`
`PROVISIONAL APPLICATION COVER SHEET
`
`ca-a>’~<“‘
`
`illrrwv A
`
`I
`
`A
`-
`—%‘}‘,'
`':::.""'--a-'
`<==:-'g_‘=i’>
`;..'“‘.‘.‘'_''''___~.....——;''—.-.,__ N This is a request for filing a PROVISIONAL APPLICATION under 37 CFR 1.53 (c).
`~
`:->“._;t”"——-c:
`DOM. -
`""--........»-—(I1
`‘cni.
`Z T}
`
`
`
`206692
`
`box —>
`
`
`
`SMITH
`
`INITIAL
`
`STATE OR FOREIGN COUNTRY)
`Norcross, Georgia
`David
`Norcross, Georgia
`John
`Fariborz
`Norcross, Georgia
`
`
`o E
`5:
`:21:
`"7 %
`-P9 “:29
`W3.’ ;__==.-=3;
`:3
`M “gm
`‘
`-__2ca
`
`0’
`
`llllllclilllll
`
`
`
`
`
`
`
`
`
`
`TITLE OF THE INVENTION (280 characters max)
`VARIABLE TRANSMISSION MULTI-WAVELENGTH OPTICAL SWITCH
`CORRESPONDENCE ADDRESS
`
`Customer Number 23460
`i
`§«
`
`El
`
`Leydig, Voit & Mayer, Ltd.
`
`Two Prudential Plaza, Suite 4900
`y
`y
`
`
`
`«
`/I I
`180 North Stetson
`
`Chicago, Illinois 60601-6780 USA.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`ENCLOSED APPLICATION PARTS (check all that apl )
`Number of Pages: 12
`I Power of Attorney
`IX] Specification
`(including any claims and abstract)
`l:l Assignment
`IX Drawins
`Number of Sheets: 11
`El Small Entity Status
`
`METHOD OF PAYMENT check one)
`
`CI A check is enclosed to cover the Provisional filing fee of $150.00.
`IX} The Assistant Commissioner is hereby authorized to charge the filing fee of $150.00 to Deposit Account Number 12-1216.
`
`The Assistant Commissioner is hereby authorized to charge any fee deficiency or credit any refund to Deposit Account
`
`Number 12-1216. A duplicate copy of this communication is enclosed for that urpose.
`
`
`
`
`1:] Other (specify)
`
`
`
`
`
`
`
`The invention was made by an agency of the United States Government or under a contract with an agency of the United States
`Government:
`
`E
`[:I
`
`No.
`Yes, the name of the U.S. Government agency and the Government contract number are:
`
`Respectfully submitted,
`
`.K Chang, RegistrationNo.41,397
`
`I 5
`
`One of the Attorneys for Applicant(s)
`LEYDIG, VOIT & MAYER, LTD.
`Two Prudential Plaza, Suite 4900
`180 North Stetson
`
`Chicago, Illinois 60601-6780
`(312) 616-5600 (telephone)
`(312) 616-5700 (facsimile)
`
`Additional inventors are being named on separately numbered sheets attached hereto.
`E]
`
`CERTIFICATION UNDER 37 CFR 1.10
`
`I hereby certify that this correspondence and the documents referred to as attached or enclosed therein are being deposited with the United States Postal
`on September 22, 2000,
`in an envelope as “EXPRESS MAIL POST OFFICE TO ADDRESSEE” service under 37 CFR l 10, Mailing L
`
`EL643535582US, addressed to Co onaofPatents and Trademarks, Washington, DC. 2023 .
`PrintedNameofPerson Signing
`Ea
`
`PROVISIONAL (Rev 6/'29/2000)
`
`
`
`FNC 1010
`
`
`
`
`
`
`
`Variable Transmission Multi—wavelength Optical Switch
`
`Inventors:
`
`Dave Smith, John E. Golub; Fariborz Farhan
`
`Assigned to: MOVAZ Networks, Inc.
