Dual Directional Wavelength Demultiplexer

1. Field of the Invention

This invention relates to fiber optic transmission lines and, more particularly, to devices for attachment to optical fibers for demultiplexing signals transmitted therein in both directions.

2. Description of the Prior Art

The field of fiber optics has progressed in a relatively few years from laboratory curiosities and decorative pieces to present-day systems of high sophistication for optical communication and data transmission. Optical fibers or light tubes are specially fabricated filaments which exhibit the property of transmitting light longitudinally along a flexible axis. Various materials which can be used in fabrication of optical fibers and the particular properties thereof are described in Derick et al. U.S. Pat. No. 3,508,589 and, in further detail, in British Pat. No. 1,037,498, cited therein. Low-loss fiber optic taps are important components for fiber optic data links and data buses. This is true because it is desirable to be able to tap a portion of a signal propagating through an optical fiber without breaking or terminating the fiber, since fiber terminations add unwanted optical losses to the system and unfavorably increase the need for highly precise fiber splicing and interconnecting arrangements. Since fiber optic transmission lines having a large number of signal taps are inherently power-starved, it is important to minimize excess losses associated with these components. Furthermore, it is desirable to have taps which can be fabricated so that the tap ratio (the power output of the tap divided by the power into the fiber in a given direction) can be conveniently tailored to the unique requirements of the given system. Efficient fiber optic taps have been reported previously whereby two fibers are cleaved, or ground and polished, at specific angles and butt joined. For example, see Karr et al., "Lightwave Fiber Tap", Applied Optics, Vol. 17, page 2215 (July 15, 1978) and Kuwahara et al., "A Semi-Transparent Mirror-Type Directional Coupler for Optical Fiber Applications", IEEE Transactions on Microwave Theory and Technique, Vol. 23, page 179 (January 1975). In these examples, the tap ratio is variable either by changing the cleavage angle or by using materials with different indices of refraction between the cleaved surfaces. However, devices fabricated by such methods are quite fragile and cannot be easily reproduced with sufficient accuracy.

It has been demonstrated that when an optical fiber is bent in the form of an arc, there is an increased tendency for light to escape from the bent region in a radiation pattern which is primarily in the plane of the bend and which is directed away from the center of curvature. See, for example, Gambling et al., "Radiation From Curved Single-Mode Fibers", Electronics Letters, Vol. 12, page 567 (Oct. 14, 1976); and Goell et al. U.S. Pat. No. 3,982,123. The tendency for light to escape from the bent region of the fibers is enhanced when a flat region is lapped and polished into the fiber surface perpendicular to the radius of the bend in the fiber.

The use of fiber optics to transmit multiplexed signals is also known. U.S. Pat. No. 4,061,577 of Bell mixes multiple signals and transmits them through a fiber optic cable for demultiplexing after conversion to electrical signals by a photodetector. However, this system is quite complex and requires the use of electronics to multiplex and demultiplex the signals. U.S. Pat. No. 3,953,727 of d'Auria et al. both multiplexes and demultiplexes by use of selective mirrors or by lenses and gratings. The signal is transmitted in the multiplexed form using fiber optics after generation by plural diodes and is reflected by selective mirrors to a lens for transmission. A reverse mirror and lens sequence is utilized to reflect the various frequency signals to different photodetectors and a grating is inserted to demultiplex the signal. Thus, d'Auria provides for dual directional transmission of multiplexed signals. In another system, patent 4,155,628 of Schlossberg suggests the use of a series of interferometer elements, each made up of four reflective elements, forming a square optical path, and a pair of beam splitters for input and exit of the signal from the square path. Light enters the square through the first beam splitter, and selectively exits each of the squares through the second splitter.

Further, the use of thin films coupled with prisms is known in the art. For instance, U.S. Pat. No. 3,584,230 of Tien discloses coupling the laser generated light to a thin film waveguide utilizing prisms and lenses. Tien obtains phase matching of the evanescent wave and the wave propagating in the thin film by appropriate alignment of the angle of the input beam. Lastly, dielectric gratings have also been used in thin film waveguides to distinguish between directions of travel. In U.S. Pat. No. 3,982,810 of Tamir et al. a serrated dielectric grating is disposed on the thin film for this purpose.

