GB2118304A - Detecting surface deviations - Google Patents

Abstract

Method and apparatus for detecting deviations in a spherical or substantially spherical surface 10 having an axis of symmetry, including an aspherical surface comprising fitting a circle 11 or part circle of a radius less than the radius of curvature of the surface to the surface and detecting any positional differences between the circle or part circle and the part of the surface adjacent the circle or part circle, and comparing the detected difference with a predetermined difference. In the case of an aspherical surface a predetermined difference may be calculated whereas in the case of a spherical surface the predetermined difference should be zero. In this way deviations and other faults in the surface can be detected. <IMAGE>

Description

SPECIFICATION Surface inspection method and apparatus The present invention relates to a surface inspection method and apparatus.

It has hitherto been difficult to rapidly and simply examine curved surfaces and difficulties are often experienced with flat surfaces. For example lenses such as spectacle lenses are still generally examined by eye using a method in which the lens is placed in front of a dark background and is illuminated from one side. Flat surfaces such as silicon wafers are also often examined by eye.

Such an examination has many disadvantages. It is time consuming, each examination of a lens taking of the order of 17 seconds; it is subjective, lenses which would, in fact, be acceptable often being rejected since the operator is unable to determine the degree of severity of the fault; in practice most damage to lenses and like components occurs by manual handling of the lenses and thus examination of the lenses is in itself a source of damage.

The present invention provides a method and apparatus for examining spherical or flat surfaces and can also be used with surfaces which are not exactly spherical but are close to spherical, for example, aspherical lenses or surfaces which have different radii of curvature in two transverse directions, such as spectacle lenses.

It is known that a circle can be placed on a spherical or flat surface and simultaneously touch the spherical or flat surface at all points on its circumference. The radius of the circle has to be less than the radius of the surface. It will be understood that if there are surface deviations, for example, bumps or pits or tilt in the surface, then the circle will not contact the surface at or adjacent the deviation.

Aspherical surfaces can also be measured since although the circle will not lie flat on the surface, providing the circle is correctly aligned with regard to the surface, the gap between the circle and the surface can be calculated. If the gap is then measured and compared with the calculated gap, then the surface deviations or errors in the asphericity can be detected.

The present invention provides a method for detecting deviations in a spherical or substantially spherical surface having an axis of symmetry (by substantially spherical surface we include not only aspherical surfaces but also surfaces which have different radii of curvature in directions transverse to one another about the axis of symmetry, such as spectacle lenses) comprising passing radiation to the surface to define at the surface a circle or part circle of a radius less than the radius of curvature of the surface, receiving radiation influenced by the surface and from said received radiation detecting any positional differences between the circle or part circle and the part of the surface adjacent said circle or part circle and comparing said detected difference with a predetermined difference.

In order to fit the circle to the surface, the axis of the circle (which is the line perpendicular to the plane of the circle and passing through the centre) should pass through the axis of symmetry of the surface. It will be understood that the "predetermined difference" will be zero in the case of a spherical surface in which case any non-zero difference detected will comprise a deviation. Alternatively, in the case of, for example, an aspherical surface, the "predetermined difference" may be a calculated value. In this case the detected difference is compared with the calculated difference and a deviation will provide a difference between these two values.

It will be further understood that it is not necessary in all instances to examine all of the circle since if the part circle and the surface are moved relative to one another, the part circle may be sufficient to be scanned across the surface.

In one arrangement the surface has a radius of curvature of infinity. In this case the surface is flat. In an alternative arrangement the surface has a radius of curvature of a finite value.

The circle or part circle and the surface are preferably moved relative to one another so that the circle or part circle is scanned across the surface.

This is most conveniently carried out by rotating the surface about its axis of symmetry.

In this case it is desirable for the axis of symmetry of the surface to pass through the circle or part circle.

In the case of the use of a beam of radiation the incident beam of radiation is preferably provided by a beam which is coaxial with the axis of the circle or circle portion and is deflected therefrom by a rotating mirror to form an incident beam which rotates about the axis of the circle and scans, at said surface, along the circle.

In an alternative arrangement, the circle is provided by the periphery of a cone of radiation.

The method also preferably includes means for detecting scatter of radiation from the surface. In this case, preferably, radiation from the surface is displayed to provide an image of the surface, standard surface faults are also provided and the image of the standard surface faults superimposed on the image of the surface so that any faults on the surface can be compared with the standard faults.

