This invention relates to an optical space switch.

R.A. Spanke in a paper entitled Architectures for

Large Nonblocking Optical Space Switches, IEEE Journal of Quantum Electronics Vol QE-22, No.6, June 1986, discusses the use of active, N to 1 combiners fabricated from optical space switch elements on Ti:LINb0 ₃ for use in nonblocking N to N optical space switch architectures. Such known N to 1 space switches are capable of selectively switching a selected one of N optical signals to an output.

According to the present invention an optical switch comprises - an optical output; n optical deflection stages each comprising a polarisation rotator responsive to a respective control signal to rotate the polarisation of an optical signal through 90°, and a deflection means for selectively deflecting optical signals according to their polarisation? the n deflection stages being serially arranged to define 2 ⁿ distinct source locations from where an optical signal is selectively deflectable successively by the n deflection stages to the optical output, a distinct combination of states of the n control signals corresponding to each location?

a plurality of up to 2 ⁿ optical sources each located at one of the 2 ⁿ source locations; and in which each deflection stage preserves the spatial integrity of the deflected optical signals. An advantage of the present invention over known optical switches capable of connecting one of N inputs to an output is that it can switch optical signals having up to two spatial dimensions. For example, each optical signal might be a complete databus or represent a page of computer memory so providing a means of optically switching information at a greatly increased rate.

All the optical signals which enter a given one of the deflection stages pass through the same polarisation rotator which therefore operates on spatially multiplexed signal paths giving a significant reduction in the effective number of cross-points and which uses few optical and electro-optical subassemblies.

By spatial integrity is meant that although an input signal may be distorted spatially as it is switched to the output, for example expanded or contracted, there is a consistant mapping between the spatial elements of the switched input and output signals which preserves the information content of the spatial modulation.

In this specification, the term "optical" is intended to refer to that part of the electromagnetic spectrum which is generally known as the visible region together with those parts of the infra-red and ultra-violet regions at each end of the visible region capable of having the polarisation rotated. The deflection means may be an optical component which supports propagation of the optical signal in two directions dependant on the polarisation of the signal, for example a bi-refringent crystal such as calcite, cr one which selectively reflects light of a given Dolarisation, such as a Doiarisinα bear, sϋlitter. Ir. this

case a prism may be conveniently used to align incoming signals with the beamsplitter such that all selected signals from the deflection stage pass from the beamsplitter in the direction of the next deflection stage or the output.

Embodiments of the invention will now be described by way of example only by reference to the accompanying drawings on which

Figure 1 is a schematic perspective view of an optical switch according to the present invention using a birefringent crystal deflection means;

Figure 2 is a schematic side view of the optical switch of Figure 1 illustrating its use with collimated optical signals; Figure 3 is a schematic end view of the optical switch of Figure 1 but using divergent optical sources and having additional lenses to focus the sources onto the output;

Figure 4 is a schematic end view of an optical switch using polarisation beamsplitters and prisms as the deflection means; and

Figure 5 is a diagramatic, perspective view of a generalised connection network formed from the N-fold array of optical switches of Figure 4 having optical sources generated by an N signal distributor. Referring to Figure 1, an 8-to-l optical switch 2 is shown which can optically switch a selected one of inputs II to 18 to an output 0. Each of the inputs I and the output 0 are in the form of a linear array of eight optical fibres 4 connectable to an optical network (not shown) the input fibres being of 1.3?m monomode fibre, the output fibres multimode. At any given time the optical signals launched from the inputs I may be spatially modulated according to the signal emitted by the individual fibres 4 of that input I which could be representative cf the instantaneous state of an eicht-bit

computer bus, for example. The fibres 4 of the output 0 receive light from the corresponding fibres of an input I. In this way one of eight, eight-bit optical buses can be switched to the output 0 for passing to the network (not shown) for processing. It will be readily appreciated that the present invention is not restricted to this example of spatially modulated signal ncr this particular application but is equally applicable to other means of forming the spatially modulated signals and to other optical networks requiring such optical signals to be switched to an output.

The optical switch of Figure 1 has three deflection stages SI to S3 each having a twisted-nematic liquid crystal cell polarisation rotator R and a birefringent calcite block deflector D, labled Rl to R3 and Dl to D3 respectively. The calcite blocks Dl to D3 have approximately 1cm square faces with thicknesses of 1.22mm, 2.44mm and 4.88mm respectively with the optical axis of the blocks D at an angle of 48° to the front face, Electronic drives controllers Cl, C2 and C3 each provide a voltage to cause the polarisation rotators R to allow light to pass through with the polarisation unrotated or rotated through 90° as required. The rotators at each deflection stage can therefore be set to rotate or not the linear polarisation of the light entering the stage.

The inputs I are arranged to launch light polarised linearly in the direction of an ordinary ray propagating in the calcite deflectors D.

A lens 6 is interposed between the inputs I and output 0 of a focal length such that the effective optical path length from the lens to the inputs I and output 0 is equal to twice the focal length of the lens. This provides 1:1 imaging magnification between the inputs I and the output G. In the embodiment of Fiσure 1 the lens has a focal

In the case of a twisted nematic liquid crystal a control signal of 0 Volts causes a 90° rotation, whereas a non-zero control signal applied across the liquid crystal cell destroys the twisted structure removing the polarisation - rotation property.

As will be explained in more detail, this optical switch provides eight possible deflections of the input optical in increments of 125?m, the number of increments determing which of the inputs I is deflected to fall on the output 0.

