This invention relates to an optical mixing device.

Optical devices for beam mixing are well known in the prior art. Beamsplitters are employed to mix two optical beams as described by A F Harvey in "Coherent Light" p1046, Wiley, London (1970). They may be used in free space or incorporated into waveguide systems.

Optical Y-junctions of various forms are known in the prior art for the production of mixed beams. Various passive symmetrical Y-junctions are discussed by Z Weissman, A Hardy and E Marom in "Mode-Dependent Radiation Loss in Y-Junctions and Directional Couplers", IEEE Journal of Quantum Electronics Vol. 25, No. 6 (1989) pp 1200-1208. Asymmetric Y-junctions are discussed by K Shirafuji and 5 Kurazono in "Transmission Characteristics of Optical Asymmetric Y Junction with a Gap Region", Journal of Lightwave Technology Vol 9, No 4 (1991) pp 426-429. Active Y- jinetions are also known, and examples are described by H Sasaki and I Anderson in "Theoretical and Experimental Studies on Active Y-junctions in Optical Waveguides:, IEEE Journal of Quantum Electronics Vol. QE-14, No. 11 (1978) pp 883-892. Each of these references discusses in detail the use of Y-junctions for beam splitting, but gives little detail on their use for beam combination or mixing. Indeed the symmetric Y-junctions, both active and passive, are inefficient splitters. Their transmission is heavily dependent on the angle of splitting; transmission is as low as 20% for splitting of a few degrees.

Prior art arrangements for mixing of more than two beams involve beamsplitter or Y-junction devices used in series. The losses of individual devices are therefore multiplied leading to very inefficient beam mixing.

If subsequent detection is required then the mixed beam, or beams, may be directed to appropriate detectors. However, in addition to their

inefficiency, prior art mixing devices based on Y-junctions also suffer from the disadvantage of only having one output port. This reduces the available information concerning the relative phases of the input beams. Therefore such prior art devices are of limited usefulness for applications such as heterodyne detection where comprehensive phase information is important.

It is an object of the invention to provide an alternative form of optical mixing device.

The present invention provides an optical mixing device for operation at a wavelength λ and including:

(a) a waveguide having an input region and an output region, and

(b) radiation supplying means arranged to provide two input radiation beams directed to the input region,

characterised in that

(A) the waveguide is a multimode waveguide,

(B) detecting means are arranged to receive radiation transmitted by the waveguide to the output region, and

(C) the dimensions of the waveguide, and the positions and spatial characteristics of the input radiation beams are in combination arranged to provide for modal dispersion in the waveguide giving rise to input radiation mixing in the output region and mixed radiation detection by the detecting means.

The present invention provides the advantage that two input beams may be efficiently mixed. Theoretically the invention might provide mixing with

100% efficiency. In practice, efficiencies of 75% have been achieved i non-optimised embodiments of the invention. The invention provides th additional advantage that phase information contained in input radiatio beams is not lost when the beams are mixed. This enables mixing to b carried out prior to detection.

The waveguide incorporated in devices of the invention may be o rectangular cross-section, of height 2a and width 2b, where b is greate than a. The detecting means may be located centrally in the outpu region. Additional detecting means may be included. These may be tw additional detecting means located within the output region distant b/2 t either side laterally of the centre of the output region, where b is a defined earlier. In such a device the waveguide may be of lengt L = 2b ² /λ, where λ is the wavelength of input radiation measured withi the waveguide.

Mixing devices of the invention may incorporate first and second detectin means. Each of these detecting means may be located within the outpu region and distant b/2 laterally from and on a respective side of th output region centre. The waveguide may be of length L equal to eithe 4b ² /λ or 8b ² /λ.

Mixing devices of the invention may be constructed with additional inpu radiation beams.

The waveguide may be formed as a hollow within solid dielectric material The dielectric material may be alumina. Alternatively the waveguide ma be formed as a ridge waveguide upstanding from a substrate. It may b formed of layers of a ternary or quaternary semiconductor material syste such as A£ _χ Ga< _| _ _χ As.

The two or more input radiation beams may be supplied by square cros section input waveguides arranged for operation in fundamental mode. Thi

provides spatial characteristics of the radiation electric field in the form of a half-cycle of a sine wave.

A mixing device of the invention arranged as a heterodyne mixer may be incorporated in an optical system which includes a coherent radiation source arranged to generate an output beam and a local oscillator beam signal, and means for collecting radiation reflected or scattered from a test region. Such a mixing device is arranged to mix the local oscillator beam and the collected radiation.