The present invention also provides apparatus for detecting defects in a spherical or substantially spherical surface having an axis of symmetry comprising radiation producing means, means for mounting said surface, means for passing said radiation to the surface to define at the surface a circle or part circle of a radius less than the radius of curvature of the surface, means for receiving radiation influenced by the surface, and means for detecting any positional differences between the circle or part circle and the part of the surface adjacent said circle or part circle and means for comparing said detected difference with a predetermined difference.

Through this specification we refer to "light" and "optical" in similar language which implies the use of radiation of an optical wavelength. However it should be understood that the techniques described are applicable, with suitable adjustment to the components, to radiation of infrared and ultra violet wavelengths and the invention is to be understood as extending to radiation of wavelengths from infrared to ultra violet.

Preferred embodiments of the invention will now be described by way of example only and with reference to the accompanying drawings in which: FIGURE lisa diagram illustrating the principle of one form of the invention, FIGURE 2 is a further diagram illustrating a variation of the arrangement of FIGURE 1, FIGURE 3 illustrates the calculation of the predetermined difference in the case of an aspherical surface, FIGURE 4 illustrates the principle of using a beam of radiation for the purpose of the invention, FIGURE 5 illustrates diagrammatically a preferred arrangement of apparatus according to the invention for carrying out the method of the invention, FIGURE 6 shows a functional diagram of the calculation apparatus forming part of the apparatus of the invention, FIGURE 7 illustrates an alternative arrangement of part of the apparatus of FIGURES, FIGURES 8a and 8b show various possible configurations of the input and output radiation in the arrangement of Fig. 7.

FIGURE 9 shows means by which the output of the apparatus of FIGURE 7 can be divided into five channels, FIGURES 10 and 11 show alternative optical arrangements for viewing different surfaces, FIGURES 12 and 13 show the locus of the scan viewed along the X axis, FIGURE 14 shows an alternative optical arrangement of part of FIGURES, FIGURE 15 shows the arrangement of FIGURE 14 adapted for use with a concave rather than a convex surface, FIGURES 16, 17 and 18 show three different arrangements of photodetectors for use with FI GURE 5, FIGURES 19 and 20 show output signals showing different types of errors in the surface, FIGURE 21 is a diagrammatic side view of a second embodiment of surface inspection apparatus according to the invention, FIGURE 22 is a graph showing various signals produced in use of the apparatus of FIGURE 21, FIGURE 23 shows, in diagrammatic form, the apparatus of FIGURE 21 applied to a surface inspection technique and, FIGURE 24 is a diagrammatic side view, similar to FIGURE 23, of a third embodiment of surface inspection apparatus according to the invention.

The basic geometrical principle of the invention is illustrated with respectto FIGURE 1.

The axes X and Y are in the plane of the paper and the Z axis is at right angles to the paper. The surface 10 to be inspected is shown here as a section through a sphere with centre C and radius OC.

The chord OP is a diameter of a circle 11 of axis 12 and radius OA with centre A and fitting the surface 10 of the sphere so that it touches the surface 10 at all points around its circumference.

If OC = R (ie radius of curvature of surface 10) = PC, ACP = ACO = e and OA = r (ie radius of circle 11)then R = sin e If the surface 10 is convex. as shown in Fig. 1,the intersection of the axis 12 of the circle 11 with the axis 13 of the surface 10 (which passes through the circle 11 at O) is at C and is to the right of O.If the surface 10 is concave the intersection C is to the left, and if the radius of the surface 10 is infinity (ie it is a plane surface 10) axes 12 and 13 are parallel with C at infinity. 20 should be less than 1800 and would typically be, for example, less than 1200 giving a maximum practical value of O as 60". In this case, therefore, 0 normally lies in the range 0 to + 600. In most cases, however 0 will be less than 40348.

Thus by determining the closeness of fit of the circle 11 to the surface 10, any defects in the surface 10 can be determined. If it is necessary to check the whole of the surface or substantially the whole of the surface then this can be arranged by scanning the circle 11 across the surface 10 which is most simply arranged by rotating the surface 10 about the axis 13. In this way the whole of the cap of the surface of the sphere between 0 and the locus of P will be scanned by the circle 11 and defects in that part of the surface 10 can be detected. The radius rand angle 0 are selected so as to scan the area of the surface 10 under consideration. Of course, if the surface 10 only extends from 0 toO and if the circle 11 is fixed at radius r then some part of the circle 11 will extend beyond the surface 10 with some loss of the active scanning cycle.Furthermore, only part of the circle 11 need be used if only a limited area is to be scanned. It will be noted that the axis 13 passes through the circumference of the circle 11 if the circle 11 is to scan up to the point 0 and thereby cover the whole of the cap of the surface 10.