The present invention is a bidirectional wavelength demultiplexer. Wavelength division demultiplexing (WDM) is commonly used to describe the superposition of several communication channels at several carrier wavelengths. It is an important technique for fiber optic transmissions since it allows for multimode optical transmission through a single fiber. The purpose of the present invention is to provide for the separation of the several wavelengths carried in an optical fiber in a straightforward spatially compact, physically rugged, optically efficient, and low-cost system.

In brief, the devices of the present invention involve the use of a solid form to support and bend an optical fiber from which a predetermined portion of the incident light is to be tapped. The form is machined from a solid material, such as aluminum, and has one curved side with a radius of curvature of between 5 and 8 cm., preferably about 6 cm. A coating of epoxy resin is placed on the curved surface of the form, after which the fiber is placed in the epoxy and is constrained to the curvature of the form. When the epoxy has been cured and a plurality of forms so prepared, the forms are mounted on a lapping and polishing fixture. The loose ends of the fibers are secured to the fixture to prevent breakage. A flat surface is lapped into the epoxy and eventually a region of the glass fiber, embedded in the epoxy, becomes exposed. Lapping of the flat surface into the fiber and the epoxy continues until the desired amount of material has been removed. The depth of removal of the fiber core is monitored by measuring the length and width of the flat region on the fiber. This measurement can be made by using a microscope with a micrometer-actuated x-y stage. The depth of the fiber removal is controlled routinely to an accuracy of +/- 10 micrometers and preferably to an accuracy of +/-2 micrometers. The lapped surface is then polished using the same support fixture on a second machine which uses a fine (e.g. 1/4 micron) polishing compound. Further details of such a fabrication procedure may be found in Application Ser. No. 15,027, filed Feb. 26, 1979 (now abandoned) and assigned to the assignee of this application.

The complete device comprising the invention further includes a prism for directing the light travelling in a given direction in the fiber to one of a pair of associated photodiode array detectors. Two photodiode arrays are provided in order to sense signals travelling in each direction in the fiber. The prism to be assembled on the fiber element has two optical grating-containing surface ends, one for each direction of light travel, and each is positioned at an angle preferably equal to 52.degree. from the plane of the polished surface. The performance of the demultiplexer is dependent on the angle of the refelction from the prism face that directs the light toward the arrays. The prism should produce an exiting beam which sufficiently separates the wavelengths of the multiplexed signal to provide minimal interference. For a radius of curvature of 6 cm., the angle may be 52.degree.+/-5.degree.. With this arrangement, nearly all of the light coupled out of the fiber is transmitted to the active area of the arrays. This arrangement, of course, is effective for both optical grating surfaces so that demultiplexing is effective for light travelling in both directions.

Each of the prism's optical grating surfaces is attached to, or fabricated on, the angular surface of the prism. The reflection of the optical signals from the grating causes the optical beam to be diffracted, i.e. the spectral components are separated, and different spectral regions are received separately by the photodiodes in each array. The arrays may be formed of discrete photodiodes or they may be separate elements of a monolithic photodiode array. In addition, the photodiodes may be active elements in an array integrated with a signal processing device such as a charge-coupled shift register. Additionally, further channel separation may be provided by optical filters interposed between the exit surface of the prism and the photodetectors. However, the fundamental means for separation of the wavelength multiplexed channels is diffraction by the optical grating.

The grating, as noted, may be assembled to the prism of the demultiplexer in a variety of ways. It may be etched or blazed into the glass of the prism or it may be a striped geometry etched in a metal film deposited on the surface of the prism. In this case, of course, the metal surface is reflective. Also, the grating may be a commercially available unit attached to the prism using optically transparent cement or epoxy. Of course, the particular channel separation desired for the unit will control the particular structure of the grating. Thus, the grating selected will depend upon the particular wavelengths to be utilized as the signal carrying channels for the multiplexed signal.

