WO2002071672A2 - (de)multiplexer with interleaver for producing a flat-top filter function and enhanced channel separation - Google Patents

Abstract

An apparatus and method are provided employing a single set of reduced-cost optics for (de)multiplexing optical signals whereby the resultant signals have a flat-top response with increased channel spacing. One aspect of the present invention uses an optical interleaver (310) to separate channels from an input multichannel (multiplexed) optical signal (302) with a select channel spacing into, for example, two input signals consisting of alternating channels at twice the select channel spacing. The separated signals are then directed to a multiple-input, such as dual-input, (parallel)(de)multiplexer device (320) which further spacially separates the channels for coupling with single channel transmission fibers (3221-n, 3241-n). The input fibers to the (de)multiplexer are preferably offset relative to each other to accommodate channel shift resulting from the interleaver.

Description

(DE)MULTIPLEXER WITH INTERLEAVER FOR

PRODUCING A FLAT-TOP FILTER FUNCTION

AND ENHANCED CHANNEL SEPARATION

TECHNICAL FIELD

The present invention relates generally to fiber-optic communications, and more particularly toward optical interleavers, multiplexers and demultiplexers having increased channel spacing and a flat-top filter function.

BACKGROUND ART

At the inception of fiber optic communications, typically a fiber was used to carry a single channel of data at a single wavelength. Dense wavelength division multiplexing (DWDM) enables multiple channels at distinct wavelengths within a given wavelength band to be sent over a single mode fiber, thus greatly expanding the volume of data that can be transmitted per optical fiber. The wavelength of each channel is selected so that the channels do not interfere with each other and the transmission losses to the fiber are minimized. Typical DWDM allows up to 40 or 80 channels to be simultaneously transmitted by a fiber.

DWDM requires two conceptually symmetric devices: a multiplexer and a demultiplexer. A multiplexer takes multiple beams or channels of light, each at a discrete wavelength and from a discrete source and combines the channels into a single multi-channel or polychromatic beam. The input typically is a linear array of waveguides such as a linear array of optical fibers, a linear array of laser diodes or some other optical source. The output is typically a single waveguide such as an optical fiber. A demultiplexer spacially separates a polychromatic beam into separate channels according to wavelength. Input is typically a single input fiber and the output is typically a linear array of waveguides such as optical fibers or a linear array of photodetectors.

In order to meet the requirements of DWDM, multiplexers and demultiplexers require certain inherent features. First, dispersive devices must be able to provide for a high angular dispersion of closely spaced channels so that individual channels can be separated over relatively short distances sufficiently to couple with a linear array of outputs such as output fibers. Furthermore, the multiplexer/demultiplexer must be able to accommodate channels over a free spectral range commensurate with fiber optic communications bandwidth. Moreover, the devices must provide high resolution to minimize cross talk and must further be highly efficient to minimize signal loss. In addition, a single device is preferably reversible so it can function as both a multiplexer and a demultiplexer (hereinafter, a "(de)multiplexer"). The ideal device would also be small, durable, inexpensive and scalable. One class of (de)multiplexers showing considerable promise is the diffraction grating- based (de)multiplexer because of its relatively low cost, high yield, low insertion loss and crosstalk, uniformity of loss as well as its ability to multiplex a large number of channels concurrently. However, grating-based (de)multiplexers typically have a Gaussian filter function. For long-haul fiber networks, large numbers of (de)multiplexers cascaded in series can cause a significant overall narrowing of the filter function, ultimately leading to large insertion loss or limiting the data rate. For smaller metro networks, it may not be necessary to cascade large numbers of (de)multiplexers in series. However, deployment of metro network equipment is extremely cost sensitive and a Gaussian filter function requires that the wavelength of the emitting lasers be locked to a particular wavelength with tight precision. But lasers tend to drift for a number of reasons, including variation in ambient temperature and aging. Providing improved lasers adds significant cost to the network equipment. In contrast, providing a flat-topped filter response places much less stringent requirements on the tolerance for the laser wavelength. Thus, for both long-haul and metro applications, it is desirable to produce a low-cost (de)multiplexer with a flat-topped filter function suitable for use on high speed networks with less expensive unlocked lasers having greater wavelength drift or with locked lasers which may drift due to age or temperature fluctuations. While a number of alternatives have been proposed for adapting grating based

(de)multiplexers to provide a more flat-topped filter function, all have disadvantages. One device which overcomes at least some of the disadvantages is described in U.S. Patent Application No. 09/675,276 entitled "Apparatus and Method for Producing A Flat-Topped Filter Response for Diffraction Grating (De)Multiplexer", which is incorporated in its entirety. The DWDM device described therein utilizes a bulk echelle grating which is highly desirable because it provides for a high angular dispersion of closely spaced channels so that individual channels can be separated over relatively short distances sufficiently to couple with a linear array of single channel fibers. The echelle grating DWDM device also provides for high resolution to minimize cross talk and is highly efficient to minimize signal loss. The '267 Application also describes a number of methods of producing a flat-top filter function in

(de)multiplexer devices. Specifically, the '267 Application discloses a grating-based DWDM device with a segmented grating having at least two surfaces angularly displaced relative to one another to produce two or more dispersed channel images per output fiber, thereby resulting in a flat-top filter function.

An optical interleaver is designed to receive a multichannel optical signal, having a first channel spacing, and separate channel subsets at a greater, second channel spacing. Heretofore, a significant use of interleavers has been to interface devices designed for one channel spacing with devices designed for a different (such as double) channel spacing. In a similar fashion to (de)multiplexers, interleavers are also optically reversible, thereby being able to interleave separate multichannel signals into a single multichannel signal.

However, further channel separation remains desirable.

SUMMARY OF THE INVENTION

The present invention provides an apparatus and method employing a single set of reduced-cost optics for (de)multiplexing optical signals whereby the resultant signals have a broad flat-top response with increased channel spacing. In particular, the present invention includes an optical interleaver with a multi-input (de)multiplexer.

