GB2189039A - Optical signal processor - Google Patents

Abstract

Light is passed through an array of cells (4a, b...) containing a non-linear medium. The light is phase- shifted by an amount depending on its intensity, so on combination with a non-phase-shifted beam (7a, b ...), interference occurs. The interference pattern intensity from each cell is detected (11a, b...) and is compared (13a, b...) with a threshold to distinguish destructive from constructive interference and thereby determine the degree of phase shift. For analogue-to-digital conversion, the light is intensity modulated by an analogue signal. A binary 0 or 1 is output according to whether destructive or constructive interference occurs. Successive cells increase in length by a factor of 2 so the instantaneous output from the array constitutes a digital code of the instantaneous intensity. <IMAGE>

Description

SPECIFICATION Optical signal processor This invention relates to optical signal processing and to high speed analogue to digital (A/D) conversion using optical components.

For accurate digital representation of an analogue signal, the Nyquist criterion requires that the sampling rate must be at least twice the highest frequency component of interest in the incoming analogue signal. This criterion severaly limits the range of conventional A/D converters because of their low speed of operation.

Electro-optic A/D converters have been designed which, because of their optical components, can operate at much higher conversion rates. These converters are based on interferometry in electro-optic crystals.

This invention uses similar interferometric methods for deriving digital code, but avoids the use of electrodes and electro-optic crystals.

According to the invention an optical signal processor comprises a plurality of intensity dependent phase modulators all reponsive to the intensity of a coherent optical input signal, the intensity of the input to each phase modulator in conjunction with the modulation rate of each modulator being such as to produce a phase relation at the outputs of the modulator which is representative of the intensity of the input signal. Preferably the modulation rate increases binarily throughout the plurality of phase modulators and equal intensity signals are input to each modulator.

The phase of each modulator output may be determined by means for causing interference between the modulator output signal and a phase reference signal to produce an interference output and means responsive to the intensity of said interference output. The phase reference signal may be derived from the coherent optical input signal. The phase modulators may each comprise a cell of a non-linear medium.

The invention includes within its scope an analogue to digital converter comprising an optical signal processor as described above and wherein said optical input signal is modulated by an analogue signal and said phase relation is used to generate periodically a digital representation of the analogue signal, the output from each modulator corresponding to one bit of the digital representation. The phase modulators may be arranged in series to provide a serial digital output, or in parallel to provide a parallel digital output.

The phase reference signal may be of considerably lower optical power than the input to the phase modulators and be passed through a cell of non-linear medium identical to that through which said input passes in order to reduce errors caused by the effect of temperature variations on the non-linear medium.

An analogueldigital converter in accordance with the invention will now be described by way of example with reference to the accompanying drawings, of which: Figure 1 is a block diagram of an A/D converter incorporating an optical signal processor in accordance with the invention; Figure 2 shows the interference patterns produced at the outputs of the converter of Fig. 1; Figure 3 shows the output of a comparator of Fig. 1; Figure 4 shows a preferred input arrangement for use with the converter of Fig. 1; Figure 5 shows an alternative configuration of the A/D converter; Figure 6 shows a Fabry-Perot interferometer; and Figure 7 shows the output against input intensity curve for the interferometer of Fig. 6.

Referring to Fig. 1, a series of interferometers 1a. 1b, 1c, etc., each have an input cube beam splitter 15a, 15b etc. One of the resulting light paths 5a, 5b, etc., includes a cell 4a, 4b etc., of non-linear medium, a neutral density filter 23a, 23b etc., and a beam splitter 9a, 9b etc. the refractive index of a non-linear medium varies with light intensity so that pulses modulated to different intensities experience different refractive indices and hence travel with different phase velocities. Materials exhibiting non-linear behaviour include LiNbO3, InSb, CdHgTe and GaAs.

The other light bath 7a, 7b etc., includes a tap-off beam splitter 17a, 17b etc., a beam splitter 19a, 19b etc., and it then rejoins the first path 5a, 5b etc., at the re-combining beam splitter 9a, 9b etc. Each tap-off beamsplitter 17 provides one output beam for its own path 7 and one as the input to the succeeding interferometer input beam splitter 15.

A. successively reduced portion of the original input light is thus applied to the successive interferometers.

The first interferometer and its input beam splitter 15a are supplied from a mode-locked laser 3 by way of an intensity modulator 2 to which a modulating analogue signal is applied.

The output from each recombining beam splitter 9a, 9b etc., is applied to a respective photodiode 1 1a, 1 1b etc. which provides one input to a respective comparator 13a, 13b etc., constituted by an operational amplifier. The other, threshold, input to each comparator is provided in common by a single photodiode 21 which derives its input from the first beam splitter 19a.

Although we describe embodiments using bulk optical components, it is envisaged that the A/D converter could be constructed equally well in integrated optics.

In operation, the analogue signal to be digitised is applied to the intensity modulator 2 which receives pulsed light from laser 3. The laser is mode-locked to provide very short precise pulses. Thus the intensity of the light emitted by the laser varies with the amplitude of the analogue signal. Digitisation of the analogue signal does of course imply sampling it and the laser 3 is pulsed at the required sampling rate.