`
`5445 Triangle Parkway
`
`Norcross, GA 30092
`
`Background of the Invention
`
`In response to the need for increased communications bandwidth, optical fiber systems
`
`operating at wavelengths near 1.5 um are rapidly replacing conventional copper-conductor
`
`electronic systems. Advantages of lightwave technology include increased signal bandwidth (10
`
`GHZ rates and higher are possible along a single fiber optic channel), immunity to electrical
`
`interference and the possibility of system expansion through wavelength division multiplexing
`
`(WDM) which superimposes more than one optical carrier on a single fiber channel.
`
`In a conventional, point-to-point, single—channel system, an electrical input signal
`
`modulates the output of a semiconductor laser. The resulting lightwave signal is coupled into a
`
`single-mode optical fiber that transports it to an optical receiver. En route, one or more
`
`optically—pumped, Erbium-doped fiber amplifiers (EDFA's) or other in—line optical amplifiers
`
`may be used to compensate for fiber absorption and other losses. Fiber optical components are
`
`used for various functions such as traffic distribution and signal routing. Finally, at the receiver,
`
`the beam exiting the fiber is focused onto a detector that converts the lightwave signal to an
`
`electronic version of the transmitter input signal.
`
`Multiple information signals may also be combined on a single fiber using wavelength
`division multiplexing (WDM).
`In this scheme, individual transmitters operate at a different,
`
`fixed wavelengths and signals are combined using all-optical multiplexers that couple the light
`
`from several input ports into a single output fiber. At the receiver, an optical wavelength
`
`separator spatially separates the individual, sing1e—wavelength channels. Various means of
`
`multiplexing and demultiplexing multiple optical carriers on a single fiber are disclosed in ———
`
`
`
`
`
`review paper on WDM technology. Dedicated detectors are then used to convert the information
`
`on each channel to an electronic format.
`
`WDM systems offer wide bandwidth capabilities and the possibilities of upgrading system
`
`capacity without changing the installed fiber base. At the present time, commercially-available
`
`systems can support 40, 2.5GHz channels yielding a total bandwidth of 100 GHz. The optical
`
`system components (fibers, amplifiers, lasers and detectors) can support a much larger
`
`bandwidth by far and it is expected that future systems will operate at higher data rates (40 GHz,
`
`for example) and include 80 or more WDM channels.
`
`Components required for the realization of a multi—wavelength optical network include
`
`simple wavelength multiplexers (MUX) and wavelength demultiplexers (DEMUX), add/drop
`
`multiplexers (ADM's), optical cross connect switches (OXC's), in-line optical amplifiers and
`
`associated gain control technologies, optical performance monitors and wavelength channel
`
`power equalizers. Wavelength conversion is another area of active development.
`
`Present—day optical WDM networks are based on components that often disadvantageously
`
`utilize the optical—to-e1ectronic—to—optical conversion process for signal regeneration, switching,
`
`power equalization and other network functions. System bandwidth and particularly, system
`
`upgradability is constrained by the complexity and cost of this pervasive electrical-to—optical
`
`conversion requirement. Transparent, all-optical components would effectively address this
`
`limitation, facilitating the design of networks in which the optical signal leaving a transmitter
`
`would not be converted to the electrical domain until it reached a receiver. A feature of the
`
`disclosed invention is that it provides an optically transparent solution to the system requirement
`
`for wavelength channel power equalization.
`
`Various WDM components are described in the following as a means of introduction.
`
`Wavelength MUX's are used to combine the output beams from a number of single frequency
`
`transmitters into a single, wavelength-divisionanultiplexed signal. DEMUX's perform the
`
`inverse function at the receiving stations, physically separating a WDM signal into single
`
`frequency beams. Both MUX and DEMUX units may conventionally use fiber or integrated
`
`optical components in order to isolate individual WDM channels. A variety of multiplexing
`
`techniques using variously, all-fiber components, integrated optical circuits, thin films and
`
`diffraction gratings have been developed.