One aspect of the present invention uses an optical interleaver to separate channels from an input multichannel (multiplexed) optical signal with a select channel spacing into, for example, two input signals consisting of alternating channels (for example, one signal consisting of even channels and the other consisting of odd channels) at twice the select channel spacing. The separated signals are then directed to a multiple-input, such as dual- input, (parallel) (de)multiplexer device which further spacially separates the channels for coupling with single channel transmission fibers. The input fibers to the (de)multiplexer are preferably offset relative to each other to accommodate channel shift resulting from the interleaver. Because the interleaver has the effect of separating every other channel of the input multiplexed signal, the signals input to the demultiplexer have twice the channel separation as in the original signal, which greatly aids in providing improved channel separation at the demultiplexer output. In addition, the interleaver can inherently produce a broader flat-top filter function for the demultiplexer device, thereby being able to accommodate greater channel drift. Moreover, the greater band width allows for higher data rate transmissions, e.g., on the order of OC-192 or greater. Moreover, the apparatus and method of the present invention have an advantage of decreasing the cost of the (de)multiplexing componentry. The present invention produces a flat-top filter function for a (de)multiplexer. When used in conjunction with a bulk optical grating having two input fibers as described herein, a single (de)multiplexer having a single set of optics may be used to provide the desired flat-top filter function. Furthermore, the (de)multiplexer used to demultiplex a 50GHz signal may have the same configuration as a multiplexer designed to demultiplex a 100GHz signal.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a schematic illustration of the optical system of the present invention; Fig. 2 is a schematic plan view of a multiplexer/demultiplexer using a bulk echelle grating in accordance with the present invention;

Fig. 3 is an enlarged cross-section of the echelle grating grooves illustrating relevant dimensions;

Fig. 4 is a graphical representation of possible step widths and riser heights at different orders which may yield a working echelle grating; Fig. 5 is a schematic representation of an example of a multiplexer/demultiplexer with a bulk echelle grating in accordance with the present invention;

Fig. 6 is a partial cross-sectional view of a pigtail template;

Fig. 7 is a perspective view of the multiplexer/demultiplexer with bulk echelle grating of Fig. 2 illustrating the potential adjustment of the components; Fig. 8 is a schematic view of a first alternate embodiment of the multiplexer/ demultiplexer using a bulk echelle grating including a pair of collimating/focusing concave mirrors;

Fig. 9 is a second alternate embodiment of the multiplexer/demultiplexer using a single collimating/focusing mirror; Fig. 10 is a third alternate embodiment of the multiplexer/demultiplexer in accordance with the present invention using an off-axis parabolic mirror as the collimating/focusing optic with the device arranged in a near-littrow configuration;

Fig. 11 is a forth alternate embodiment of the multiplexer/demultiplexer of the present invention using a concave echelle grating; Fig. 12 is a schematic elevation of a pigtail harness having a one-dimensional input array of fibers and a two dimensional output array of fibers;

Fig. 13 is a schematic representation of a multiplexer/demultiplexer having stacked multiplex fibers and a two-dimensional array of single channel fibers; Fig. 14 is a plot of the system response versus wavelength for the multiplexer/ demultiplexer of Fig. 5;

Fig. 15 is a side view of a grating for producing a flat-topped filter response;

Fig. 16 illustrates the Gaussian pass-band produced by each section of the grating of Fig. 15 and the resulting flat-topped filter function;

Figs. 17A and 17B are plan views of an alternate embodiment of a grating for producing a flat-topped filter response;

Fig. 18 is a schematic cross-section of a thermally expanded TEC fiber core;

Fig. 19 is a schematic cross-section of a focusing lens operatively associated with a receiving/transmitting end of an optical fiber;

Fig. 20 is an illustration of the channel shift at the outputs of the interleaver incorporated into the present invention;

Figs. 21A-C illustrate means by which channel shift may be accommodated; and

Fig. 22 illustrates the flat-top filter function of the interleaver incorporated into the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

One embodiment of the device of the present invention for use in optical communication systems is illustrated schematically in Fig. 1. The device 300 includes an interleaver/deinterleaver 310 optically coupled to a multiplexer/demultiplexer 320. In the present discussion, the multiplexer/demultiplexer ("(de)multiplexer") 320 will be discussed in terms of a demultiplexer and the interleaver/deinterleaver 310 ("(de)interleaver") will be discussed in terms of a deinterleaver. However, it will be appreciated that the description applies equally to a multiplexer and an interleaver with the functions of the inputs and outputs reversed. The device 300 further includes an input waveguide or fiber 302, for carrying an original multiplexed optical signal containing multiple channels, and a plurality of output fibers 322ι._n and 324j._n, each for carrying a demultiplexed optical signal containing a single channel. The exemplary deinterleaver 310 provides two multiplexed, multichannel signals, carried by fibers 312 and 314, to the demultiplexer 320. The fibers 312 and 314 each carry one-half of the original multichannel signal, such as the alternating odd and even subchannels, respectively. Each of the output fibers 322j carries a single subchannel (such as one of the odd subchannels) and each of the output fibers 324; carries a single channel of the alternate subchannel (such as one of the even subchannels). By way of example and not limitation, the input fiber 302 may carry a multiplexed signal consisting of multichannels at 50GHz spacing to the deinterleaver 310. The deinterleaver 310 divides the 50GHz spacing input signal into two lOOGHz spacing signals, described as an odd channel lOOGHz input carried on fiber 312 and an even channel lOOGHz input carried on fiber 314. Each of the odd channel lOOGHz input 312 and the even lOOGHz input 314 can be provided to an input of the dual-input demultiplexer 320. Exemplary 50GHz (de)interleavers known in the art include fiber Bragg grating-based interleavers and Mach- Zehnder interferometer-based interleavers. Representatives sources of interleavers include New Focus Oplink, Chorum, Avanex and Wave Splitter. A (de)multiplexer suitable for use in optical communications systems in accordance with the present invention is illustrated schematically in Fig. 2. It includes a pigtail harness 12 consisting of an input waveguide 14, a plurality of output waveguides 16 arranged in a linear array adjacent the input fiber 14, a collimating/focusing lens 18 and an echelle grating 20, all of which are optically coupled. For the sake of clarity, only seven output waveguides are illustrated (the center output waveguides underlies the input fiber 14 in Fig. 2 as can be seen with respect to elements 312 and 322 of Fig. 13). Furthermore, the waveguides 14, 16 are preferably single mode optical fibers. As will be discussed in greater detail below, in a preferred embodiment, 90 or more output waveguides may be associated with a single input waveguide, depending upon the bandwidth channel, separation and other factors. As used herein, "optically coupled" or "in optical communication" mean any connection, coupling, link or the like, by which optical signals carried by one optical element are imparted to the "coupled" or "communicating" element. Such "optically communicating" devices are not necessarily directly connected physically to one another, but may be separated by a space through which the optical signals traverse or by intermediate optical components or devices.