The operation of the optical switch 2 of Figure 1 will now be explained with reference to Figure 2 which shows the side view of the switch 2.

Each deflector stage SI, S2 and S3 operates as the digital light deflector described by .Kulcke, T.J.Harris, K. Kosanke and E. Max in a paper entitled "A Fast, Digital-Indexed Light Deflector" IBM Journal January 1964. If the optical signal entering a calcite crystal is linearly polarised in the horizontal direction it will pass straight through the crystal as the ordinary ray, if polarised in the vertical direction it will be deflected as if propagates as the extraordinary ray to exit the crystal D displaced relative to an ordinary ray but parallel to it. The displacement between the ordinary and extraordinary rays is directly proportional to the thickness of the deflector crystal. Thus by employing stages with thicknesses in the ratio 1:2:4, 8 possible increments of displacement are possible. A set of exa plary rays 10, to 10 _g one from each of the inputs I, and all arranged to be horizontally polarised is shown. If all the rotators R are set to leave the polarisation of light unaltered on passing through it, each ray 10 will remain horizontally polarised and pass through the optical switch 2 undeflected. The located output is located relative to the inputs

such that the optical signal from the optical fibres 4 of input II of Figure 1 will be coupled to the output fibres 4 of the output 0. If the rays entering Dl, D2 or D3 are vertically polarised they will be deflected 1, 2 or 4 125?m units by the stages Dl to D3 respectively.

Consider now rotators Rl, R2 and R3 all set to rotate the polarisation of the optical signal through 90° by the application of the appropriate control signals from the controllers Cl, C2 and C3. The horizontally polarised signals from the inputs I are rotated by rotator Rl to have vertical polarisation and consequently they are all deflected by unit of 125?m away from the undeflected line 11 shown as a dotted lines for ray 10, and 10„. Rotator R2 rotates the polarisations back to the horizontal polarisation so all the signals pass undeflected through the second stage S2. Rotator R3 again rotates the polarisation, this time from horizontal to vertical so the deflector D3 causes a four-unit deflection of all the rays. The net, five unit deflection means the rays from input 6 fall on the fibres 4 of output 0 of Figure 1. The other eight possible settings of the bi-state controllers Cl, C2 and C3 will each selectively cause a respective distinct input I to be connected to the output 0. The different thickness deflectors D can be arranged in any order. Further deflection stages can be provided to provide a larger number of switchable inputs as desired there being 2 ⁿ switchable inputs for an n-stage optical switch. For these larger inputs it may be convenient to locate the rotators at the focal point of a pair of convex lenses of equal focal length to reduce the area of rotator needed to rotate or not the optical signals passing through the stage.

Referring now to Figure 3 there is illustrated hov: the three deflection staσeε S of FJ

divergent beams of which only that from I _g is shown in clarity. Only optical signals from input I- would be directed by lenses 14 and 16 to be focussed on the output 0, the other inputs being focussed to a position displaced from the output 0, when all three rotators 12 are activated. The other inputs I can be selectively switched to the output by activating other combinations of rotators R to provide different total deflections as described for non-divergent beams with reference to Figure 2. The use of divergent beams as shown in Figure 3 is particularly attractive for fibre optic circuits.

Referring now to Fig 4, there is shown the three deflection stages of an optical switch 16 in which the deflection means are formed from a polarisation beamsplitters Bl to B3 and prisms PI to P3 to define eight distinct locations which can be directed the output 0. At each stage half the ray paths enter a beamsplitter B in one direction, the other half at right angles to that direction via the corresponding prism P. All the optical signals have the same polarisation on entering a stage.

If vertically polarised the half entering the beamsplitter B from the prism P will be deflected to the next stage, the remainder being deflected away. If the optical signals are horizontally polarised those entering the beamsplitter B from the prism P will pass undeflected out of the switch, the remaining half passing through undeflected to the next stage. This arrangement can therefore select which of the eight inputs I is to be output to the output 0 by setting the polarisation rotators R accordingly to determine which half of the optical signals entering a given stage will be passed to the next stage and, eventually, the output 0.

Electro-optic switches other than twisted-nematic liquid crystals can be used as polarisation rotators, for example PLZT or iron αarnets in all the above e_r_boc_in.er.ts.

Referring now to Figure 5 there is shown a generalised optical connection network having eight inputs 20 formed by cascading a passive optical signal distributor 22 as described in the applicant's co-pending application GB 8721472 with eight optical space switches 2 according to the present invention arranged in parallel. The signal distributor 22 creates eight equal intensity but spatially separated copies of each of the spatially modulated inputs 20 which copies provide spatially modulated inputs to each optical space switch 2. Each optical space switch 2 therefore has eight inputs, each being a copy of a different one of the eight spatially modulated inputs 20. As each optical space switch 2 can select one of the eight copies to be the corresponding output, the combination

Q acts as an eight way N x N space switch having 8 setting. The total member of control signals and crosspoints in such a generalised optical connection network is the theoretical minimum of only N log ₂ N. The minimum loss however is 10 log, _Q N(dB) due to the passive input signal distributors.

A generalised permutation network could alternatively be formed by replacing the optical distributor with parallel optical space distributors of the type discussed in the above referenced paper by Kulke et al. Again, these can be extended to have N from 1 to a number greater than 8 to provide larger N, N x N optical connections. The distributor loss and 1st order cross-talk are eliminated but it will have only Nl settings and requires twice as many control signals and cross-points as the generalised optical switch. These could be made asynchronously self-routing since the control code line could be set optically.