`
`
`
`
`
`
`
`In a typical fiber-optic network application, optical signals occupy a plurality of fiber lines
`
`and must be interconnected for signal distribution, multiplexing and demultiplexing of signals
`
`from numerous locations, protection routing and other functions. Such coupling and switching of
`
`signals amongst fibers is realized by optical switches or crossconnects. Optical cross connect
`
`switches (OXC's) provide optical connections between input and output ports that may be
`
`reconfigured in response to control signals. In WDM applications, OXC’s may be used to
`
`connect individual wavelength channels from a single input port to different output ports. If the
`
`optical crossconnect further distinguishes among wavelength channels so as to allow
`
`redistribution of wavelength content among fibers in a network, this more specific crossconnect
`
`is called a wavelength crossconnect.
`
`A variety of crossconnects have been disclosed using a number of different optical
`
`technologies. Cross connect designs include devices based on optical interferometry (Mach-
`
`Zehnder and directional coupler switch arrays, SOA-based switches, A0, E0 and liquid crystal
`
`switches, thermo-optical (silica-on-silicon) and polymer optical waveguide switches and
`
`optomechanlical switches including those based on MEMS (micro—electro—mechanical systems).
`
`US Patent #6,097,859 to Solgaard (Solgaard), et.al. discloses a Inultiwavelength cross-
`
`connect switch based on an array of MEMS mirrors. In this device, optical WDM signals are
`
`received by a plurality of input ports. An input lens system collimates these beams and directs
`
`them to a diffraction grating that reflects different wavelength channels at different angles while
`
`preserving the spatial separation of the individual input ports. The resulting two-dimensional
`
`array of beams (the width equals number of input beams and the height equals the number of
`
`wavelength channels) is imaged onto an array of electronically-actuated micro-mirrors (MEMS
`
`array). Each beam is reflected by an individual micro-mirror at an angle that is a function of the
`
`applied voltage.
`
`In a first embodiment of Solgaard’s patent, the electronically—controlled elements of a
`
`second micro—mirror array redirect the beams from the first MEMS device to an output optical
`
`system.
`
`Single-wavelength beams are reflected by the second array and recombined into
`
`WDM outputs signals by a second grating. Optical lenses on either side of the grating collimate
`
`the beams from the MEMS array and couple the WDM beams to the output ports of the switch.
`
`According to Solgaard the output optical system may be a mirror-image of the input system. By
`
`adjusting the voltage on the two MEMS arrays, a variable connection between input and output
`
`
`
`ports may be established for each wavelength channel in the system. These connections are,
`
`however, subject to the constraint that no two inputs in a single wavelength channel may be
`
`connected to a single output.
`
`In alternative embodiments, a planar fold mirror is used to eliminate the second MEMS
`
`array and output optical system. A fold mirror can be used to reduce the lens and diffraction
`
`grating component count and to make a more compact geometry. In this case, for an NXN fiber
`
`wavelength crossconnect, 2NxW mirrors are required, where W is the number of wavelength
`
`channels being switched.
`
`Optical add/drop multiplexers (OADM’s) allow selected wavelength channels to be added
`
`or dropped from a WDM lightwave signal. In its simplest embodiment, an OADM has 4 ports —
`
`‘input’ ,
`
`'outp11t', ‘add’ and ‘drop’. Typically, an optical trunk line enters the switch via the ‘input’
`
`port and exits at the ‘output’. Individual wavelength channels may be switched from the ‘input’ to
`
`the ‘drop’ port or from the 'add' port to the ‘output’. Connectivity may or may not be provided
`
`between the ‘add’ and 'drop‘ ports. Note that the OADM is an application-specific example of a
`
`general 2x2 fiber—optic wavelength switch, which is the lowest reasonable p0rt—count wavelength
`
`crossconnect (WXC). Larger port—count VlD(C’s allow one of several add inputs to be switched
`
`to the output while the input signal is connected to one of several drop ports.
`
`US Patent 5,960,133 to Tom1inson(Tomlz'nson) discloses a MEMS—based ADM switch that
`
`also uses a grating to separate the inputs into spatially-distinct, single wavelength beams. In
`
`contrast to the Solgaard switch, however, in Tomlinson the beams are switched by a single
`
`reflection from a MEMS array.
`
`The overall WDM network performance is optimized when the power on individual
`
`wavelength channels is uniform across the channel spectrum. Factors contributing to the
`
`inequality of WDM channel powers include source power nonuniformity, different path routing
`
`of the channels which comprise the WDM spectrum, differential wavelength-channel gain in
`
`optical amplifiers, wave1ength—dependent loss along the fiber path and within the components
`
`which comprise the overall network.