As illustrated in Fig. 2, the multiplexer/demultiplexer 10 is in "near littrow configuration," meaning that the input incident beam λι__n and the channels diffracted off the surface of the grating λ_\, λ₂, λ₃, λ₄, λ₅, λ^, λ₇ are generally along the same optical axis (that is, they trace a very close path) and the lens both collimates the input beam λι__n and focuses the diffracted channels λ λ₇ to the output fiberslό.

The echelle grating 20, like other gratings such as echellette gratings, uses interference between light wavefronts reflected from various portions of its ruled surface or steps 22 to divide the incident beam consisting of a plurality of channels λι.„ having a select channel spacing within a select wavelength range λι__n into separate channels of wavelength beams λi- λ₇ which are angularly dispersed by the grating into output waveguides some distance away. Referring to Fig. 2, the channel separation (D) of the device, which is the product of the focal length of the focusing/col limating optic 18, the angular dispersion and the incremental channel spacing is equal to the distance S between the center of adjacent output waveguides

16. The echelle grating 20 is particularly suited to use in optical communication systems because of a unique combination of properties: 1) it provides clear channel separation notwithstanding channels being closely spaced (such as 0.4 nm or less); 2) it provides large spatial separation of channels over relatively short distances; and 3) it is highly efficient in the range of optical communications wavelengths.

Referring to Fig. 3, for the purpose of this specification, echelle gratings are a special grating structure having groove density (1/d) of under 300 grooves/mm and a blaze angle θ_b of greater than 45° which typically operate at an order of diffraction greater than 1. In combination, these features enable a multiplexer/demultiplexer to efficiently separate closely spaced channels over a relatively small focal length (e.g., 5 inches), permitting a small form factor (on the order of 10 inches in length or less).

Consideration of certain external and performance constraints point to the desirability of echelle gratings for DWDM. The external constraints include the following:

1) Minimize focal length, with a focal length of under 6 inches desired. 2) Center wavelength in near infrared, approximately at the center of the C-band,

1550 nm.

3) A minimal channel spacing (e.g., 0.4 nm or less).

4) Large free spectral range, 150 nm.

5) System number in the range of 4-8. 6) Rugged, minimum cost system.

The performance constraints include:

1) Resolution greater than 20,000 (although resolution greater than 10,000 may be suitable for some applications).

2) High dispersion. 3) Flat response across spectral range.

4) High efficiency or low loss, (>75%).

5) Minimize polarization dependent loss. The external constraints of ruggedness size and cost minimization as well as performance constraints of ease of alignment and high efficiency dictate a littrow configuration, which simplifies the system optimization analysis.

Fig. 3 illustrates the echelle grating geometry and the variables set forth below. θ = blaze angle α = incident angle β = diffracted angle In littrow, λ = θ = s β b = step (reflective surface) size d = 1/groove density a = riser size An examination of a number of constraining factors discussed above illustrates the utility of echelle gratings for DWDM.

1. Constraining Factors: / number (f) in range of 4-8 and resolution ("R")>20,000.

Result: For a grating in littrow configuration,

R > where W is the illuminated width of the grating. Thus,

W=(20,000/2)(1550nm), or W~1.55cm. W x / = /l (focal length), or /« = 1.55 cm x 8 * 124

2. Constraining Factors: FC > 124 mm and channel separation of at least 80μ.

Result: For an echelle grating in littrow, dispersion

where m = order of diffraction. Thus, assuming channel separation to be at least 80μ, λ =4 x 10^"4μ and β = 1.2 x 10 μ, m > 1.5b. 3. Constraining Factors: FSR (free spectral range) > 150

Result: FSR = — , which implies m = or m < 10. m 150

4. Constraining Factors: Wish to provide a flat response over the bandwidth.

Result: The diffraction envelope must have a broad enough maximum so that loss is minimized at the extremes of the wavelength range. This dictates b< 8.5 μ. An order greater than 7 spreads the light too much across the diffraction peak, resulting in unacceptably low efficiency. Thus: b < 8.5μ and m ≤ 7.

5. Constraining Factors: High efficiency. (>85%)

Result: Efficiency is a function of step size. A step size must be selected providing a channel width capturing 90% of the signal at a select order, b > 3μ yields suitable efficiency.

6. Constraining Factors: Limitations on m from 4. and 2. above. Result: 1.5 < m < 7.

7. Constraining Factors: For an echelle grating in littrow mode: - -^ . Result: a = 3.88μ at m = 5

= 4.65μ at m = 6 = 5.43μ at m = 7 Fig. 4 illustrates that these constraints and results provide a range of values for a and b at a given range of suitable orders (m). Simulations aimed at maximizing efficiency and minimizing polarization dependent loss optimize around blaze angles and groove frequencies that fall in the range of echelle gratings, i.e., 45<θ_b<78° and d<300 grooves/nm. Furthermore, limitations on manufacturing further dictate that only echelle gratings can provide the necessary results within the external and performance constraints.