In order that the invention may be more fully understood, embodiments thereof will now be described, by way of example only, with reference to the accompanying drawings, in which:-

Figure 1 is a schematic sectional plan view of an optical device of the invention in the form of a mixer for use in heterodyne detection;

Figure 2 is a sectional view on line II-II in Figure 1 looking in the direction of the arrows;

Figure 3 graphically illustrates the variation of intensity coupling coefficients for rectangular waveguide EH^ modes with variation in the aspect ratio of the waveguide;

Figure 4 shows field amplitude distributions for various lower order rectangular waveguide modes;

Figures 5 and 6 illustrate variation in electric field intensity distribution as a function of position along multimode waveguides with aspect ratios of 3 and 6 respectively.

Figure 7 provides the phase variation along each of the intensity distributions in Figures 5 and 6;

Figures β and 9 illustrate variation in electric field intensit distribution as a function of position along a multimode waveguid for two input radiation beams which are respectively in phase and i antiphase with one another;

Figures 10 and 11 graphically illustrate relative modal amplitude of odd and even numbered waveguide modes excited in a rectangula waveguide by a fundamental mode input beam;

Figures 12 and 13 schematically illustrate devices of the inventio for use in heterodyne detection;

Figure 14 schematically illustrates a laser radar syste incorporating a mixing device of the invention;

Figure 15 schematically illustrates a further mixing device of the invention.

Referring to Figures 1 and 2, there are shown sectional views of a optical device of the invention in the form of a mixer indicated generally by 10. The mixer 10 incorporates a rectangular block 12 with a rectangular cross-section hole running through it to define a rectangular waveguide 14. The waveguide 14 has constant rectangular cross-section and reflecting walls 16a to 16d; it is of height 2a, width 2b and length L, these dimensions being respectively parallel to x, y and z Cartesian co¬ ordinate axes indicated by 18 and 20. Of these, x is referred to as vertical (perpendicular to the plane of Figure 1 ) and y and z as horizontal (in the plane of Figure 1), for ease of expression. The origin of the co-ordinate system is defined, for the purpose of this specification, to be such that dashed line A-A in Figure 1 indicates the plane z = ^• 0, and walls 16a to 16d lie in planes y = -b, x •= +a, y = +b and x = -a respectively. The waveguide 14 has an input region 22 in the plane z = 0, and an output region 24 in the plane z = L.

The parameters a, b and L are employed to preserve generality, specific values will be described later. However, in this example b > 2a.

The mixer 10 also incorporates two square cross-section input waveguides 26 and 28. The input waveguides 26, 28 have output apertures 30 and 32 arranged in the input region 22 of waveguide 14 such that their centres 30a and 32a are located at x = 0, y = -b/2, z = 0 and x = 0, y = +b/2, z = 0 respectively. A detector 34 is arranged in the output region 24 of the waveguide 14 such that its centre is located at x = 0, y = 0, z = L.

The waveguides 14, 26 and 28 are formed of alumina. The detector 34 is a mercury cadmium telluride detector with associated circuitry of known kind.

The operation of the mixer 10 will now be described in general terms; a more detailed theoretical analysis will be given later. The input waveguides 26 and 28 receive input radiation from a coherent source (not shown), and each carries a fundamental EH<u mode radiation beam. These radiation beams in the input waveguides 26 and 28 provide two fundamental EHιι mode inputs to the rectangular waveguide 14, in which a number of EH _j m _j modes are excited in consequence. These modes interact with each other as described in detail later. The effects produced by the interaction depend on the relative frequencies and phases of the input beams. Two input beams which are of like frequency and which are in phase with one another produce a single central maximum which is centred at the point x = 0, y = 0, z = L and which is incident on the detector 34. However, when two input beams of like frequency are in antiphase, the input field is regenerated with two maxima incident on respective sides of the detector 34.

If however the two input radiation beams differ slightly in wavelength, for example by virtue of a relative Doppler shift, the electric field at the output region 24 varies between a single central maximum and two laterally positioned maxima. This variation occurs at the beat frequency

of the two input radiation beams. The intensity of light incident on the detector 34 therefore varies at the beat frequency. The device 10 may therefore be used as a heterodyne mixer. For example, one of the input waveguides 28 may carry a received signal obtained from a target zone by reflection or scattering of radiation an output from a coherent source; the other input waveguide 26 may then carry a local oscillator signal obtained from a like source and employed in the device 10 for mixing with the received signal. Any beat frequency obtained by this mixing indicates Doppler frequency shift produced in the target zone from motion of reflectors and/or scatterers of the source radiation.