`
`Generally, the lowest channel power determines the system performance in terms of
`
`crosstalk and signal-to—noise ratio, so it is necessary to adapt the system to accommodate the
`
`weakest channel. The preferred method is to equalize channel powers to that lowest channel
`
`power level.
`
`
`
`
`
`Thus, it is advantageous to actively compensate for the effects of wavelength-dependent
`
`amplifier gain in optical amplifiers and wavelength-dependent transmission loss in a WDM
`
`system. Typically, fiber amplifiers are themselves optimized for uniform gain across the WDM
`
`spectrum (reference). Conventionally, optical power equalization may be accomplished by
`
`sensing the power of each wavelength channel and adjusting the power to the appropriate
`
`transmitter. Realization of this scheme is complicated by the fact that transmitters and power
`
`measurement instruments are often separated by large distances — a situation that makes the
`
`associated electronic power control system both complex and expensive. Therefore, it is
`
`preferred to locally resolve the power inequalities at the OXC level, which also prevents
`
`propagation of channel power inequities across the network. Propagation of power transients can
`
`have an extremely deleterious effect of WDM system performance (reference).
`
`A self—contained unit consisting of an optical performance monitor (OPM), control
`
`electronics and a rnulti-wavelength optical loss modulator may also be used for WDM power
`
`equalization. References to an 0PM and power equalization schemes. In operation, the OPM
`
`measures the power in each WDM channel and transmits this information to a control processor.
`
`The processor then adjusts the wavelength-dependent transmission of the modulator to flatten the
`
`power distribution curve. While eliminating the need for a distributed control system, this
`
`system requires an additional optical component (the loss modulator) to be added to the optical
`
`network.
`
`A multi—wavelength OXC or ADM with electronically variable transmission functions for
`
`each wavelength charmel would eliminate the need for a dedicated optical loss modulator in a
`
`self-contained power equalizer. To our knowledge, variable-transmission optical switching
`
`components are unknown in the prior art.
`
`It is, therefore, a primary object of this invention to
`
`provide improved, rnulti—wavelength, OXC and ADM switches with electronically-controllable
`
`transmission.
`
`Brief Description of the Invention
`
`Electronic control of the optical transmission of individual wavelength channels in a multi-
`
`wavelength, MEMS optical switch is achieved by intentionally spoiling the spatial mode match
`
`between the free—space beams inside the switch and the output optical coupler. This technique
`
`may be applied to any optical switch regardless of the number of inputs, wavelength channels, or
`
`
`
`
`
`the specific switching scheme employed. The invention is further not to be construed as
`
`restricted to those switches containing a MEMS optical beam switching methodology.
`
`According to a preferred embodiment of the invention, the optical throughput of each
`
`Wavelength channel may be controlled by using a mirror array with elements that can be rotated
`
`in an analog fashion about two orthogonal axes. Angular displacement in a first, switching
`
`plane, is used to perform an OXC, ADM or other switching function while angular displacement
`
`about the orthogonal axis is used for power control. MEMS switches with single axis mirror
`
`arrays may also be used for output power control. In this case, the coupling of each beam to a
`
`designated output port is adjusted by variations in the switching angle.
`
`A complete power equalization system incorporating a Variable-transmission MEMS
`
`switch is also disclosed. In this system, an optical performance monitor measures the power
`
`spectra at the switch output ports and transfers the data to a control processor. This unit
`
`generates electronic signals that adjust the angles of individual micro-mirrors within the switch
`
`to optimize the output power spectra.
`
`Brief Description of the Figures
`
`Figure 1 is an optical schematic diagram of a 2x2 ADM with variable transmission according to
`
`the invention
`
`Figure 2 is a diagram detailing the switching function of the ADM
`
`Figure 3 is a diagram detailing optical power control operation of the ADM.
`
`Figure 4 is a power equalizing ADM system incorporating the switch of Figure 1.