The associated photodiode arrays are either cemented directly to the upper surface of the prism using an optically transparent epoxy or supported adjacent the prism so that the demultiplexed light leaving the fiber is directed toward the appropriate photodiode. The photodiode can be packaged in a standard header/lid assembly using a lid with a window in it. More specifically, the photodiodes can be packaged in a hybrid microelectronic assembly whereby a preamplifier is included in the assembly and a window is provided on the packaged lid.

The fiber/prism/grating/photodiode system is preferably mounted rigidly in a suitable package, such as an aluminum box or molded enclosure, which supports a pair of opposed optical connectors and an electrical connector needed to interface with the device. In such an embodiment, the radiating light which is tapped from the fiber is highly localized by virtue of the coupling from the flat region of the bent fiber to the photodiodes using the described optical prism and optical grating system.

A better understanding of the present invention may be had from a consideration of the following detailed description, taken in conjunction with the accompanying drawings in which:

FIG. 1 is a view of a portion of one embodiment of the invention in a preliminary stage of fabrication;

FIG. 2 is a view of the portion of FIG. 1 following a succeeding step in the fabrication process;

FIG. 3 is a schematic view of the fiber after treatment as in FIG. 2;

FIGS. 4A and 4B are schematic views of the second portion of one embodiment of the present invention;

FIG. 5 is a schematic of an assembled unit; and

FIG. 6 is a schematic of the completed unit.

FIG. 1 illustrates a form 10 to which an optical fiber 12 is affixed by cured and hardened epoxy resin 14. The form 10 is machined of solid material, such as aluminum, to develop a curved side 16 with a circular arc having a radius of cuvature between 5 and 8 cm., preferably of 6 cm. Sufficient epoxy is applied to result in an adequate support surface when the part is lapped, as shown in FIG. 2. After epoxy 14 has been cured, form 10 to which fiber 12 is attached, together with a plurality of other forms and fibers which have been similarly prepared, is mounted on a lapping and polishing fixture (not shown). In this step, the loose ends of fibers 12 are secured to the fixture to prevent breakage. A flat surface 18 is lapped into the epoxy, as shown in FIG. 2. During lapping in this fashion, a small section of fiber 12 becomes exposed and lapping of the flat surface 18 continues into fiber 12 and epoxy 14 until a predetermined amount of fiber has been removed. The depth of penetration of the fiber core is monitored by measuring the length and width of surface 18 along fiber 12. This measurement is made using a microscope and the depth of fiber removal is controlled routinely to an accuracy of +/-10 micrometers, preferably within +/-2 micrometers. If the radius of cuvature of the fiber is less than about 4 cm., fiber 12 tends to crack as soon as the lap cuts through the buffer coating or cladding (not shown) which surrounds the fiber. Therefore, the preferred radius of curvature for fiber 12, which is determined by curved surface 16 on form 10, is 6 cm. After lapping, the flat surface is polished using the 1/4 micron particle size polishing compound.

As indicated in FIG. 3, which is a schematic representation of a portion of fiber 12 lapped to provide planar surface 18', dimension h is the distance from the fiber axis to flat surface 18'. This dimension affects the performance of the optical fiber demultiplexer as will be discussed hereinbelow.

Double optical grating faced optical prism 20 of FIG. 4 is mounted on the flat surface of the fiber to couple and diffract a fraction of the light travelling in each direction in the fiber away from the fiber to an associated photodiode array. The percentage of the light coupled from the fiber to the photodiode array is controlled by the depth of the flat surface 18' (FIG. 3) lapped into the fiber. Only light propagating in one direction is coupled into each array; thus two reflective optical grating surfaces and two photodiode arrays are provided for in the present invention. Prism 20 is provided with mating surface 22 which eventually contacts flat surface 18' of FIG. 3. Light enters the prism through surface 22 travelling toward one of the two optical grating surfaces 24 on the prismatic sides. The surfaces are at an angle .phi. of 52.degree.+/-5.degree. from the perpendicular as shown. This angle, of course, varies with the radius of curvature previously described, but for the preferred radius of curvature of 6 cm., 52.degree. is the appropriate angle. The light is reflected inside the prism from optical grating surfaces 24 and outward to the photodiode array, not shown. In its preferred form the prism used in the present invention has a length (x) of 0.50 inches, a height (y) of 0.16 inches and a thickness (z) of 0.05 inches.