The effects of mode interaction within the device 10 are the result of a specific example of a more general phenomenon. They arise from the form of excitation of the rectangular waveguide 14 and the relationship between the waveguide length L, waveguide width 2b and radiation wavelength λ. In the device 10 the length L is given by

L - * (1)

where λ is the wavelength of the local oscillator radiation in the rectangular waveguide 14. The wavelength of a Doppler shifted received signal varies with time. As will be described later, changes in the waveguide length L and in the location and form of the input to it alter the form of the electric field at the output region.

The theoretical propagation characteristics of a rectangular waveguide will now be analysed. It is assumed that the waveguide has height 2a, width 2b and is bounded by a homogeneous dielectric material with complex dielectric constant ε. It is also assumed that this dielectric material (which provides the waveguide walls) is highly reflecting and not significantly attenuating for required propagating modes. The waveguide has height, width and length dimensions which are parallel to the x, y and z axes respectively. It has normalised linearly polarized modes of the kind EH _j m. The electric field contribution E _an (x,y,z) of the mn ^th mode

EH _jj m at the point (x,y,z) has been calculated by Laakaann et al in Appl. Opt. Vol. 15, No. 5, pages 1334-1322, May 1976 as follows:

cos /mπx\ cos /nπy

Emn ,y,z) (ab) sin [la ! sin \~2 )] 'mn- (2)

where

m is the mode number relating to the field dependency along the x axis,

n is the mode number relating to the field dependency along the y axis,

z is the distance along the z axis,

Yen ⁼ (Pmn ⁺ --vn- ' -~ ^e Propagation constant of the mn" ¹ mode, β^ and O _JJJJJ being the mn ^t " mode's phase and attenuation coefficients, and

"cos" above "sin" indicates the former applies to even mode numbers (m or n as appropriate) and the latter to odd mode numbers.

The phase coefficient β _™ is given by:-

2π

^■■ mn -ftsr λn ) B ),1f (3. 1

If the negative term in parenthesis in Equation (3.1) is small compared with unity (paraxial radiation approximation), which is satisfied in practice, then the binomial theorem may be used to rewrite Equation (3.1) as:-

2ιt

^J mn -iteMsr} (3.2 )

where a, b, m and n are as previously defined, and λ is the wavelength o the radiation propagating in the waveguide.

Equation (2) sets out the electric field contributions obtainable from al linearly polarized modes of a rectangular waveguide. It is calculated o the basis that the electric field contribution of each mode is zero at th side walls 16a and 16c of the waveguide, ie at y = +b and -b. This i satisfied if the waveguide 14 has reflecting side walls 16.

The first situation to be considered is that of a rectangular waveguide of side 2a by 2b excited by radiation propagating as a singl fundamental EH ₁₁ mode from a square section waveguide of side 2a connecte to one end of the rectangular waveguide and arranged coaxially therewith. c

The single EH^ mode is coupled to the various EH^ modes of th rectangular waveguide. That is it becomes decomposed into a linea combination of the EH^ modes with respective complex multiplicativ amplitude coupling coefficients A^. For the case of excitation of the rectangular waveguide modes EH^ by a square waveguide fundamental mode EH the coefficients ?i _mn are given by:-

Essentially the A^ amplitude coupling coefficients are the coefficients of a Fourier series which represents the field at the input region. The EHJJJJJ modes are mutually orthogonal, and in consequence the coefficients A _g n can be calculated from overlap integrals of the form:

+b +a Ann - f J BH?ι • ---mn • dy.dx. (5)

From Equations (2) to (5) it is possible to calculate how the intensity coefficients of the excited rectangular waveguide modes vary

as a function of b/a, the ratio of the widths of the rectangular and square waveguides. Figure 3 illustrates the variation of 1,^ with b/a; ie the effect of varying the waveguide aspect, or width to height, ratio. Figure 3 indicates that 1^ = 0 except when m = 1 and n is odd. This is because of the symmetric nature of the excitation conditions. Consequently, the modes excited are only the symmetric modes EHu, EH 3, EH 5 etc.