`
`Figure 5 is a figure showing optical crossconnect components in their various roles in a
`
`transparent WDM network including multi—fiber wavelength crossconnects (top left), optical
`
`add-drop multiplexers, a multiple—LAN distribution network (lower left), a local access
`
`wavelength OADM node.
`
`Figure 6 is another view of the OADM 3—space cross connect design including a fold mirror.
`
`Figure 7 shows the mirror tuning scheme used for switching and equalization on orthogonal
`
`directions, based on calculations for the design of figures 8 and 9.
`
`Figure 8 is the package of one design of a crossconnect showing dimensions.
`
`Figure 9 is a to-scale figure of the crossconnect of figure 8, with the fold mirror out of plane and
`
`separated by 10 mm.
`
`
`
`
`
`Figure 10 is a schernatic of the power equalization scheme described in ths invention, applied to
`
`the specific case of a folded-mirror multiwavelength crossconnect.
`
`Figure 11 shows electronics circuitry which is appropriate to the drive of the MEMS mirrors for
`
`the purposes of power equalization.
`
`Description of the Preferred Embodiments
`
`Referring to Figure 1, a 2x2 multi-wavelength, add-drop multiplexer (ADM) 100
`
`embodying the features of the invention is schematically shown. (100 is not shown in the figure
`
`— DS) Optical fibers 105 connect the four ports of the switch to an optical network. Signals may
`
`enter the switch through the input port 110 or add port 114 and exit through the drop port 1 16
`
`and/or output port 112. All fiber channels are wavelength division multiplexed with 8
`
`wavelength channels in the 1.5 um communications band that are separated by a wavelength
`
`increment, AX, suitable to WDM communication systems, referring to approximately 0.4 and 0.8
`
`nm respectively, more specifically 50-100 GHZ apart in frequency for dense WDM systems.
`
`In a typical application the ADM 100 is used to add or subtract individual wavelength
`
`channels from a signal that enters through the input port 110 and exits through the output port
`
`112. Under the control of an external control signal, the ADM may allow a wavelength channel
`
`to pass through the switch from input to output or, alternatively, to route the input signal to the
`
`drop port and simultaneously connect the add port to the output.
`
`In Figure 1, signals entering the switch 100 are coupled into an optical concentrator 120
`
`that decreases the physical spacing between the input and output channels. (This is the subject of
`
`another invention — how should I handle that ~ DS?) Typical single mode fibers have a diameter
`
`of 125 um and the I/O ports must be spaced at least this far apart. The concentrator of another
`
`invention of some of the authors typically uses integrated waveguides to reduce this spacing to
`
`approximately 30 um at the output surface. (Should my and John’s notes be incorporated into
`
`the invention disclosure as an appendix - DS) The optical lens 125 collects the beams exiting the
`
`concentrator, collimating them nominally into a single beam that is reflected by the diffraction
`
`grating 130. This grating is oriented with its grooves parallel to the page of the figure and the
`
`individual wavelength channels are dispersed into a fan of beams in the direction perpendicular
`
`to the page. The lens 135 focuses the light from the diffraction grating to form two, 8-element
`
`columns of beams that are reflected by the input and add columns of the micro-mirror array 140
`
`
`
`
`
`onto the fold mirror 145. Each array column has 8 -elements, corresponding to the 8 wavelength
`
`channels in the input and add beams. The spacing between the lens 135 and fold mirror 145 is
`
`adjusted to locate the Gaussian waists of the input and add beams at the mirror surface. (Why are
`
`we restricting ourselves to folded systems?) - DS
`
`The fold mirror reflects the individual beams back to the MEMS array where they are
`
`reflected by the micromirrors in the 8—ele1nent output and drop columns. These beams are then
`
`collimated by the lens 135 , remultiplexed by the grating 130 and focused onto drop and output
`
`channels of the optical concentrator 120 by the lens 125, according to the individual control
`
`signals applied to determine the switch states independently for each wavelength channel..
`
`Switching of the input and add beams in an individual wavelength channel is accomplished
`
`by adjusting the angle of two mirrors in the MEMS array. Two states are possible — a first state
`
`in which an input channel is coupled through the switch to the output and a second state in which
`
`an input channel is coupled to the drop port and an add channel is coupled to the output.