Since sin 0 = Tand must always be less that 1, it R follows that r < R. This means that when testing very small lenses, such as might be used in a microscope, the radius r may have to be less than 1 mm and this defines the lower limit of r. This fact influences the mechanical design of the scanner.

An aspherical surface 10A is illustrated in Figure 2.

An aspherical surface is characterised by the fact that separate elemental areas have different centres of curvature C. If the surface is a solid of revolution, all the centres of curvature will lie on the X axis and the solid may be regarded as made up of an assembly of disks whose centres lie on the X axis and of appropriate radius determined by the ordin ate of P which is the point at which the circular locus of radius r intersects the desired surface. The method depends on the fact that whereas a circular ring fits perfectly on a sphere, it will rock backwards and forwards when fitted to an aspheric. The amount of rock (ie the gap between the circle 11 and the surface 10) will determine the degree of aspher icity.

Fig 2 shows two points P1 and P2 on the aspherical surface 10A. The component of displacement of P in the X direction is brought about by a rotation Izr (see Fig 3) of the point P about its rotation axis 12 (AC). In one complete rotation P the point P traces the base circle 11, centre A, of a right circular cone with its apex at C and semi angle e shown in Fig 3.

Taking as before OA = rand dropping a perpendicular PB on to OA, we see that BA = r cos 2 and OB = r(1-cos ).

Therefore the coordinates of P (x, y, z) are x = r(1-cos ) sine y = r(1-cos ) cos e and z = r sin Thus in order to follow a desired elemental circular section of the aspherical surface 10, by rotation of P about the x axis, say at point P1, the scan axis 12 is tilted angle e about 0 so that the scan locus of the point P intersects P1 for angular position (Zil.

The value of m and e for a given r can be computed given the Cartesian equation of the aspheric surface required.

For optical applications of aspheric surfaces which are solids or revolution, the maximum departure from the nearest fitting sphere is usually very small, involving a change in O of less than t 1".

Thus by suitable analysis, one can determine the gap between the circle 11 and surface 10 for an aspheric surface which, in most practical applications, will not be very great.

The principle so far outlined may be used to examine the shape of surfaces. This can be done by determining the degree of fit of the circle 11 to the surface 10.

Figure 4 shows the principle in the use of an optical probe for examining the surface 10. An illuminated pin hole 16 is arranged at the focus of a collimator lens 17, the parallel incident beam 18 being reflected offthe surface 10 at P and the reflected beam 21 being collected by a further collimator lens 19 of larger diameter which focuses a spot image 20 of the pinhole 16 on to an optical image displacement measuring device 22 which comprises an array of photocells or an array of optical fibres, the outputs of which fibres are fed to an array of photocells. Tilt of the surface 10 at P (owing to either a deviation in the surface 10 or to the asphericity of the surface 10) is measured by means of the displacement of the spot image 20 across the image displacement measuring device 22.

The width of the incident beam 18 at P must be sufficiently small to prevent the aberration due to the oblique incidence on surface 10 causing excessive spread of the image 20 falling on device 22.

Some compensation for this can be brought about by appropriate design of collimator lens 17 at least for a certain range of radii of curvature of surface 10.

If geometrical displacement is required instead of surface tilt, the angle measured can be integrated with respect to the distance travelled from the known reference point such as the pole 0 of the surface 10.

The principle illustrated with respect to Figure 4 is utilised in the apparatus in Figure 5. In Figure 5 there is provided a laser 26 producing a laser beam 27, the beam 27 passing through a beam splitter 28, the portion of the beam at 29 being arranged to be coaxial with the axis 12. In the case where the surface 10 is a sphere the axis 12 passes through the centre of the sphere 10.

The beam portion 29 is incident on a tilted plane mirror 31 and is thereby reflected from the mirror 31 to be incident on the sphere 10 at the point P. The mirror 31 may allow a small proportion of the laser beam 28 from the laser to pass directly through the mirror. The mirror 31 is mounted in a suitable bearing illustrated diagrammatically at 32 so as to rotate about the axis 12. The bearing 32 is adjustable whereby, for surfaces 10 of different radii, it may be adjusted so that at all times the mirror 31 rotates about the axis 12 which passes through the centre of curvature of the surface under examination.A spherical mirror 33 is provided in such a position as to intercept the reflected beam 35 and reflect the beam back to the point P whence it retraces its path and is reflected by the beam splitter 28 to the image displacement measuring device 22, the output of the image displacement measuring device 22 being fed to a suitable computing apparatus 34 and a pictorial output display being provided by a video display 36.

