The present invention is directed generally to transmitting light. The unique advantage of the technology disclosed herewith is to multiplex light at different wavelengths or carrier frequencies at low losses over a very large temperature range.
For the transmission of optical signals, generally two different approaches are presently being deployed. In a first approach, the transmission of multiple signals is facilitated over a single optical fiber with each signal having a different carrier frequency.
This approach is commonly referred to as dense or coarse wavelength division multiplexing (DWDM or CWDM) with CWDM technology being commonly used in long reach (LR) or extended reach (ER) transceiver products.
A second approach facilitates the transmission of optical signals over a strand of optical fibers with one signal per fiber. The latter approach is henceforth referred to as parallel technology and is commonly used in short reach (SR) transceivers based on the 40GBASE-SR4 or 100GBASE-SR10 communication standards. The parallel transceiver technology does not utilize a key advantage of optical fibers compared to copper wires: the ability of a single optical fiber to carry multiple data signals of different carrier wavelengths in a single fiber without resulting in compromises on signal integrity.
Two main transmitter optical engines are used in these parallel transceiver technologies, both of which are important prior art to this disclosure and incorporated herewith. As a representative example of the first group, U.S. Patent No. 2009/0016734 A1 to Hamazaki discloses an optical block for parallel transmission using planar waveguide technology.
As an example of the second group of parallel technologies, U.S. patent application Ser. No. 12/316,211 to Noguchi discloses an optical turning block based on lens arrays. Particularly to the latter technology, many comparable topologies exist including those in which the light is propagating straight from the sources to the fibers and turning the signal electrically instead of optically. The Prizm LightTurn connector, a trademark of US Conec Limited, for example, is a device that terminates a strand of 12 optical fibers to couple light to and from an array of photodetectors or light sources, respectively, with one fiber for each light source or photodetector. An array of curved reflective surfaces focuses the individual emission onto the facet. Similarly, U.S. Pat. No. 7,887,243 to Abel et al. discloses a miniature mechanical transfer (MT) optical coupler. However, the above disclosures do not discuss the option to multiplex a plurality of wavelengths into a single optical fiber. The main technologies utilized there are based on dispersive optical elements such as filters, diffraction gratings or, less common, prisms as discussed below. U.S. Pat. No. 7,184,621 to Zhu and U.S. Pat. No. 6,941,047 to Capewell et al., which are hereby incorporated by reference, describe a system of cascaded thin film filters that can multiplex and demultiplex four or more wavelengths to and from a single fiber. U.S. Pat. No. 4,299,488 to Tomlinson and U.S. Pat. No. 4,198,117 to Kobayashi, which are hereby incorporated by reference, describe a multiplexing and demultiplexing technique in which a reflection grating multiplexes multiple wavelengths incident from a set of fibers and demultiplexes into a second set of fibers. While these technologies can utilize the same optical engine to multiplex and demultiplex optical signals, they require a very careful alignment of the constituent optical components to function at high data rates with low insertion loss.
Another important prior art is the combinatorial problem of optimizing the packing of two-dimensional unit objects in an envelope circle of minimum diameter. In the context of this disclosure, the envelope circle corresponds to the object size when imaging a plurality of light sources onto a single optical fiber facet. Reis discussed the dense packing of equal circles 301 in an envelope circle 302 (G. E. Reis, Dense Packing of Equal Circles within a Circle, Math. Mag. 48 (1975) 1, 33-37). Cantrell discussed the optimized packing of eight equal squares 303 in an envelope circle 304 (D. W. Cantrell in communication with E. Friedman, March 2002, DWCantrell@sigmaxi.org).
Finally, numerous well-established technologies exist using arrayed waveguides to multiplex and demultiplex different wavelengths, particularly for dense wavelength division multiplexing applications. As an example, U.S. Pat. No. 5,002,350 to Dragon discloses the use of arrayed waveguides as an optical multiplexer and demultiplexer. The size of this type of multiplexer is preventing it from utilization in transceivers with a compact form factor.
The present invention discloses two main embodiments to optically multiplex a plurality of wavelength division multiplexed signals. The embodiments utilize a system of lenses and a planar waveguide combiner structure, respectively, to simultaneously multiplex and spatially turn the signal by a designated angle.
Simplification of the electrical conduction path from the electrical connector to and from the light emitter and detector, respectively, is a desirable design criterion for optical transceivers. Parasitic inductances and capacitances can substantially compromise the signal integrity, particularly at higher data rates. At the same time, it can be important to limit the interconnect cabling complexity by allowing the ingress and regress of a plurality of optical signals over a single optical fiber. The devices disclosed herewith present compact solutions to both of these system demands.
In accordance with the present invention, it has been discovered that the proposed wavelength division multiplexing device operates nearly independent from the polarization distributions of the light sources and is very insensitive to the precise wavelengths of the channels, therefore making it highly tolerant to errors in center wavelength, spectral width, or wavelength shifts due to temperature variations.
It is believed that the reason for this highly polarization and wavelength insensitive behavior is the absence of a dispersive optical element such as a filter, prism, or diffraction grating to multiplex the optical signal of the proposed embodiments. While some polarization or wavelength dependence is introduced through material dispersion and small polarization-dependent Fresnel reflections at optical interfaces, these effects are minimal due to operation near surface normal incidence.