In designing a functioning multiplexer/demultiplexer, a number of design parameters were selected that were dictated by many of the external and performance constraints set forth above. An exemplary configuration is illustrated schematically in Fig. 5, with like elements having the same reference number as Fig. 2. The dictating constraints and their effect on the exemplary bulk echelle grating DWDM are as follows:

1. Channel Characteristics

Currently optical communications utilize what is know as the "C" band of near infrared wavelengths, a wavelength band ranging from about 1528-1565 nanometers (nm). This provides a bandwidth or free spectral range of 37 nm available for channel separation. Known prior art multiplexer/demultiplexers require a channel spacing of 0.8 nm or even 1.6 nm, resulting in a possibility of only between 48 and 24 channels. Because echelle gratings provide markedly superior channel dispersion, a much smaller channel spacing of 0.4 nm was chosen, resulting in a possibility of 93 channels over the C band. As the tuning range of semiconductor lasers increases and optical communications expand beyond the C band to include the "L" band (about 1566-1610 nm) and the "S" band (about 1490-1527 nm), a total bandwidth of about 120 nm or more is foreseeable, creating a possibility of the multiplexer/demultiplexer accommodating 300 channels or more per input fiber.

Current optical communications operate primarily at a channel frequency of 2.5 GHz, known as OC48. At OC48 the channel width λ ₈ = 0.02 nm. Optical communications are currently beginning to adopt a frequency of 10 GHz, known as OC192, at which the channel width λ i₉₂ = 0.08 nm.

2. Fiber Dimensions

Standard single mode optical fiber used in optical communications typically have an outer diameter of 125 microns (μ) and a core diameter of lOμ. Optical fibers having an outer diameter of 80μ and core diameter of 8.3μ are available, model SM-1250 manufactured by

Fibercore. In this example, both the input fiber 14 and the output fiber 16 are single mode and share the 80μ outer diameter. Assuming the output fibers 16 are abutted in parallel as illustrated in Fig. 5, this results in the core centers being spaced 80μ, or a required channel separation D of 80μ at the select focal length. Because fibers of different outer diameter are available and fibers cladings can be etched away, it is possible that the 80μ spacing can be reduced, with core spacing of 40μ or less being foreseeable, which could enable shorter focal lengths or different echelle grating designs having lesser angular dispersion. The spread of the beam emitted from the fiber was 10° at the e-folding distance, although it was later found to be 14° at the 1% point. 3. Form Factor

The design was intended to provide a high channel density in a form factor consistent with or smaller than used in current multiplexer/demultiplexer devices. A total length of between 10-12 inches was the design target. To accommodate all the optics and harnesses, a maximum focal length of 5 inches (127 mm) was chosen. As discussed above, in light of the constraining factors of the number between 4-8 and a resolution R>20,000, a focal length of

124 was ultimately dictated.

4. Di spersi on Limitati on s

In order to prevent the loss of data, it was necessary that the dispersion of the echelle grating be constrained. The initial .4 nm channel spacing at the echelle grating was required to be about 80μ of separation at the output fibers (corresponding to the core spacing). On the other hand, the 0.08 nm channel width of OC192 frequencies could not disperse to much greater than the fiber core aperture over the focal length. Thus: channel separation OΛnm , ., channel width (OC192) 0.08rcm while > fiber spacing 80μ core diameter 8.3μ

5. Grating Design

The variables affecting grating design are:

1) wavelength range

2) efficiency

3) dispersion (D)

4) desired resolution λ)

Fig. 3 is a cross-section showing the principle echelle grating dimensions including: blaze angle (θ_b), wavelength range and groove density (d).

For design of the grating, 150 channels centered on 1550 nm was chosen. This results in a physical size of the spectral image of (number of channels) x (maximum separation), or 150 x 80μ = 12,000μ. This desire to have 90% of the intensity contained in 12,000μ constrains the size of b. The far field pattern of the diffraction grating is

7tb 7td

N = number of lines illuminated, β = — sin θ_b and α = — sin θ_b

A A

Spread-sheet calculations show that b < 5.5λ (or b < 8.5μ) is necessary to make the spectral image > 12,000μ at its 90% intensity point.

To minimize loss, i.e., maintain adequate efficiency, b > 2λ. Thus 2λ < b < 5.5λ. (Condition A).

In littrow mode, the angular dispersion is:

Δ x (linear separation) = (ΔΘ)(/ )(Δλ) = (Δλ) - rø

80μ < — (4 x 10^"4 μ)(l .2 x 10⁵ μ) b

1.6b, m > > 1.6b„ However, for OC192, dispersion must be constrained to contain the .08 nm channel width in a lOμ core, so that m < 3.34b_μ.

Thus, 1.67b < m < 3.34b (Condition B).

The desired resolution (R) = — = N • m. ' Δλ Here, λ = 1550 nm and Δλ = .08 nm, yielding a required resolution R = 19,375 or approximately 20,000. Assuming a beam size at the grating of 2.1 cm (based upon a ft = 124 cm and 10° divergence):

NT = — PC —²-¹) -, p = l ,i^•nes/ /cm = — ! cosθ_b d

~_{n nnn} 2Λxl ^'2 cm 2ΛxlO^~2 cm

Thus, 20,000 < m = m , or dcosθ b b<1.05 m (Condition C).

To align the order m with the diffraction peak in littrow mode, we know

„ _ mλ , or a must have the values: ^{a "}

(Condition D)

Only as θ_b increases to greater than 45° is it possible for conditions A and D to be satisfied. Assuming θ_b = 60°, and m = 5, a = 3.38μ b = 2.24μ d = 4.48μ.

Hence, all of conditions A-D are satisfied.

Selection of the precise groove density and blaze angle are also affected by the polarization dependent loss and manufacturing constraints. For the embodiment illustrated in Fig. 5, use of an interferometrically controlled ruling engine to machine the line grating drove the selection of a line density evenly divisible by 3600. Considering these various factors led to selection of groove density d = 171.4 grooves/mm and m = 6. This leads to a = 3.88μ, b = 3.55μ, and a corresponding blaze angle of 52.6° for this example. However, this methodology shows that for a focal length between 30-125 mm and an order of 5-7, potential blaze angles range between 51° and 53° and the groove density carries between 50 and 300 grooves/mm to provide linear channel separation of between 40-125 microns and an angular dispersion of the echelle of between 0.091 and 0.11 degrees/nm.