The forms of some of the lower order EH^ waveguide modes are shown as electric field amplitude distributions in Figure 4. These were obtained by computation, and are shown as graphs (a) to (f) in quasi-three dimensional form. The coordinate axes are shown at (g) . The axes x, y and z correspond to transverse vertical, transverse horizontal and longitudinal directions in the multimode waveguide as before. The graphs (a) to (f) correspond to modes as follows:-

(a) : EHu; (b) : EH ₁ ; (c) : EH31;

(d) : EH ₁₂ ; (e) : EH ₁₃ ; (f) : EH ₂₂ .

Of these, (a), (c) and (e) are symmetric modes and (b), (d) and (f) are antisymmetric modes. To clarify this, let E(x) and E(-x) respectively be the electrical field amplitude distributions associated respectively with positive and negative parts of the x axis in Figure 1; E(x = 0) is on the z axis. Let E(y) and E(-y) be the equivalents for the y axis.

For a symmetric mode:-

E(x) = E(-x) andE(y) = E(-y) (6.1)

For an antisymmetric mode, either one of or both of (6.2) and (6.3) below apply:-

E(x) = -E(-x) (6.2)

E(y) - -E( -y) ( 6.3

In the initial situation considered the symmetric input provides for onl symmetric modes of the multimode rectangular waveguide to be excited.

The transverse electric field distribution in an xy plane distant z fro the input region is E _z given by:-

E ₂ = £ Ann . EHnn (7

The field intensity distribution in xy planes distant z from the inpu region is |E _Z | ² , the square of the modulus or magnitude in Equation (7)

|E ₂ | ² has been computed as a function of distance z along the rectangula waveguide for two values of b/a. In both cases, the waveguide width (2b was 3 mm, and its height (2a) was 1 mm in one case and 0.5 mm in th other. This corresponds to b/a = 3 and b/a = 6, and the computatio results are given graphically in Figures 5 and 6 respectively. Figures and 6 give the field intensity I = |E _Z | ² as a function of position across the rectangular waveguide for each of a series of values of z alon this waveguide. In both cases the computation was based on a radiatio wavelength of 10.59 microns (C0 laser) and an active waveguide length of 425 mm given by Equation (1 ) .

As illustrated in Figure 3, when b/a = 3, only the modes EHu, ^EH 13' ^EH 1 and EH ₁₇ are excited, and these have approximate relative powers 0.32 0.33, 0.13 and 0.02 respectively. When b/a = 6, the modes EHu ^{t0 EH} 1 1 are excited with respective relative powers from 0.27 to 0.02.

In Figure 5, an initial central maximum 80 indicates the electric fiel intensity distribution I at the input region to the rectangular waveguide At this point (z = 0), the modes EHu ^{t0 EH} 17 ^{give rise to} electric field which are in phase with one another and interfere constructively t produce the maximum 80. Moving down the length of the rectangula

waveguide, ie as z increases, the modes EHu ^{fco ra} 17 ^{aove out of} phase with one another. This is a consequence of Equations (2) and (3), in which the phase coefficient ^{' '} β^ and therefore also the propagation constant γ^ are dependent on the mode numbers m and n.

The spatial rates of change of these modal electrical field contributions therefore differ along the z axis, ie axially of the rectangular waveguide. This changes the form of the interference between modal field contributions, and gives rise to a variety of electric field intensity distributions extending transversely. The distributions are indicated by curves such as 81 and 82 in xy planes at respective values of z. Approximately two thirds of the distance down the rectangular waveguide, the intensity distribution is given by a curve 83 having three similar maxima. A distance L along the rectangular waveguide, the intensity distribution is shown as a curve 84 having two well separated maxima 84a and 84b. The maxima 84a and 84b are located with their centres at the points x = 0, y = -b/2, z = L and x = 0, y = +b/2, z = L respectively. They are in phase with one another.

Turning now to Figure 6, this shows transverse electric field distributions along the length of the rectangular waveguide when its cross-sectional aspect ratio b/a is 6, as previously mentioned. As indicated in Figure 3, the effect of increasing b/a from 3 (as in Figure 5} to 6 (as in Figure 6) is to reduce power coupled to rectangular waveguide modes EHi1 and EH ₁ 3 and increase power coupled to modes EH ₁ 5 to EH ₁ 3. Since higher order modes receive more power, the degree of structure and definition in Figure 6 is increased over that in Figure 5. In Figure 6, the field distribution in the plane of the input region is indicated by a curve 90 with a single maximum 90a. As before, due to the modes EHu to EH ₁ 3 having differing β^ values, the transverse intensity distributions change with distance z along the rectangular waveguide. Curves 91 to 95 indicate locations at which there is field intensity division into multiple maxima of substantially equal form and magnitude. The curves 91, 92, 93, 94 and 95 have six, four, three, five and two

maxima respectively. Curve 93 in particular has three well defined maxim 93a, 93b and 93c. The curves 91 to 95 are located at distances from th waveguide input region of L/3, L/2, 2L/3, 4L/5 and L respectively, wher L is the waveguide length as has been said. These lengths can b expressed as 2L/6, 2L/4, 2L/3, 4L/5 and 2L/2. Accordingly, there is a inverse relationship between number of maxima and distance.