In a preferred embodiment illustrated in
When multiplexing optical signals with tighter wavelength spacing, the same epitaxial growth can be used with an additional structure that introduces a systematic phase shift during each round trip in the laser cavity. In this case, the VCSEL apertures are circular and can be packed in close spatial proximity. The packing problem is therefore defined as packing unit circles 301 into an envelope circle 302.
When multiplexing coarse wavelength division multiplexed (CWDM) signals, preferably different epitaxies are used to generate the light emission. This forces separate die placement with die footprints being typically of square or rectangular shape. The packing problem can therefore be described as packing squares 303 into an envelope circle 304.
The magnification of the object can be approximated by the ratio of the numerical aperture (NA) of the light source and the numerical aperture of the optical fiber and a topology factor that describes the radius R of the envelope circle normalized by the separation of light emitting element centers. For example, when multiplexing 4 CWDM VCSELs with a beam divergence half angle of 20 degrees into a single multimode fiber with an NA of 0.2, the magnification is approximately given by
in which s denotes the VCSEL-to-VCSEL aperture center separation, fcol denotes the focal length of the four individual collimating lenses (for example, lenses 101 in
For comparison, in a linear arrangement of the VCSEL sources, the magnification required to still satisfy the numerical aperture of the fiber increases to about 7.3. The larger magnification results in a larger spot at the facet of the multimode fiber, therefore resulting in reduced alignment tolerances or larger insertion losses or both.
The more colors that are multiplexed into a single fiber, the more demanding the alignment process will become for a given fiber NA. While a larger fiber NA such as those from photonic crystal fibers can of course push this boundary, using the preferred packing configuration remains an important design factor.
In an exemplary structure, the emissions from 4 light sources, each having a different wavelength, are combined into a single 50 μm core multimode fiber. Both the collimating lenses and the focusing lens are described by aspheric surface sagitta described by
in which R is the base radius of the lens, k is its conic constant, and αj are the higher order deviations from a spherical shape. Four collimating lenses are arranged in a 2×2 configuration with base radius R of 190 μm, a vanishing conic constant k, and higher order deviations α1 of 0.201 mm−1, α2 of −43.354 mm−3, and α3 of −135.015 mm−5. The center-to-center separation s of the lenses is 250 μm. The single focusing lens is defined by a base radius R of 920 μm, a conic constant k of −0.093, and higher order deviations α1 of −0.01 mm−1, α2 of −0.354 mm−3, and α3 of −0.046 mm−5. The parameters can be can be adjusted to yield vanishing factors α1 if desired.
In a third preferred embodiment, the multiplexing device includes a receptacle for a transistor outline (TO) header or a transistor outline (TO) cap, a 2×2 array of optically transmissive curved surfaces to individually collimate the emission of 4 light sources of different wavelengths, a focusing lens to multiplex the four collimated beams onto a single multimode fiber facet, and a receptacle for a standard fiberoptic connector such as Lucent connector (LC), standard connector (SC), straight tip (ST), or fiber channel (FC). As before, the embodiment can be generalized to multiplex more than one group of light beams of different wavelengths into an array of optical fibers and include a receptacle for a strand of fibers such as a mechanical transfer (MT) fiber connector.
In a fourth preferred embodiment illustrated in
In another preferred embodiment illustrated in
As in the first embodiment, the wavelength division multiplexer operates again without the use of an explicit dispersive optical element such as a filter, prism, or diffraction grating. The multiplexing device can furthermore feature an array of curved surfaces 504 to increase the coupling efficiency from the light sources into the individual input waveguides 502. Furthermore, the embodiment can include a curved surface 505 at a designated distance from the output facet of the output waveguide 503 to increase the coupling efficiency from the single output waveguide to the optical fiber. Mechanical alignment features 506 can optionally establish a designated distance between the light sources and said array of curved surfaces 504. Alignment features 506 can furthermore be used to establish a designated lateral spatial alignment between the light sources and the curved surfaces 504. Similarly, mechanical alignment features 507 can optionally establish a designated distance between curved surface 505 and an optical fiber entrance facet. Alignment features 507 can furthermore be used to facilitate a lateral spatial alignment between curved surface 505 and an optical fiber entrance facet.
The waveguide combiner can be fabricated by defining a mold that includes grooves or channels that are subsequently filled and cured with a polymer that exhibits a higher index of refraction than the body of the mold. Alternatively, the multiplexing device can be fabricated by defining a mold that features a recess into which a polymer waveguide slab can be inserted and attached. The polymer waveguide combiner is fabricated independently and is inserted and secured into the predefined cutout.
Preferably all transmissive surfaces of the multiplexing device embodiments described above are coated with a dielectric layer structure that is antireflective in the wavelength range of the light sources.
Both the bent-waveguide-based and lens-based multiplexing mechanisms could in principle be used as a demultiplexer as well when incorporating filter structures at the photodetector end to discriminate between the different wavelength channels. However, in both cases a minimum loss of 6 dB when demultiplexing 4 wavelengths cannot be avoided due to the splitting of the incident polychromatic emission into 4 polychromatic beams, each having at least 6 dB less optical power. For demultiplexing, the introduction of a dispersive mechanism, such as through a cascaded filter structure, prism, or diffraction grating is the preferred technique.
Depending on the application, separate devices for multiplexing and demultiplexing may be desirable, while for other applications, a symmetric physical mechanism for multiplexing and demultiplexing is preferred.
The United States Government has a paid-up license in this invention and the right in limited circumstances to require the patent owner to license others on reasonable terms as provided by the terms of Contract No. FA8750-05-C-0110 awarded by the United States Government.