The pulsed coherent modulated light from laser 3 enters interferometer 1 a and is split by cube beam splitter 15a. This divides the beam into two components, one of which travels along path 5a and the other along path 7a.

Path 7a leads to tape-off beam splitter 17a, where the beam is again split into two components, one of which continues along path 7a and the other entering the second interferometer 1b via path 8a. Light on path 5a passes through the cell 4a of non-linear medium where it is phase shifted by an amount depending on its intensity. Path 7a is of constant refractive index and of equal length to path 5a. Thus when light on path 5a emerges from cell 4a it is phase shifted with respect to light on path 7a.

The beams are recombined at beam splitter 9a and they interfere constructively or destructively according to their relative phase. Before recombination, the beam on path 5a passes through an appropriately weighted neutral density filter 23a. This equalises the light intensities so that the interference pattern is symmetrical and not biassed towards the more intense beam.

The component of light travelling along path 8a from beam splitter 1 7a enters interferometer 1 b where it is split between the two paths Sb, 7b. Light which does not enter this interferometer travels onto the third and so on to the nth interferometer. At each interferometer, one light beam passes through cell 4 and is phase shifted, the other is not.

The lengths of cells 4 increase by a factor 2 at each successive interferometer. Thus cell 4b is twice the length of cell 4a, cell 4c is twice the length of cell 4b, and so on. This means that twice the phase difference is introduced by cell 4b as is introduced by cell 4a for the same intensity, and four times the phase difference is introduced by cell 4c as is introduced by cell 4a.

Fig. 2 shows the interference pattern produced when the pairs of beams are recombined at beam splitters 9a, 9b and 9c. The intensity varies from zero at perfect destructive interference to a maximum at perfect constructive interference, which maximum increases with increasing input intensity.

The intensity is measured by photodiodes 11. Light travelling on path 7a is split by beam splitter 19a, one component going to recombine with the beam on path 5a at beam splitter 9a, and the other component at output 20 being applied to photodiode 21. The intensity of this output 20 varies with the analogue signal amplitude, and provides a threshold intensity. As mentioned above the more intense beam of each pair passes through an appropriately weighted neutral density filter 23a, 23b etc., before interference takes place so each intensity output. can be compared to the single reference or threshold intensity 10 on Fig. 2, from beam splitter 19a. Beam splitters 19b, 19c etc., could be replaced with mirrors.

If the intensity at beam splitter 9 of the interference pattern exceeds the threshold value at 21 then constructive interference has taken place, and the associated comparator passes a corresponding signal, binary digit 1.

If the interference pattern intensity at 9 is less than the threshold value, destructive interference has taken place and the associated comparator transmits a corresponding signal, binary digit 0.

An alternative thresholding method is to arrange for all the non-phase shifted component intensities on 7a, 7b etc., to be read and compared with their respective interference pattern intensities at beam splitters 9a, 9b, etc. In this case, successive neutral density filters 23a, 23b etc., are independently weighted. All elements 19a, 19b etc., are in this case beam splitters and each comparator is linked to a different photodiode 21a, 21b, etc. Thus, in this embodiment of the light intensity is halved at each beam splitter 15a, 17a and 19a the light intensities approaching beam splitter 9a on paths 5a and 7a will be respectively at 50% and 1221% of the original intensity from modulator 2.To equalise the intensities before recombination therefore, filter 23a should reduce the intensity of the beam on path 5a by 75%.

An array of n cells 4 enables an n-bit digital code to be generated. The most significant bit is generated by the interferometer with the shortest cell (4a) and the least significant bit is generated by the interferometer with the longest cell. To count in binary, the "units column" (least significant bit) changes twice for each change in the "2's column", four times for each change in the "4's column", and so on. Thus, to provide the means for binary counting, the phase difference produced by the longest cell (least significant bit) is twice that produced by the next longest cell, and m times that produced by the mth cell.

Fig. 3 shows the output of one comparator corresponding to the interference pattern intensity at beam splitter 9a, say. The line 10 indicates the threshold intensity separating destructive interference (below line 10) from constructive interference (above line 10).

Fig. 4 shows a preferred arrangement of beam splitters illustrated here for 8-bit code.

Light from intensity modulator 2 is split by cube beam splitter 25 into two components.

Each component is split again at cube splitters 27 and again at cube splitters 29. The eight components from cube splitters 29, all of equal intensity, enter interferometers 1a, 1b...1h. Each component is then split by cube splitters 31, one beam of each resulting pair of beams going along path 5a, 5b etc., and the other along path 7a, 7b etc. in this arrangement, all beams enter the interferometers at the same intensity so neutral density filters 23 are not required.

A possible modification of the system is to provide asymmetic splitters at 15 in Fig. 1 or at 31 in Fig. 4 so that a higher proportion of light traverses paths 5 than paths 7. Since the phase shift produced at cell 4 is proportional to intensity, use of a higher intensity enables use of a shorter cell to produce the same phase shift. In this case, filters 23 would be differently weighted to equalise the components before recombination.

Fig. 5 shows the interferometers arranged in series. This allows more efficient use of the light, which travels sequentially from interferometer 1a to interferometer 1b and so on, as opposed to the almost parallel configuration of Fig. 1, or the parallel configuration of Fig. 4.