In the example of Fig. 5, the echelle grating has a groove density of 171.4 grooves/mm and a blaze angle of 52.6°. The echelle may be formed from one of several known methods. For example, it may be formed from an epoxy layer deposited on a glass substrate into which a master die defining the steps is pressed. The steps are then coated with a highly reflective material such as gold. The steps may also be precision machined directly into a glass or silicon substance and then coated with a reflective material. A further option is the use of photolithographic techniques described in McMahon, U.S. Patent No. 4,736,360, the contents of which are hereby expressly incorporated by reference in its entirety.

The lens 18 may be a graded index (GRLN) optic with spherical surfaces or a compound lens with one or more surfaces that might not be spherical (aspheric). The use of lenses or a single lens to collimate the beam and focus the dispersed light limits spherical aberrations or coma resulting from the use of front surface reflectors that require the optical rays to traverse the system in a off-axis geometry. A first type of potential lens uses a radially graded refractive index to achieve near-diffraction limited imaging of off-axis rays. A second type of lens actually consists of at least two individual pieces cemented together (doublet). Another option uses three individual lens pieces (triplet). These pieces may individually have spherical surfaces, or if required for correction of certain types of aberration, aspheric surfaces can be utilized. In this case, the lens would be referred to as an aspheric doublet or triplet.

In the example illustrated in Fig. 5, the lens 18 is an aspheric singlet of a 25.4 mm diameter having a spherical surface 26 with a radius of curvature of 373.94 mm and an aspheric surface 28 with a radius of curvature of 75 mm and a conic constant of -0.875. The average focal length in the 1520-1580 nm wavelength range is 125.01 nm. Thus, the distance

A from the center of the spheric surface to the emitting end of the input and output fibers 14, 16 is about 125 mm. The average distance between the aspheric surface 28 and the center of the surface of the grating 20 is about 43.2 mm.

In the pigtail 12 represented in Fig. 2, the input and output fibers terminate in the same plane. This is also the case with the example illustrated in Fig. 5. In some configurations, however, the input and output fibers 14, 16 are on slightly different axes and do not terminate in the same plane. The fibers 14, 16 of the pigtail are precisely located by being fit into a template 34 illustrated schematically in Fig. 6. The template 34 has a plurality of parallel v- shaped grooves 36. The template and v-shaped grooves are preferably formed by etching the grooves 36 into a silicon substrate. In the example in Fig. 6, the grooves of the template are spaced 80μ.

The example configuration of Fig. 5 is shown in perspective view in Fig. 7. To facilitate alignment, the pigtail 12, the lens 18 and the grating 20 have limited freedom of movement in multiple directions. Once they are moved into position, they are secured by clamps or a suitable bonding agent. The lens 18 is held stationary. The pigtail 12 is movable by translation along the x, y and z axes. The input and output fibers can be moved independently along the x axis. The echelle grating 20 is fixed against translational movement except along the z axis. It can be rotated about each of the x, y and z axes. Other possible combinations of element movement may also yield suitable alignment.

The dimensions and performance criterion of the DWDM device 10 of Fig. 5 are summarized as follows:

Fibers: SM-1250 (Fibercore) • Outer diameter 80μ

• Core diameter 8.3μ

• / Number 4-8 Lens: Aspheric singlet

• Average focal length (fiϊ) = 125 Optical Signal: λ = 1528-1565 nm channel spacing = 0.4 nm Grating:

• d = 5.83 μ

• θ_b = 52.6° • order = 6

System Performance:

• D (linear separation) = 80μ

• Resolution (R) = 20,000

• Efficiency = 75% Fig.. 14 is a plot of the system response (y-axis) versus wavelength (x-axis) for the grating described above at a 100 GHz (0.8 nm) channel spacing over the 1528-1565 nm bandwidth at an average insertion loss of 7.5 db. This plot illustrates the flat insertion loss across the bandwidth.

As an alternative to the use of a littrow configuration as well as the use of collimating lenses, concave mirrors may be used for collimating and focusing the incident beam. A first alternate embodiment of a concave mirror dense wavelength multiplexer/demultiplexer 40 is shown schematically in Fig. 8. Single mode input fiber 42 emits a divergent incident beam 44 consisting of multiplexed channels onto the surface of a collimating/focusing concave mirror 46. The collimated beam 48 is then directed in an off-axis manner to the surface of an echelle grating 50. The echelle grating disperses the channels according to their wavelength in the manner discussed above with respect to Figs. 2 and 5 and the dispersed channels 52 are reflected off axis off the front surface of the concave collimating/focusing mirror 54. The collimating/focusing mirror 54 then focuses and reflects the various channels to a corresponding fiber of an output fiber array 56. As alluded to above with respect to the discussion of the embodiments of Figs. 2 and 5, use of surface reflecting optics such as the collimating mirror 46 and the concave focusing mirror 54 requires that the optical beams traverse the system in an off-axis geometry which creates significant aberrations (spherical aberrations and coma) that significantly limit the performance of the system. However, the use of the front surface reflecting optics has the potential of facilitating a more compact form factor than is possible with littrow configurations using a single optical lens. As should be readily apparent, combinations of front surface reflecting optics and lenses can be used in non-littrow configurations where necessary to balance form factor minimization requirements and optical aberrations.

Fig. 9 is a schematic representation of a second alternate embodiment 80 using a single concave mirror as both a collimating and focusing optic along the optical axis. In this embodiment, input fiber 82 directs a beam consisting of multiplexed channels to the surface of the concave mirror 84. A collimated beam 86 is reflected off the echelle grating 88 which diffracts the multiplexed signal in the manner discussed above. The demultiplxed channels are then reflected off the surface of the concave mirror 84 and directed into the array of output fibers 92. While the embodiment 80 contemplates the mirror 84 being spherical and therefore having a constant diameter of, for example 25 cm, a slightly parabolic or aspheric mirror may be used to improve image quality, if necessary.

Fig. 10 is a third alternate embodiment 100 using an off-axis parabolic mirror as the collimating/focusing optic. In this embodiment, multiplexed light from the input fiber 102 is directed off the front surface of an off-axis parabolic mirror 104 which in turn directs a collimated beam of light 106 off the surface of an echelle grating 108. The multiplexed light is reflected off the surface of the echelle grating 108 back to the surface of the off-axis parabolic mirror 104 and dispersed to respective output fibers 106. In this embodiment, the echelle grating is in near-littrow configuration, thereby directing light back to the output fibers 106.