Figure 7 shows the variation along the y axis of the phase of th resultant electric field for the waveguide dimensions from which Figure was derived. Curves such as 100 to 105 are shown, which correspond t curves 90 to 95 respectively. Each of the phase curves such as 10 indicates the phase variation of the electric field across the rectangula waveguide for a respective value of z, and corresponds to a respectiv intensity distribution in Figure 6. The vertical scale of the phas representation φ is shown at 106, where an interval of 2ιτ is indicated The field distributions at 90 and 95 are of constant phase as indicated b straight lines 100 and 105. However, curve 103 for example has a centra region 103a which differs in phase to its two outer regions 103b and 103c The regions 103a to 103c give the phases of associated maxima 93a to 93 in Figure 6. In consequence, the central maximum 93a is out of phase wit the outer maxima 93b and 93c, which are in phase with one another. Sinc curves 100 and 105 are in phase, they produce reversible properties; i two in-phase inputs 95a and 95b would give rise inter alia to one outpu 90.

Figures 5, 6 and 7 relate to specific values of b/a. More generally, fo the situation initially considered, only EH _1n modes are excited because o the EHι symmetry of the excitation from the input radiation beam. At th rectangular waveguide input region, the phase is constant. For the cas involving arbitrary b/a values, using Equation (3) the phase coefficien βlp of mode EH _p is given by:-

3lι (8

and the phase coefficient β* _] q of mode EH _1q is given by:-

2π

Plq T J afl] (9)

By subtraction of Equation (9) from Equation (8) and rearranging, the phase difference between modes EH _1p and Hι„ at guide length z is χ _z given by:-

Xz - βl - βlq * ϋ Ξ . (p ² - q ² ) ( 10)

16.b ²

If the condition is imposed that a 2π phase difference exist between the modes, Equation (10) becomes:-

χ _z - l ± . (p ² -q ² ) - ITS. (11)

16.b ²

and the propagation distance z (say z _2π ) in Equation (11) in rectangular waveguide that gives rise to a 2π phase difference between modes EHι_ and EH „ is given by:-

z ₂ π ' '~ ⁽ 12 ⁾

(p ² -q ² ).λ

For the case of the EHu ~~- ^EH 1n ^ao -~- (--- ~- ^e fundamental and n ^th highest order odd mode) z _2π is given by

z _2π - ² -- ² (13)

(n ² - l).λ

Combining Equations (2) and (13):-

-2κ - I - ⁽ 14 ⁾

(n ² - 1)

With n = 3,5,7,9,11 16L/n ² - 1) z _2π is 2L, 2L/3, L/3, L/5, 2L/15 ...

As fractions of a propagation distance 2L in rectangular waveguide which results in the EHu and EH ₁₃ modes being back in phase, the relative length ratios are 1, 1/3, 1/6, 1/10, 1/15 etc. This shows that there is

a harmonic relationship between the EH _1n modes of the rectangular guide Equation (4) shows that the propagation distance z _n which gives rise t a 2π phase shift between the fundamental EH ₁₁ mode and the next highest order EH ₁₃ mode also gives rise to a 2τt phase shift between th fundamental and all other EH _1n modes (n odd) . This results i reproduction of any symmetric input electric field after a distance z _2π provided that there is only excitation of odd EH _1n modes. A symmetri input field is also produced periodically at distances of tz _2π , where "t is an integer number if there is sufficient length of rectangula waveguide available.