Figure 6 shows in diagrammatic form some of the components of the computing apparatus 34, (which may be provided by a microprocessor), there being provided an input line 37 from the measuring device 22, a processing unit 38, a memory store 39, a first acceptance output line 41, a second rejection output line 42, and a third video output line 43.

In operation the mirror 31 is rotated about the bearing 32 and hence the incident beam 27 scans across the surface 10 to trace the circle 11. As is clear from above, if the surface 10 is a perfect sphere then the image displacement measuring device 22 will receive a spot 20 of light which will remain stationary as the mirror 31 is rotated and the point P is scanned around the circle 11. If however, there is any flaw in the perfect spherical surface 10 then the beam will be displaced and this displacement will cause displacement of the spot 20 on the device 22.

The severity of the deviation in the surface 10 will determine both the displacement of the spot 20 across the device 22 and also the period of time over which it is displaced and this displacement is fed to the computing apparatus 34 which compares the signal with parameters preset in the memory store 39. It thereby provides signals either to the acceptance output line 41 or rejection output line 42 so that the surface 10 is accepted or rejected depending upon the severity of the deviation. The position of the deviation may be indicated on the video display 36. In practice, of course, the sphere 10 is mounted in suitable bearings 44 to rotate about its axis 13 and as it is rotated a complete area of the cap of the sphere is examined for deviations.

The tilt of the scan mirror 31 with respecttothe beam portion 29 is (y-0)/2. When testing a flat surface 0 = 0 and the mirror angle is Ywhich could 2 be typically 221/2 when the beam 27 is incident at 450 to the surface 10. The maximum value of O is likely to be 45" and taking the minimum value of tilt of the mirror 31 to be 10 , then y = 65". This large angle of incidence will ensure that light refracted and trans mitted by the surface 10 will be lost from the system.

Strictly speaking, the radius of curvature of mirror 33 should be equal to twice the distance from B (the point at which beam portion 29 strikes mirror 31) to device 22, so that the parallel beam would focus on device 22, but this would mean that mirror 33 could not be truly spherical. As a compromise, the centre of curvature of mirror 33 can lie on the axis BC but as the beam 27 is narrow the spread of laser radiation will not be too significant. The part of mirror 33 which is reflecting the beam can be thought of as a flat mirror of variable tilt as mirror 33 is moved along the axis BC. In an alternative arrangement, mirror 33 could be a small spherical mirror fixed to and rotating about axis 12 with mirror 31 with its focal point at device 22.

Figure 12 shows the view along the axis 12 from the mirror31 of Figure 5. As mirror 31 rotates through an angle 8 the surface 10 slowly rotates about the X axis. The circle 11 from this view will only be a true circle if R = infinity, that is if the surface 10 has a flat surface. In this case, P has coordinates z = r sin 0 andy = r1 - cos ).

For convenience, if instead of rotating the surface 10 about the X axis we rotate the circle 11 about 0 the coordinates of A, that is the centre of circle 11 are z = rcos x, y = r sin as shown in Figure 13.

The equation of the circle 11 is Z2 +y2 -2rfrcos-+ysinoc)=O If the whole surface 10 can be covered with N scans of the beam and the angular frequency of the rotation of the surface 10 is w, then the angular frequency of P is Nw. Atypical value of N might be 200.

Using the apparatus of Figure 5, the mirror 31 might typically be rotated at 400 Hz, and one complete surface could then be scanned in half a second and for an inspection spot of diameter 1 mm, an isolated defect such as a fine scratch would produce a pulse of duration 20 Ps.

As mentioned earlier, the mirror 31 allows some light to pass through along the axis 12, and this can be used to aiign the axis 12 with the centre of curvature C of the surface 13. Thus if light reflected by the mirror 31 is blanked out, the light passing through mirror 31 being viewed, the apparatus is arranged so that the incident and reflected beam along the axis 12 coincide. The axis 12 is then accurately aligned with the centre C.

Different arrangements of image displacement measuring device 22 may be used. Figure 16 shows a first arrangement in which there is provided a rectangular array of photocells 51. The normal position of the beam spot 20 is shown. The position of the beam spot 20 is determined by the particuiar photocell 51 on which it is incident.