A fourth alternate embodiment illustrated in Fig. 11 uses a concave echelle grating 107 configured to be the optic which collimates and focuses the incoming beam. This embodiment eliminates the need for the collimating/focusing lenses or concave mirrors of alternate embodiments one - four.

Various modifications can be provided to the basic echelle grating demultiplexer structures illustrated schematically in Figs. 2-11 to further increase the channel carrying capacity of single mode optical fibers. As alluded to above, it is foreseeable in the future that advancements in optical amplifier technology will enable bandwidth in excess of the current 60 - 80 nm bandwidth used in optical communication. Such broad bandwidths tax the ability of an echelle grating DWDM to effectively multiplex and demultiplex the entire bandwidth, particularly in the frequencies at the edge of this broad band. Accordingly, it would be useful and desirable to use a network of echelle grating DWDM devices with each device optimized to multiplex/demultiplex light in a portion of the broad spectral range. For example, assuming future amplifier technologies enable bandwidths on the order of 120 -180 nm, each echelle grating DWDM could be optimized to function with a portion, for example Vi, of the bandwidth, 60-90 nm.

The bulk optic echelle DWDM of the present invention is able to simultaneously demultiplex signals from a number of input fibers. In each of the echelle grating DWDM devices illustrated in Figs. 2-11 above, light in the exemplary demultiplexer 320 of Fig. 1 is spatially resolved in a single dimension, vertically in a direction transverse the dispersion direction. Consequently, the input fibers 312 and 314 carrying signals to the demultiplexer 320 may be vertically stacked in a linear array and the output fibers 322]._n and 324ι__n may be arranged in a corresponding two dimensional array for the demultiplexed signals from the input fibers. This concept is illustrated schematically in Fig. 12, an elevation view of a pigtail harness containing the input and output fibers of the demultiplexer 320. First and second input fibers 312 and 314 lying in a vertical linear array are optically coupled to first and second horizontal rows of output fibers 322]._n and 324ι__n, respectively. Thus, a one dimensional input array produces a two-dimensional output array. It will be appreciated that, while the present example is limited to two input fibers 312 and 314, and only nine output fibers in each of the first and second rows of output fibers 322ι__n and 324ι__n, the actual number of output fibers will correspond to the number of input channels and will be a function of the channel separation and input bandwidth, and may easily exceed 90 output fibers per output row. Each output fiber has a core center, and the output fiber core centers are spaced a distance equal to the linear separation of the grating at the device focal length. Further, the number of corresponding input and output arrays may be greater than two and is largely a function of external factors such as the space available for the pigtail harness. As should be appreciated, this configuration allows a single demultiplexer to demultiplex channels from a number of input fibers, thereby minimizing the number of echelle grating DWDM devices required for a multiple input fiber optical system. This further illustrates the flexibility and scalability of the echelle grating DWDM devices in accordance with the invention.

The use of the echelle DWDM devices for demultiplexing the split wavelength bands provides the many advantages discussed above with regard to the embodiments illustrated in

Figs. 2-11. However, the present invention could be practiced with other DWDM devices such as fiber Bragg grating devices, integrated waveguide arrays or the like. With an echelle spectrograph permitting wavelength spacing of .4 nm, a device for providing a total wavelength range of 120 nm will allow up to 300 channels to be demultiplexed from a single fiber. Furthermore, this system is scalable.

Fig. 13 is a schematic representation of a preferred embodiment of a stacked input bulk optic echelle DWDM device 320. Input beam 'l-io from input fiber 312 is directed to the collimating/focusing optic 340 and a collimated beam is then directed off the reflective surface of the reflective echelle grating 342. The diffracted channels

λ'₂ then return through the collimating/focusing optic 340 and are dispersed to the fibers comprising the first output row 322 as illustrated by λY The collimating/focusing optic has an optical axis 344 and the input fiber 312 and the output row 322 are equally spaced from the optical axis 344 of the collimating/focusing optic in the vertical direction. In a like manner, a multiplexed input beam λ ι__n is emitted from the input fiber 314 and its various channels λ²ι, λ² are diffracted to the second horizontal output row 324. With respect to each of output rows 322 and 324, the centers of the optical fibers in the row are each spaced a distance from the centers of adjacent optical fibers in the row equal to the channel separation of the echelle grating 342 at the focal length of the focusing/collimating optic 340. The propagating ends of the output fibers as well as the propagating ends of the input fibers all lie in a plane spaced the focal length of the collimating/focusing optic from the collimating/focusing optic.

Fig. 15 illustrates an adaptation to the diffraction grating of Fig. 5 for providing a broadened pass band or a more flat-topped filter function. As should be apparent to those skilled in the art, the adaptation is applicable to diffraction gratings other than echelles. The grating 200 of Fig. 15 is identical to the grating 20 of Fig. 5 except it is divided into two sections 202 and 204 that are angularly displaced relative to one another. The angular displacement is in fact very small, and is greatly exaggerated in Fig. 15. Assuming a configuration illustrated in Fig. 5 with a focal length of 135 mm, the total angular displacement is on the order of 10 - 50 arc-seconds with an angular displacement of about 15 arc-seconds believed to be preferred. The angular displacement is chosen so that with the grating incorporated in the (de)multiplexer 10 of Fig. 5, the optical signal diffracted by each section 202, 204 is offset in a direction of dispersion relative to the portions of the optical signal diffracted by the other section 202, 204. The offset is preferably on the order of 20μ at the receiving/transmitting ends of the optical fibers. Obviously the angular displacement is a function of the desired offset and focal length.

The desired result is illustrated in Fig. 16. Each segment produces is own Gaussian function, the function 206 corresponding to section 202 and function 208 corresponding to section 204. The superposition of these two filter functions 210 approximates the desired flat-topped filter function.

In the grating 200 of Fig. 15, the first and second sections 202, 204 are preferably planar and are preferably formed in a single substrate 212. They intersect along a line of intersection 214. The angular displacement is preferably chosen so that, as illustrated, the angle φ between the sections 202, 204 is greater than 180 degrees. Parallel grooves 216 are preferably formed in the planar sections 202, 204 parallel to the line 214. This simplifies manufacture of the grating 200. The grating can be manufactured as discussed above or using holographic techniques.