The prism to be attached to the fiber is illustrated in FIGS. 4A and 4B. The index of refraction of the glass used in the prism is important to consider. It has been determined that for best results the index of refraction of the prism, which is preferably glass, should be greater than but nearly equal to the index of the fiber core. For example, when using a fiber with a core index of 1.61, such as the Galite 3000 LC fiber manufactured by Galileo Electro-Optics, Inc., prism glass with an index of 1.62, such as F-2 glass manufactured by Schott Glass Company, is most appropriate. "Galite" is a registered trademark, Registration Nos. 1,058,328 and 1,138,265. The optical grating surfaces are metallized, for example, by vacuum deposition of gold or aluminum and then selectively etched to remove portions of the film. Thus they act as diffraction gratings. The light beam from the fiber strikes the grating, is diffracted and reflected toward the top surface. Thus as the light travels away from the grating it is demultiplexed by the separating of the various wavelengths of the carrier waves. The angles of the reflecting surfaces and the widths between the grating reflective surfaces are chosen so that the reflected light beams are diffracted and separate into discrete wavelength bands. This helps reduce reflection losses at the prism-photodiode array interface. The preferred angles, as noted with regard to the specific arrangements disclosed herein, are both 52.degree..

Diffraction grating 26 is adhered to surface 25 in FIG. 5 prior to mounting prism 20. In this form, the diffraction grating is adhered by, for instance, optically clear adhesive such as epoxy to surface 25. However, as noted above, the diffraction grating can be produced by vacuum deposition of metal on surface 25 followed by selective etching, or by blazing the surface to produce the grating. The method of producing the grating is not critical and numerous methods are known to those skilled in the art. Also, photodiode arrays 28 which contain preamplifier circuits are adhered to the upper surface of prism 20 by the same type of optically transparent adhesive. Each array is provided with a common lead 30 and plural individual wavelength output leads 32. Thus, as shown, wavelength multiplexed light may be demultiplexed into three separate signals or channels. Additional channels could be provided by using additional photodiode elements in the arrays.

In FIG. 5, form 10 is shown with fiber 12 mounted and lapped in place to form contact surface 18. Prism 20, provided with surfces 25, diffraction grating 26, photodiode arrays 28 and contact surface 22, is positioned over the flat lapped space. It is then cemented in place using clear optical epoxy, for example. At this point in the process of producing the present structure, the photodiode array, if not already positioned, may be cemented to the prism over the region from which the diffracted beam exits the prism. The beams are shown as the arrows and indicated as P.sub.1t and P.sub.2t. A portion of a light signal input at P.sub.1I is tapped off as signal P.sub.1t with the remainder continuing on through as P.sub.10. A multiplexed light signal can be input to the fiber, the photodiode array placed on the prism, and the electrical response of the array monitored. When the desired wavelength separation and the overall output photocurrent of the array are maximized, by moving the array, it has been properly positioned and is cemented in place. Two photodiode arrays are required in the dual directional demultiplexer of the present invention. The second array is positioned in the same manner as the first, using P.sub.2I and P.sub.2t to determine its position. After positioning and adjusting, arrays 28 are indepenently cemented in position.