`
`Figure 2 is a detailed diagram of the beams in a single wavelength channel as they are
`
`reflected by the rnicromirror array 160 and fold mirror 165. The input beams are far enough
`
`from the focusing lens to be spatially distinct as they enter the figure. The elements of the
`
`micromirror array are adjusted to couple light from the input beam 150 to the oppositely-directed
`
`drop beam 154 and the add beam 156 to the output 152.
`
`The input beam 150 travels from the input micrornirror 172 to the fold mirror and returns to
`
`the drop micromirror 176. The angles of the two elements of the micromirror array are
`
`electronically adjusted so that the drop beam propagates in essentially the opposite direction
`
`from the input. The optical path lengths between the focusing lens and fold mirror are adjusted
`
`to focus the input beam on the fold mirror, i.e. the beam waist is in the plane of the mirror.
`
`Similarly, the add beam 156 is directed by the micromirror 174 to the fold mirror in such a way
`
`that it is reflected to the micromirror 178 that couples it to the output port. The waist of this
`
`beam is similarly located in the plane of the fold mirror.
`
`Advantageously, the micromirror array and fold mirror are separated by a distance that is
`
`equal to the Raleigh range of the input and add beams. The Raleigh range is defined as the
`
`distance from the Gaussian beam waist after which the beam diameter has increased in size by a
`
`factor of root 2. It is equal to W032/7». This condition minimizes the ratio of spot size to beam
`
`separation at the micromirror plane, minimizing the cross-talk between beams and losses due to
`
`
`
`
`
`leakage around the mirror edges. This invention does not require that the separations between the
`
`micromirror array and the fold mirror is equal to the Raleigh range, only that that separation is a
`
`preferred maximum.
`
`In order to control the power of the beams leaving the switch Via the output and drop ports,
`
`the corresponding micro-mirror elements are designed to tilt about a second, perpendicular axis
`
`in the plane of the array. Tilting a pair of elements about this axis translates the corresponding
`
`beam on the surface of the optical concentrator, thereby decreasing the percentage of the light
`
`coupled out of the switch. By displacing the optical beam perpendicular to the line defining the
`
`optical ports, power coupling is decreased in either direction While not sacrificing the degree of
`
`isolation between optical channels. That is why it is preferred that the misalignment is
`
`orthogonal to the optical access line. One can misalign within the access line, but that
`
`misalignment will typically increase coupling into other ports which is negligible up to a point
`
`but deleterious thereafter.
`
`Figure 3 is a schematic diagram illustrating this process. In the Figure, the micro—mirror
`
`array 200 has four columns corresponding to the four I/O ports and 8 rows corresponding to the
`
`individual wavelength channels in an 8~wavelength WDM system. As an example, we consider
`
`the switching a single wavelength input beam 205 to a drop beam 210. Not shown in the figure
`
`is the add beam that would be switched to the output port in the preferred embodiment.
`
`To switch the input beam, the micro-mirror in the 'i' column of the array 220 is tilted about
`
`a first, vertical axis to steer the reflected beam towards the drop output. After reflection by the
`
`fold mirror 215, the beam strikes a second, 'd' element 225 of the array. The vertical—axis tilt of
`
`this element is adjusted to efficiently couple the beam leaving the array to the drop channel 230
`
`of the optical concentrator 235. In the demonstrated embodiment, an angular displacement of
`
`3.25 degrees is required.