The preferred embodiment described above with respect to Fig. 15 could be altered in a number of ways and still perform the function of producing a flat-topped filter function. For example, instead of being a planar grating, the grating could be convex as illustrated in Fig.

11. Or, the grooves could be transverse the line of intersection. Another alteration could be having the angle φ be less than 180 degrees. Also, the preferred embodiment illustrates only first and section sections 202, 204. It should be understood that three or more sections could be provided, each angularly displaced from the other, to produce three or more Gaussian functions to modify the superimposed function as desired. Further, while the single embodiment of Fig. 16 shows the first and second planar sections 202, 204 in essence inclined about a parallel axis, first, second or more planar sections could be inclined about unparallel axes as desired. While the preferred embodiment shows the grating sections formed in a single substrate, they could be formed in multiple substrates suitably supported in operative positions to achieve the same result. These examples of alterations are not intended to be limited on the scope of the invention, but merely to illustrate some of the alternatives within the scope of the invention.

Another variation to the grating to produce a flatter bandpass is ruling the grating in two halves (side-by-side in the dispersion direction) with two slightly different blaze angles and line spacings. This is illustrated schematically in Figs. 17A and 17B. Fig. 17A is a schematic elevation view of a bulk grating 250, again preferably an echelle grating such as that described with reference to Fig. 5. First section 252 has a line spacing S and a second section 254 has a slightly greater line spacing R. Referring to Fig. 17B, which is a schematic cross-sectional view of the grating 250 of Fig. 17A along line B-B, first section 252 has a blaze angle P and a second section 254 has a slightly different blaze angle Q. Alternatively, only either one of the line spacing or blaze angles of segments 252, 254 could vary. More than two rulings could be used if desired. This would be accomplished during the ruling of the master grating. Once again, 20 arc-second precision will be required and the line density will need to change concomitantly by a very small amount. Such variations in the grooves could produce the filter function illustrated in Fig. 16. Typically grating manufacturers only have -0.1 degree control over the absolute blaze angle on a grating, but it should be possible to change the blaze angle accurately by a very slight amount after ruling half of the grating using a high precision fixture. The ruling density can be controlled to almost arbitrary precision. Polarization dependent loss, resolution and efficiency concerns could make this option difficult to pursue.

Altering the grating as discussed above may not in all circumstances provide a complete solution to providing a flat-topped filter response. For OC 48 operation, the bit rate is 2.5 Gbs. This corresponds to a signal bandwidth of 2.5GHz or 0.02 nm. The spot size of the signal light at the input output pigtail of Fig. 5 if approximately lOμ for OC 48 operation. Any change in wavelength immediately begins to compromise insertion loss as the wavelength of the light varies since the effective aperture of the fiber is also approximately lOμ. (Note that a change in wavelength will displace the focused beam relative to the fiber core into which it is directed.) For higher bit rates (e.g. OC 192 or 10 Gbs), the signal bandwidth increases to 0.08 nm. The spot size at the output pigtail of Fig. 5 increases to more than 20μ. Off-setting the diffracted signal portions as discussed above will further increase the effective spot size and, particularly for OC 48 and higher bit rates, create the potential for loss of data.

Obviously, if the core size itself is large enough in a particular (de)multiplexer, the problem might not be present. More likely, however, the increased effective spot size must be accommodated. One way to address this problem is to decrease the spot size by minimizing dispersion. For example, if possible, the receiving/transmitting ends could be moved closer to the focusing/collimating lens.

A more promising way to address this problem is to provide a structure operatively associated with the receiving/transmitting ends of the fibers for radially expanding the effective size of the fiber core. One potential structure is to use thermally expanded core fibers (TEC fibers) as the multiplex and single channel fibers in the fiber pigtail array that is used to capture the (de)multiplexed signal. A TEC fiber illustrated in Fig. 18 is produced from a fiber 270 having a core 272 with a first refractive index doped with a diffusing agent surrounded by a cladding 274 having a second refractive index. When a portion 276 of the fiber is heated to high temperatures (700-1400°C) locally for a given period of time (about 10 minutes - 5 hours), the diffusing agent diffuses into the cladding and varies the refractive index of the cladding to essentially expand the core. An exemplary diffusing agent is GeO₂. Prior to heating there is a step variation in the index of refraction between the cladding and the core. After heating the change in the index of refraction between the core and the cladding is more "Gaussian" in profile, which results in an expanded effective size of the core

278 and the mode field diameter of the fiber. Referring to Fig. 19, the core has an essentially adiabatic taper in the portion.

As one example, the unexpanded core diameter (at 272) is about 8.2μ and the total diameter of the cladding is between 80-125μ. After heating, the effective core diameter (at 108) is increased to 15-24μ, although an increase of up to 40μ (approximately a factor of 5) is available. Provided that the length of the taper is sufficient (the length of portion 106 is about 4 mm), there is no additional loss incurred by the use of the TEC fiber; however, the numerical aperture (angular acceptance) of the fiber is slightly decreased. As a result, it is likely that increases in the core size will be limited to a factor of 2-3. TEC fibers are discussed in greater detail in Kihara and Haibara (1996) J. Lightwave Technology 14:2209- . 2214, and Finegan, U.K. Patent No. GB 2,219,869, each of which is incorporated by reference in its entirety. Yet another potential structure is to provide a focusing microlens in operative association with the core of each fiber. This structure is illustrated schematically in Fig. 19. Here lens 280 is placed in front of each fiber 282 with the fiber core 284 within the focal length of the lens. Such a structure is shown in Martin, U.S. Patent No. 6,284,695, the disclosure of which is incorporated by reference in its entirety. Referring now to Fig. 20, in order to use an (de)multiplexer described above with respect to Figs. 2-11, with the interleaver as contemplated by the present invention, either of the multi-channels output fibers or the single channel output fibers should preferably be skewed relative to one another to accommodate an effective channel shift resulting from the channel separator of the interleaver. The channel shift can be seen by comparing the relative positions of the odd channel lOOGHz signal 313 and the even channel lOOGHz signal 315 in