The components and operation are exactly as for Fig. 1 except that tap-off beam splitters 17a, 17b etc. are here situated in paths 5a, 5b etc. with emergent beams either travelling along connecting paths 8a, 8b etc. to the next interferometer ib, 1c etc., or continuing along paths 5a, 5b etc. to recombining cube beam splitters 9a, 9b etc.

Beam splitters 15a, 15b etc. and 17a, 17b etc. are highly asymmetric so that the maximum possible light passes through cells 4a, 4b etc. Only a low light level is required at the photodiodes. For example, of the light at 15a, only 1% might be directed along path 7a, whilst 99% continues through cell 4a, with a further 1% being tapped off at beam splitter 17a to enter recombining splitter 9a. Thus, high light intensity is maintained through the successive cells 4a, 4b etc. Again, a single threshold may be used as shown in Fig. 5 at photodiode 21, or a separate threshold may be taken for each interferometer.

The properties of non-linear media are dependent, to differing extents, on temperature.

Temperature instability could therefore lead to inaccurate digital conversion. This problem could be alleviated by introducing into paths 7a, 7b etc. a cell of non-linear medium similar to that in paths 5a, 5b etc. This would eliminate common mode refractive index variations in the cells caused by temperature variations, and provided the optical power in path 7a, 7b etc. is a sufficiently small fraction of the optical power in path 5a, 5b etc., the phase shift introduced by the extra "corrective" cell in path 7a, 7b etc. will be negligible in comparison with that introduced by cell 4a, 4b etc.

To make the system fully optical for use, for example, in an optical computer, electronic thresholding means may be replaced by optical devices, such as Fabry-Pérot cavities, as shown in Fig. 6. The cavity comprises a cell 25 of non-linear medium, and two partially reflecting mirrors 23 at each end of the cell 25. The mirrors 23 could simply be the end faces of the cell 25, partially silvered. Light is reflected back and forth inside the cavity, and constructive interference occurs only if the cavity length I is equal to an integral number of half-wavelengths of the light i.e.

nl I= 2 The output intensity therefore remains very low unless this wavelength condition is fulfilled. The non-linear medium effectively changes the wavelength of light according to its intensity, since phase velocity (and therefore also wavelength) decreases with increasing intensity. The length of the cell 25 can be chosen such that light above a certain intensity level is strongly transmitted, but light below this level is very weakly transmitted. The transmission curve is shown in Fig. 7. The line of rise is very steep, consequently the determination of whether the intensity is above or below the threshold level is more accurate.

For a cell 25 of fixed length, the threshold intensity is fixed, so a bias light input of variable intensity must be applied to compensate for the changing intensity of light at the output of the intensity modulator 2, which is of course reflected on the interference pattern intensity, as shown on Fig. 2. To hold constant the intensity level at the changeover from constructive to destructive interference, the input to the Fabry-Pérot interferometer is biassed in the opposite direction to the change of intensity. As the reference light level on output 20 rises, the bias intensity falls and vice versa.

This bias signal can be provided by a Fabry Pérot interferometer as in Fig. 6, but used in the reflective mode. Most of the light not transmitted through the cavity is reflected.

The reflection characteristic is thus the inverse of the transmission characteristic, falling with rising input intensity, as shown in Fig. 7.

Claims (10)

1. An optical signal processor comprising a plurality of intensity dependent phase modulators all responsive to the intensity of a coherent optical input signal, the intensity of the input to each phase modulator in conjunction with the modulation rate of each modulator being such as to produce a phase relation at the outputs of the modulators which is representative of the intensity of said input signal.

2. An optical signal processor according to Claim 1 wherein the modulation rate increases binarily throughout the plurality of phase modulators and equal intensity signals are input to each modulator.

3. An optical signal processor according to Claim 1 or 2 wherein the phase of each modulator output is determined by means for causing interference between the modulator output signal and a phase reference signal, to produce an interference output and means responsive to the intensity of said interference output.

4. An optical signal processor according to Claim 3 wherein said phase reference signal is derived from said coherent optical input signal.

5. An optical signal processor according to any preceding claim wherein each said phase modulator comprises a cell of a non-linear medium.

6. An analogue to digital converter comprising an optical signal processor according to any preceding claim wherein said optical input signal is modulated by an analogue signal and said phase relation is used to generate periodically a digital representation of said analogue signal, the output from each modulator corresponding to one bit of said digital representation.

7. An analogue to digital converter according to Claim 6 wherein said phase modulators are arranged in series to provide a serial digital output.

8. An analogue to digital converter according to Claim 6 wherein said phase modulators are arranged in parallel to provide a parallel digital output.

9. An analogue to digital converter according to Claim 6, 7 or 8 and Claim 5 wherein said phase reference signal is of considerably lower optical power than the input to said phase modulators and said phase reference signal is passed through a cell of non-linear medium identical to that through which said input passes in order to reduce errors caused by the effect of temperature variations on said non-linear medium.

10. An analogue to digital converter substantially as hereinbefore described with reference to the accompanying drawings.