`In the example of Figure 3, the percentage ofthe drop beam that is coupled into the power
`
`drop channel 230 of the optical concentrator 235 may be Varied by tilting the individual array
`
`elements about their horizontal or Vertical axes. In normal ‘drop’ operation, one would adjust
`
`the angle of the ‘i’ mirror element 220, primarily in the horizontal direction to optimize
`
`placement of the input beam on the drop mirror element 225. The angle of the 'd' element is then
`
`adjusted to align the beam waist with the output plane of the concentrator. In this way, the
`
`vertical displacement introduced by the 'i' element repositions the waist at the concentrator in a
`
`
`
`
`
`predictable and controllable manner. Displacement of the beam from a condition of perfect
`
`alignment decreases the percentage of light coupled into the concentrators drop channel,
`
`reducing the amount of power leaving the switch. In the preferred operation of the switch,
`
`optimal alignment will be performed by optimal positioning of the beam on the appropriate
`
`concentrator element, specifically by controlling the horizontal alignment. Vertical misalignment
`
`is the preferred method of reducing power coupling to the chosen output channel. As an
`
`example, for singlemode fiber cores illuminated by 1.5 micron light, misalignment in any
`
`direction away from the fiber core by one micron resultis in about 1 dB of attenuation, and
`
`further misalignment further exacerbates this coupling degradation in a dramatic manner well
`
`known to those skilled in the art. Angular displacement off the waveguide axis is another, less
`
`preferred means of reducing coupling.
`
`In other embodiments of the invention, 2—axis MEMS arrays may be used to control the
`
`output of prior art switches. For example, they may be used to add output power control
`
`capability to the prior art designs disclosed in Tomlinson and Solgaard. A 3x3 (that is, 3 input
`
`port by 3 output port) power-controlled OXC switch with N wavelength channels may also be
`
`constructed using 6 x N two—axis mirror array and a switching scheme similar to that of Figure 2.
`
`Physically, the number of input/output ports is limited by the maximum tilt angle of a single
`
`array element. Devices having many elements may be constructed using conventional
`
`technology. The number of wavelength channels is limited by IC fabrication technology and
`
`may exceed 100.
`
`Fabrication of silicon MEMS arrays that can be tilted about two orthogonal axes is
`
`described in the prior art (see, for example, "Development of a silicon 2-axis micro-mirror for
`
`optical cross-connect" by Andrew S. Dewa, and John W. Orcutt published on pp. 93~96 of the
`
`Technical Digest of the Solid State Sensor and Actuator Workshop, Hilton Head Island, SC, June
`
`4-8, 2000). In an exemplified embodiment of Figures 1,2 and 3, a 4 x 8 array of 400 um x
`
`400nm mirrors with a center spacing of 495 pm in the wavelength direction and 595 pm in the
`
`orthogonal, switching direction may be used. In the design of the attached figures showing a
`
`particular embodiment of a crossconnect, power variations of 10 dB should be able to be
`
`achieved with a 0.3 degree tilt about the orthogonal direction.
`
`Figure 4 is a schematic block diagram of a complete power—controlled switching system
`
`incorporating the preferred embodiment of Figure 1. 8-channel, WDM signals are transported to
`
`10
`
`
`
`
`
`the ADM switch 305 by single—mode optical fibers 310 (Note — we do not have to restrict
`
`ourselves to singlemode), entering through the input port 312 and add port 314. Signals in each
`
`wavelength channel are routed to the output 316 and drop 318 ports under the control of the
`
`electronic switching input signal 320. Optical fibers 323 are used to transport the WDM signals
`
`from the switch to other network components and to taps 325 that direct a small portion of the
`
`output signals to an optical performance monitor (OPM) 327. This unit measures the power in
`
`the individual wavelength channels of the 'output‘ and 'drop' signals and transmits the resulting
`
`spectral data to the control processor 330. This processor compares the power spectra to optimal
`
`distribution fiinctions and generates power control signals to correct the deviations. These
`
`signals are transported to the ADM switch on the power control signal line 335 which controls
`
`the tilt angle of the individual MEMS elements about the power control axis. This resulting
`
`feedback loop may be used to actively optimize the power spectra of the signals leaving the
`
`ADM switch. Note -— optimizing the output powers should be construed generally. As a rule, one
`
`would prefer to maximize the power on each channel, but, for reasons of system uniformity, it is
`
`preferred to equalize the powers of all channels which are within a specified range of power
`
`(some too-low or too—high power signals may need correction outside the proposed means — e.g.