Fig. 20. Figs. 21A-21C illustrate how the channel shift can be accomodated. In the multi- input device described above and illustrated schematically in Fig. 12, mutli-channel or input fibers 312, 314 are vertically stacked as illustrated in Fig. 21 A and they produce a vertically and horizontally aligned output matrix illustrated by the two dimensional array of single channel fibers 322, 324. When the interleaver is used to affect a channel separation, either the multi-channel input fibers 312, 314 should be displaced one-half of the input signal channel spacing as illustrated in Fig. 21B (or 41 μ for a lOOGHz signal) or all the single channel output fibers 322, 324 should be offset one-half the output channel spacing as illustrated in Fig. 21C. This latter configuration is considerably less desirable because of inherent fabrication complications. That is, the channel spacing of the output signals increases from right to left as illustrated in Fig. 21C. By way of example, the channel separation between the first pair of adjacent output fibers may be 85μ whereas the channel separation at the left most adjacent output fibers may be 95μ. The second row of output fibers must be offset one-half this amount of channel separation. Obviously, this is more complicated than simply offsetting the input fibers one-half the channel separation of the input signal.

The interleaver 310 naturally produces a flat-top filter function. Interleavers can be provided which cause only a small added insertion loss, on the order of ldB or less. The flat- top filter function of the interleaver multiplied by the Gaussian input signal results in a relatively flat-top filter function, as illustrated in Fig. 22.

A further advantage of incorporating an interleaver into the present invention is that it increases the effective channel separation of the input signal by 100% and broadens the effective pass band filter function by a factor of two. For example, as described above, if the signal input to the interleaver is 50GHz (a 0.4nm channel separation), the resultant output signals are lOOGHz (a 0.8nm channel separation). This increased channel separation at the input allows a dual-input (de)multiplexer configured for demultiplexing lOOGHz signals to be used in combination with the interleaver to separate the 50GHz signals. By way of example, a lOOGHz signal has a .8nm channel spacing at the input and a 50GHz input signal has a .4nm channel spacing. For an otherwise identical (de)multiplxer designed for a lOOGHz signal to accommodate the 50GHz signal, the output spacing of the fibers would have to be decreased by 50%, the focal length of the device would have to be doubled or the angular dispersion would need to be increased. In any case, the pass bands would be twice as narrow, limiting high data rate transmissions. Instead, with the present invention, the interleaver has the effect of changing the 50GHz spacing input signal to two lOOGHz spacing input signals having an .8nm channel spacing and no change need be made to the (de)multiplexer configuration. Thus, the present invention has the further advantage of allowing 50GHz spacing signals to be dispersed with a (de)multiplexer configured to separate lOOGHz signals.

Claims

CLAIMS What is claimed is:

1. An apparatus for use in optical communications systems, comprising: an optical interleaver comprising: an input port for receiving from a multi-channel optical fiber a multi-channel optical signal having a first channel spacing x and comprising a plurality y of interleaved channel subsets; and a plurality y of output ports, each for transmitting one of the plurality of interleaved channel subsets as a multiplexed optical subsignal having a second channel spacing equal to x times y; and a multiple-input grating-based (de)multiplexer optically coupled to said interleaver, comprising: a pluralty y of input ports, each for receiving one of the multiplexed optical subsignals from a corresponding output port of said interleaver; and a plurality of output ports, each for transmitting a single-channel optical signal.

2. The apparatus of claim 1, wherein said plurality of (de)multiplexer input ports are displaced relative to each other.

3. The apparatus of claim 1, wherein: said plurality of (de)multiplexer output ports are arranged in y rows; and the output ports of a first row are displaced horizontally relative to the output ports of a vertically adjacent row.

4. The apparatus of claim 1, wherein said multiple-input (de)multiplexer comprises a dual-input grating-based (de)multiplexer.

5. The apparatus of claim 4, wherein said plurality of (de)multiplexer input ports are displaced one-half the channel separation of the input signal relative to each other.

6. The apparatus of claim 4, wherein: said plurality of (de)multiplexer output ports are arranged in two rows; and the output ports of a first row are displaced horizontally by one-half the channel separation of the output signal relative to the output ports of a second row.

7. The apparatus of claim 1, wherein; the multi-channel optical signal comprises an even channel subset and an odd channel subset; and a first output port of said optical interleaver transmits the even channel subset and a second output port of said optical interleaver transmits the odd channel subset.

8. A method for use in optical communications systems, comprising the steps of: receiving at an input port of an optical interleaver a multi-channel optical signal having a first channel spacing x and comprising a plurality y of interleaved channel subsets; transmitting each of the plurality of interleaved channel subsets as a multiplexed optical subsignal having a second channel spacing equal to x times y; receiving at each of a plurality y of input ports of a multiple-input grating-based

(de)multiplexer, optically coupled to the interleaver, one of the multiplexed optical subsignals from a corresponding output port of said interleaver; and transmitting a single-channel optical signal from a plurality of output ports of the (de)multiplexer.

9. The method of claim 8, further comprising the step of displacing the plurality of (de)multiplexer input ports relative to each other.

10. The method of claim 8, further comprising the steps of: arranging the plurality of (de)multiplexer output ports are arranged in y rows; and displacing the output ports of a first row horizontally relative to the output ports of a vertically adjacent row.

11. The method of claim 8, wherein y=2.

12. The method of claim 11, further comprising the step of displacing the plurality of (de)multiplexer input ports one-half the channel separation of the input signal relative to each other.

13. The method of claim 11 ,further compring the steps of: arranging the plurality of (de)multiplexer output ports in two rows; and displacing the output ports of a first row horizontally by one-half the channel separation of the output signal relative to the output ports of a second row.

14. The method of claim 8, wherein: the multi-channel optical signal comprises an even channel subset and an odd channel subset; and said step of transmitting each of the plurality of interleaved channel subsets comprises the steps of transmitting the even channel subset from a first output port of the optical interleaver and transmitting the odd channel subset from a second output port of the optical interleaver.