Equations (11) to (14) may be rewritten to determine z _π , the propagatio distance in rectangular waveguide over which an intermode phase change o π is introduced. By inspection of these equations, it is seen that:-

1

- -2n 8L/(n ² - 1) (15

L and 2L are the waveguide lengths over which z _π and z _n are introduced and L = 2b ² /λ from Equation (1). In consequence, z _n and z _2n are bot proportional to b ² , and may be arranged to occur at prearranged distance along a rectangular waveguide by suitable choice of the waveguide width

Returning to the mixer 10, of Figures 1 and 2, the appropriate mod structure within the waveguide 14 is the reverse of that illustrated i Figures 5 and 6. That is there are two fundamental EHu ^mo ~- inputs, length L of rectangular waveguide, and a single centrally positione detector or output means. However, as previously mentioned, when ther are two or more input radiation beams the relative phases of the input are important, and must be selected appropriately for the desired output

Referring now to Figures 8 and 9, the variation of electric fiel intensity distribution I with distance along a multimode waveguide i illustrated. These drawings relate to two equal intensity inputs in-phas and in anti-phase respectively. In Figure 8, two initial maxima 110 an

112 indicate the electric field intensity distribution I at z = L. They are positioned on the y-axis at -b/2 and +b/2 respectively.

The relative phases of the maxima 110 and 112 are indicated at 114. As for beam splitting, modal dispersion occurs in the waveguide and after a length L (ie at z - L) a single maxima 116 is produced, positioned on the y-axis at y = 0. In Figure 9, two initial maxima 120 and 122 indicate the electric field intensity distribution I at z = 0. They are positioned at y = -b/2 and y = +b/2 respectively. The relative phases of the maxima 120 and 122 are indicated at 124. Again modal dispersion occurs in the waveguide, but after a length L (ie at z = L) two maxima 126 and 128 are produced. They are positioned on the y-axis at -b/2 and +b/2 respectively. Thus the input electric field intensity distribution has been reproduced after a length L of waveguide. For phase conditions intermediate the two extremes illustrated in Figures 8 and 9, three output maxima will be produced at z = L. The respective amplitudes will be dependent on the relative phases of the inputs.

The output effects produced by the relative phases of a plurality of inputs to the rectangular waveguide are a result of the modes excited.

This is discussed below with reference to Figure 10 and 11 which graphically illustrate relative modal amplitudes for the three lowest odd and even EH^ modes respectively, as an input waveguide bearing a fundamental mode beam is offset from the centre of the rectangular waveguide input region. In the device 10 the two inputs are located within the input region 22 at y = ± b/2. As can be seen from Figure 10 the odd modes, EHu, ^EH 13 ^an - ^EH 15 -~- excited with identical amplitudes by inputs in these two positions. However, as can be seen from Figure 11 the even modes, Hι ₂ , EH ₁₄ and EHig are excited with amplitudes of identical magnitude but opposite sign by two such inputs. Therefore when the two inputs are in phase with each other the excitations of the odd modes sum to produce twice the amplitude of a single input at -b/2 or +b/2 whilst the even modes cancel each other out. As was shown earlier, excitation of only odd modes (n = 1,3,5 etc) leads to the two inputs

summing to form a single maximum at the output region 24. When the two inputs are in antiphase with each other the odd modes cancel out and the even modes sum to produce twice the amplitude of a single input at -b/2 or +b/2. Again, as was shown earlier, this input condition produces two maxima at the output region 24.

The mixer 10 may be designed for operation with radiation from a C0 ₂ laser of wavelength 10.59 ym. Its dimensions may be 2a = 0.6 mm, 2b = 1.2 mm and L = 2b ² /λ - 106 mm. The setting of b = 2a indicates the minimum width and hence length for which the mixer 10 may be constructed. However, with b = 2a the three output maxima produced for most input phase conditions are not fully resolved. The inner tails of the laterally positioned maxima overlap with the tails of the central maximum. Thus a detector 34 of width 2a will never receive zero intensity, but the intensity will vary with the beat frequency. The detector 34 may be narrower than 2a if required. Alternatively the dimensions of the mixer 10 may be 2a = 0.6 mm, 2b = 1.8 mm and L = 2b ² /λ = 153 mm. In this case b = 3a and the three output maxima will overlap less and therefore be better resolved. However to achieve full resolution of the three maxima a rectangular waveguide with b = 4a is necessary, and this requires L = 2b ² /λ = 272 mm if 2a = 0.6 mm.

The waveguides 14, 26, 28 of the device 10 may be constructed, for use at 10.59 μm of materials other than alumina, eg BeO, Si, Macor or metal. Furthermore, the square waveguides 26 and 28 may be replaced by other forms of waveguide. For instance, a device of the invention may incorporate square section waveguides of side 2a with sides at 45° to the x and y axes, or elliptical guides with the major and minor axes parallel to the x and y axes. However the square of the waveguide depth should be an integral multiple of the product of multimode waveguide length and wavelength in that waveguide. Other shapes such as diamond or octagonal guides may also be used. In addition optical fibres may be used.

