, like location 1220, is a potential exit point for the light. However, quarter wave plate 1218 and mirror 1214 do extend as far as location 1224, so light beam 1222, like light beam 1202 at location 1220, passes through the quarter wave plate, in this case quarter wave plate 1218, reflects from the mirror, in this case mirror 1214, and passes back through the quarter wave plate, re-entering block 1204 at location 1224, propagating perpendicular to surface 1208 just before it enters block 1204. As a result of passing twice through quarter wave plate 1218, the polarization of the light beam, now labeled 1226, is again linear but in a direction perpendicular to the direction of polarization of light beam 1222, and parallel to the direction of polarization of the input light beam 1202. So light beam 1226 propagates perpendicular to surface 1208, straight across block 1204 to location 1228, another potential exit point on surface 1206, but lower down, with respect to the orientation of
The process described above repeats, with the light beam moving further down block 1204, until the light beam reaches a potential exit point location which is not covered by a mirror. In the case illustrated in
It should be noted that device 1200 can be built to displace the input light beam by a variety of different possible distances, and to have an output beam propagating in the same direction as the input beam, or the opposite direction, depending on which of the potential exit points have a mirror covering them. In addition, by using a moveable mirror for mirror 1212 and/or mirror 1214, the displacement of the light beam can be changed dynamically. It should also be noted that, in contrast to a device with two parallel mirrors which a light beam enters and exits at an oblique angle to the mirror surfaces, the input beam enters the birefringent block in device 1200 substantially perpendicular to one of its faces, and leaves the block substantially perpendicular to one of its faces. This fact may make it easier to couple device 1200 to optical fibers and to other optical elements. All of these advantages to device 1200 may apply also to the device shown in
If input beam 1202 enters device 1234 with the same linear polarization state as the input beam shown entering device 1200 in
If input beam 1202 has a different linear polarization state when it enters device 1234, orthogonal to the polarization state of the input beam in
In some embodiments of the invention, output port 1248 is located at location 1240 instead of at location 1246, so light exiting through output port 1248 is not laterally displaced at all, only light exiting through output port 1232 is laterally displaced. In some embodiments of the invention, output port 1248 is located on surface 1208, rather than on surface 1206, and the light exits device 1234 going in the opposite direction to the input beam, as well as laterally displaced, for this polarization state of the input beam.
In some embodiments of the invention, there is a controllable polarization rotator 1239, through which input beam 1202 passes before reaching device 1234, which controls the polarization of input beam 1202, and hence controls in which direction the light beam is displaced, and whether it leaves device 1234 through output port 1232 or output port 1248.
It some embodiments of the invention, a configuration similar to that shown in
In some embodiments of the invention, two or more of the modules or devices described, which use blocks of birefringent material and quarter wave plates, are arranged along a single long block of birefringent material, with one device arranged after the other device. At the end of the first device, the light immediately continues to propagate in the second device, without any need for the light to leave the first device through an output port and to enter the second device through an input port, and so on for any other devices arranged on the same block.
The invention has been described, in some cases, in the context of the best mode for carrying it out. It should be understood that not all features shown in the drawing or described in the associated text may be present in an actual device, in accordance with some embodiments of the invention. Furthermore, variations on the method and apparatus shown are included within the scope of the invention, which is limited only by the claims. Also, features of one embodiment may be provided in conjunction with features of a different embodiment of the invention. As used herein, the terms “have”, “include” and “comprise” or their conjugates mean “including but not limited to.”
While this invention has been particularly shown and described with references to preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the scope of the invention encompassed by the appended claims.
This application is a Continuation of PCT application serial number PCT/IL2007/000984 filed on Aug. 7, 2007, which in turn claims the benefit under 35 USC 119(e) of U.S. Provisional Application No. 60/835,913, filed Aug. 7, 2006, both of which are incorporated herein by reference in their entirety.
| Number | Name | Date | Kind |
|---|---|---|---|
| 5872878 | Bergmann | Feb 1999 | A |
| 6041151 | Mendlovic et al. | Mar 2000 | A |
| 6396629 | Cao | May 2002 | B1 |
| 6560015 | Cao | May 2003 | B1 |
| 6636651 | Li | Oct 2003 | B2 |
| 6804467 | Colbourne et al. | Oct 2004 | B2 |
| 6807329 | Zalevsky et al. | Oct 2004 | B2 |
| 7072111 | Iwatsuka | Jul 2006 | B2 |
| 20030086168 | Copner et al. | May 2003 | A1 |
| 20050174639 | Zalevsky et al. | Aug 2005 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20090231705 A1 | Sep 2009 | US |
| Number | Date | Country | |
|---|---|---|---|
| 60835913 | Aug 2006 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/IL2007/000984 | Aug 2007 | US |
| Child | 12366051 | US |