Finally, at this point, the demultiplexer, as shown in FIG. 6, is encapsulated in a potting compound 34 and provided with optical connectors 36. An exemplary potting compound is Ablecast No. 402. "Ablecast" is a registered trademark, Registtration No. 1,033,479. The resulting unit may be connected by connectors 36, in line, for tapping off portions of the multiplexed signal from a throughput fiber. Also, an optional optical filter 40 has been inserted between the prism and the fiber to remove unwanted signals from the coupled multiplexed signal or to select particular signals for directing to the photodiode array.

In selecting the type of grating, the size of the prism, the photosensitive diode arrays, etc., the following factors are of importance. Generally, either the particular wavelengths to be utilized, or the photosensitive diodes are selected first. These, of course, must be matched in order for proper functioning. Next, the unit size is determined, should it be important to miniaturize the unit. It should be noted that the farther the diodes are away from the diffraction grating, the better the separation of the various wavelengths of demultiplexed signals. However, there is an inherent power loss with travel through a thick prism. Thus, since photosensitive diodes are commercially available which are very sensitive to small or low power light signals, the geometry of the system is more controlled by the need to minimize power loss, and the height y of the prism is generally controlled by the spacing between the photodiodes, the grating character and the spacing in the grating. Also, the angle of impingement of the multiplexed light upon the diffraction grating affects the rate of separation of the various wavelengths, along with the grating pitch. Thus, the particular wavelengths of light available, and the effective wavelength sensitivities of the photosensitive elements along with the geometry of the system control the specific design of the unit.

The use of an angle of 52.degree. for the diffraction grating generally assumes that the grating fully separates the multiplexed signals into the various wavelengths contained therein. Where less efficient gratings are utilized, often a centrally positioned signal containing a portion of the signal of each wavelength is present. When this occurs it is desirable to minimize the effect of this signal by placing a light absorptive element, such as the signal absorber 38, at the point where the central multiple wavelength signal exits the prism. Alternatively, the angle of the grating can be changed from the preferred 52.degree. angle to an angle that directs this central signal away from the photodiode array and still directs the demultiplexed signal toward the array in a higher order diffraction pattern.

For example, if it is desired to isolate five different channels of wavelength multiplexed optical data produced by injection laser diodes having center wavelengths equal to 805, 825, 850, 870 and 885 nm., the first-order diffraction from the grating would be determined by the equation

where .lambda. is the optical wavelength, d is the pitch of the grating and .theta. is the angle of the diffracted rays relative to the grating normal. In order to minimize cross-talk it would be necessary to maximize the derivative of the diffraction angle with respect to the optical wavelength. Thus, differentiating the above equation with respect to the change in wavelength,

in radians per .mu.m. Since the tangent of .theta. increases rapidly with increasing .theta., so long as .theta. is greater than 45.degree., the wavelength discrimination will improve substantially for increased values of .theta.. However, the efficiency of the demultiplexer would depend on the diffraction angle and on polarization of the incident light. This efficiency is affected by the diffraction angle, blaze angle, and optical polarization and it diminishes to zero as the diffraction angle approaches 90.degree.. By substituting in the above equations, it will become apparent that the pitch of the grating should be slightly more than one-half the value of the longest wavelength to be diffracted. Thus, if an angle of 52.degree. is chosen for use with the 5 injection laser diodes mentioned above, a grating having 1850 lines per mm. would be desired. This is within the range of commercially available diffraction gratings. It should be noted that, due to the close spacing of the diffraction angles, for example, the above units result in diffraction angles varying from 48.1 to 54.9 degrees, and the optical signal incident on the grating must be very well collimated. Thus the system must be well designed in order to prevent unacceptable power losses from occurring.

Although there have been described above specific arrangements of a dual directional demultiplexer for fiber optics in accordance with the invention for the purpose of illustrating the manner in which the invention may be used to advantage, it will be appreciated that the invention is not limited thereto. For example, although the invention has been disclosed in the context of association with dual directional demultiplexers using visible light, the principles of the invention are equally applicable to systems not using visible light and the like. Accordingly, any and all modifications, variations or equivalent arrangements which may occur to those skilled in the art should be considered to be within the scope of the invention as defined in the appended claims.