The mixer 10 may have various additional features. It may be advantageous for the remaining area of the output region 24, not occupied by the detector 34, to be made from an absorbent material, or bear an antireflective coating. This would prevent radiation reflecting back into the rectangular waveguide 4 and interfering with the desired mode structure.

Referring now to Figure 12 an alternative embodiment of the invention is illustrated schematically. It consists of a mixer 130. The mixer 130 operates in a very similar manner to the mixer 10 and therefore the following description concentrates on areas of difference. Parts common to the mixer 10 of Figure 1 are like referenced with the addition of an asterisk superscript (*) . The essential difference between mixers 0 and 130 is the presence of two additional detectors 132 and 134, and associated circuitry (not shown). The two additional detectors 132, 134 are linked together such that their output signals are combined.

The additional detectors 132, 134 are located in the z = L plane with their centres at x = 0, y = ± b/2. Thus in operation as a heterodyne mixer a central output maximum when produced would be incident on the detector 34* whilst lateral output maxima would be incident on the additional detectors 132, 134. As the intensity incident on the central detector 34* falls, that incident on the lateral detectors 132, 134 rises. The mixer 130 does not obtain any additional information over that obtained by the mixer 10 but, depending on operating conditions, may provide an improved signal to noise ratio.

As discussed above the resolution of the three output maxima depends on the width of the rectangular waveguide 14*. If it is desired that the maxima are fully resolved and the detectors 34*, 132, 134 are 2a wide then b must be set at 4a and the length will be 32a ² /λ. However if the length of the mixer 130 is required to be shorter the detectors 34*, 132, 134 may be narrower. For instance if b = 5a, and the detectors 34*, 132 and 134

are 1.2a wide then the mixer 130 will be L = 12.5a ² /λ long and ha reasonable performance.

Referring now to Figure 13, a further two input mixer of the invention illustrated schematically. The mixer is indicated generally by 170. incorporates a rectangular waveguide 172 of width 2b, height 2a and leng L - βb-/λ. The width 2b must in general be equal to or greater than 4 and in this case 2b = 4a. The waveguide 172 has input and output regio 174, 176 respectively at its mutually opposite longitudinal ends. As f previously described embodiments Cartesian co-ordinates will be employ to describe positions within the device 170, the axes and origin a similarly defined. Two input square waveguides 178, 180 are connected the input region 174 such that input radiation beams provided by them a centred at x = 0, y = -b/2 and x = 0, y = +b/2 respectively. T detectors 182, 184 are located in the output region 176 centred at x = 0 y = -b/2 and x = 0, y = +b/2.

The two input square waveguides 178, 180 each provide a fundamental mo input. As described earlier, these may be a local oscillator signal a a Doppler shifted received signal respectively. When the two inputs a either in phase or in antiphase with one another, two substantially equ intensity maxima are produced in the output region 176, centred on x = 0 y = ±b/2. When the received signal in waveguide 180 is 90° ahead of t local oscillator signal in waveguide 178 then a single maximum is produc in the output region centred on x = 0, y - +b/2. When the received sign is 270° ahead of the local oscillator signal then a single maximum centr on x = 0, y = -b/2 is produced. Thus, the intensity incident on each the detectors 182, 184 varies at the beat or difference frequency.

Referring to Table 1 the relative dimensions of the various embodimen described, of two input mixers of the invention, are summarised. It c be seen from Table 1 that a compromise has to be reached between t length of the device and the resolution of the output maxima. In ma applications the length will be the more important criterion. There m

however be manufacturing processes which make relatively straightforward the production of devices each having a rectangular waveguide the same width as the sum of the widths of associated input waveguides.

Table 1

Relative Dimensions for Various Embodiments of Two Input Mixers of the nvention

A further alternative to the embodiments described above will now be described. It incorporates a multimode waveguide of width 2b greater than 4a, but with two input waveguides positioned at y = ±(b - a) . That is the input waveguides are positioned adjacent to the sides of the input region of the multimode waveguide, with a gap therebetween. This device structure will lead to some distortion of the electric field structure described earlier, in particular of the central output maximum. However it will provide for zero intensity at the centre of the output region when the two inputs are in phase. The distortion may thus be acceptable if the result is improved discrimination at the detectors. Suitable relative dimensions may be 2b = 4.5a and L = 10.125a ² /λ.