`
`dead lasers) which means adjusting all channel powers until the equal the weakest acceptable
`
`channel power —- this is common in the current art. For other system reasons, exact equalization
`
`may not be preferred, but the current invention is capable of providing programmable power
`
`control to adapt to all system requirements. In an alternate embodiments of the power
`
`equalization system, the wavelength-dependent transmission of optical switches incorporating
`
`single-axis MEMS arrays may be varied by adding the control processor output to the switching
`
`signal. In this case, the optical coupling between the free space beam inside the switch and the
`
`output is port is adjusted by translating the beam in the switching plane. In comparison to the
`
`two-axis embodiment, single axis systems may be realized using simpler, single axis MEMS
`
`arrays but suffer from increased potential for crosstalk between channels.
`
`The invention is embodied by any multi~wavelength MEMS optical switch that uses the tilt
`
`angle of individual micro-mirrors to provide independent control of the transmission of each
`
`wavelength channel. Both single and dual axis mirror arrays may be used in a variety of
`
`switching configurations, although the two—aXis components are preferred. Switching schemes
`
`include, but are not limited to, those described in Tomlinson and Solgaard in addition to the
`
`11
`
`
`
`
`
`preferred scheme described above. More generally, the invention discloses the use of
`
`controlled misalignment in fiber optical devices in order to introduce a controlled loss for the
`
`purposes of enabling power equalization. This is not restricted to WDM systems. It is further not
`
`restricted to matching power levels of optical channels but for achieving any desired channel
`
`power profile according to network management needs.
`
`
`
`12
`
`
`
`
`92>?:o_mm_Ema§0min
`
`...>H.
`
`
`
`£8‘Nouswfi
`S/m.
`
`:EmmwEm:m.flLoE,mEm>
`
`
`
`.so
`
`55%mBswfi
`
`wmflfimcfirwo5mEw>
`
`
`
`
`v8&5
`
`oAE
`
`mégamgc%§>
`
`J$..w..u..é.%
`
`
`
`DAD:
`
`
`
`wemmoou<&-w\1%%-.»i.:aemN2:
`
`mos:OhZ<A88%HEB
`
`oofifififiKOQmmm
`
`8%N83
`
`
`
`
`
`somawopwwmxonfmfi53>
`
`
`
`Hoosaoommouooboz:
`
`
`
`vEo>BoZE8332_.oo§%<
`
`
`
`
`
`
`
`
`
`.838wasxomwsufimmouomafifl
`
`
`
`%.:<fiésmnoézmzmz5%seaon
`
`8:?
`
`
`
`oEwowsmaoxo85802:5HUGM
`
`
`
`
`E:me2m22.3E.53502
`
`
`
`3:533mumsSE2
`
`
`
`3:533mum».no.32o
`
`a£353:3.838OWNu_o<
`
`Ea§E-m_o
`
`mSo
`
`5
`
`no.5
`
`§Q<O
`
`_
`
`
`
`.SO+»_oU<
`
`ao5Tc_
`
`
`
`w:Es,HHOCJZB25%
`
`
`
`
`
`mwsfiowgm3:0mwoo:§Q<Oz<q-2%_:2"802
`
`
`
`
`
`
`
`
`
`
`
`ommxommBo::oo..mm8UE330
`
`.w«AAU.w»mm...
`
`:2:52...
`
`wEswfi
`
`
`
`mmmoom.5nm..E:2
`
`%_:w®>§
`
`poambsoocoo
`
`BE
`
`
`
`
`
`5.38E3.955
`
`B£mo>§>m§m2
`
`
`
`Ewan325Hm
`
`wnumhbH0
`
`5.32A2
`
`mamaHA
`
`mPfimfi
`
`
`
`
`
`3&3oewam_§ao.I2o<o
`
`
`
`
`
`
`moummzmswmSfiomHEBEHONO
`
`pooacoo$80
`
`35>:
`
`fa
`
`
`
`
`
`owmmswo8mpomfioSEEpmsméIWBBOQ.8::o§
`
`
`
`3owswwm
`
`
`
`
`
`
`
`9/ED>5548.32m2m:Z
`
`mo_:Wo.:o2m_
`
`S2&5
`
`
`
`SowsoummomSEEHm::.<
`
`
`
`m2Z80¢EmmaBfim
`
`20%
`
`SE0Eom$33Eafio
`
`
`
`