Embodiments of the invention may be constructed with alternatives to detectors located in the rectangular waveguide output region. For

instance output waveguides may be positioned to accept the maxima in place of the detectors, with detectors located within the output waveguides remote from the output region.

The invention is not limited to mixers for use in heterodyne detection. There are many applications where it is desired to combine radiation beams prior to detection for which embodiments of the invention may be used. In addition the invention is not limited to mixing devices with two inputs.

The invention is also not limited to use of hollow core optical waveguides for use of wavelengths of 10.59 μm. Provided the devices are constructed from appropriate materials for the wavelength of operation, they may be constructed for use over a wide range of wavelengths. For instance they may be constructed, using semi-conductor layer technology, from Gaλs and AδGaAs for use with radiation from Nd-YAG lasers (free space wavelength λ = 1.064 μm).

Referring now to Figure 14, a laser radar system incorporating a device of the invention, in the form of a heterodyne mixer, is illustrated schematically. The system is indicated generally by 200. It incorporates a laser radiation source 202, a beamsplitter 204, an acousto-optic modulator 205 and a mixing device of the invention 206 similar to the device 10 of Figure 1. It also incorporates optical waveguides 208, 210, 212, 214 and 215 directing radiation between optical components and to and from a target zone (not shown) . The mixing device 206 incorporates a detector 216 and associated circuitry 218.

The system 200 operates as follows. Radiation emitted by the laser 202 passes down the waveguide 208 to the beamsplitter 204. It is split into first and second beams. The first beam passes down the waveguide 210 and becomes an output beam to the target zone. The second beam passes along the waveguide 212 to the acousto-optic modulator 205 where it undergoes a small frequency shift. The frequency shifted beam passes to the mixing device 206 and is employed as a local oscillator signal. Radiation

reflected or scattered from the scene is received by the waveguide 214, and passes to the mixing device 206. It forms a received signal including Doppler shifted components reflected from moving objects or particles. The Doppler shift may be an increase or decrease in frequency.

The mixing device 206 operates as previously described for the device 10. Radiation passing to the mixing device 206 along waveguides 212 and 214 is mixed and the intensity incident on the detector 216 varies with the difference frequency of the local oscillator signal and the received signal. An output electrical signal is provided, by the associated circuitry 218, at 220. The purpose of the acousto-optic modulator 205 is to avoid the loss of decreased frequency components after subtraction of the local oscillator frequency, and which would otherwise correspond to negative frequencies; ie a source frequency to which becomes downshifted by a Doppler frequency shift f _D becomes a frequency f _Q -f _D . Subtraction of the local oscillator frequency (also f ₀ ) apparently produces -f _D , which does not exist. To avoid this, subtracting a modulation frequency f _A from the source frequency produces a local oscillator frequency f ₀ -fχ. A negative Doppler shift then gives a frequency f __-- _! which can be arranged to remain positive.

Referring now to Figure 15, there is shown a further embodiment of a mixing device of the invention indicated generally by 230. It incorporates two square section fundamental mode input waveguides 232 of side 2a, these being connected to a rectangular multimode waveguide 234 of cross-section 2a by 2b and length 4b ² /λ. As before, λ is the wavelength of radiation in the rectangular waveguide 234 for which the device 230 is designed. The rectangular waveguide 234 has two detectors 236 with associated output lines 238.

The input waveguides 232 have central axes 240 offset by b/2 on respective sides of a rectangular waveguide central longitudinal axis 242. The detectors 238 are centred on respective input waveguide axes 240.

The device 230 operates as follows. Radiation beams (not shown) are inpu to the input waveguides 232 and propagate therein in fundamental mode. O reaching the rectangular waveguide 234 they undergo modal dispersion. I the two contributions of radiation reaching the rectangular waveguide ar in phase with one another, the detectors 236 receive respective radiatio intensity maxima of like magnitude centred on axes 240. If however on input contribution leads the other in phase by 90°, then that detector 23 which is aligned with the input waveguide 232 providing that contributio receives 85% of the radiation intensity reaching the right hand end of th rectangular waveguide 234 and the other detector receives 15%. Variatio in the input phase as a function of time produces a variation in eac detector signal between 15% and 85% of the sum of the signals. Unlike th device 170 of Figure 13, the theoretical modulation depth is only 70% instead of 100%, but against this the device 230 is more compact becaus it has a rectangular waveguide 234 half the length of the equivalent 172 for the device 170.