In an alternative arrangement shown in Figure 17 there is provided a quadrant type detector in which four quadrant photocells 53 are provided. The cross section of the beam spot 20 will, unless there is deflection, provide an equal output on each of the quadrant photocells 53, but any movement away from the centre will produce changes in voltages on the photocells proportional to the deviations or tilt of the surface at the point of inspection. In the arrange ment of Figure 18 the quadrant photocell of Figure 17 is provided in the centre and is surrounded by an annular photocell 54.In this arrangement, the quadrant photocells 53 provide the information regarding the position of the beam 20 and if there are other defects on the surface 10 such as those which would produce scattering of the light (such as scratches) then the cross section of the beam 20 will expand momentarily into a bar of light extending beyond the central quadrant photocells 53 and provide an output on the photocell 54. Thus the arrangement of Figure 18 detects not only deviations in the surface 10 but also deviations which produce scatter. Error signals due to scatter will usually have time durations at least 100 times less than errors due to surface deviations.

Figure 10 shows the passage of the beam 29 when testing a flat surface 10 (such as a mirror or semi-conductor wafer) in which case the centre of curvature of the mirror 33 is at B1. Other parts of the apparatus are as arranged in Figure 5.

Figure 11 shows an arrangement of the components of Figure 5 to examine a concave spherical surface 10.

Other arrangements of the apparatus of Figure 5 are possible. For example the mirror 31 can be replaced by a rotating prism to refract the beam from the laser or two acousto-optical beam deflectors in tandem and with their deflection pianes at right angles.

Afurther alternative of part of the apparatus of Figure 5 is shown in Figure 7. In this case, after the beam has been reflected from the surface 10 it passes to a mirror 56 in place of mirror 33, the mirror 56 reflecting the beam directly back to the mirror 31, the beam from the mirror 56 passing parallel to the part of the beam incident on the surface 10, and the beam from the mirror 56 being reflected by the mirror 31 to an optical light guide 57, the optical light guide 57 comprising a bundle of optical fibres.The configuration of the fibres at the input to the optical light guide 57 is shown in Figure 8a and includes a central fibre 62 to take the unscattered laser beam and fibres distributed into four quadrants 58, 59, 60, 61 so that displacement in two directions at right angles to the laser beam can be detected as well as the proportion of light scattered out of the direct beam. Figure 8b shows the output face where the light guides from the four quadrants 58, 59, 60 and 61 have been separated into four bundles and the central light guide 62. In this way the outputs from these five sets of light guides can be separated on to five separate photocells in a manner shown schematically in Figure 9. The beam 27 from the laser 26 is reflected by a mirror 64 along the axis 12 and the light returning from the surface 10 is divided as above described in to the four bundles 58, 59, 60, 61 and 62.

The light guide 57 is connected to a hollow shaft 66 which rotates about the axis 12.

The light beams from the five bundles 58 to 62 are collimated by five lenses 67 and deviated by the axicon 68 so as to fall on to corresponding photon detectors 71 to 75. The electrical connections to these photon detectors are made by sandwiching them between two glass plates, each of which has a transparent electrically conducting layer. The effect of reflection of light by these layers is minimised by the use of an appropriate index matching optical cement which is electrically insulating.

In the above described arrangements, the circle 11 has been provided by scanning a light beam about the axis 12. This rotation can be avoided by the arrangement shown in Figures 14 and 15 when only surface deviations such as scratches are to be observed. In Figure 14 the laser 26 provides a beam which is reflected along the axis 12 by a small mirror 71 and it is brought to a focus by a lens 75 before diverging as a solid cone of light to illuminate the whole of the solid circle OP of the surface 10. The rays near the edge of this cone are reflected back by mirror 33 and the light scattered by defects is collected by lens 72 to form an image of the surface 10 onto a TV camara 73. The rays near the axis 12 are eliminated by the use of a small circular stop 74 placed at the centre of the lens 75 or alternatively by the use of a small hole at the centre of the mirror 71.

A small pinhole 76 is placed at the focus of lens 75 and antireflection coatings on the lens will reduce the level of unwanted reflected light. Thus a complete circle 11 of light is produced by this arrangement and rotation of the surface 10 about the axis OC will enable a scan of the cap of the surface.

Figure 15 shows an arrangement somewhat similarto Figure 14 but for use with a concave surface 10.

The TV camera 73 will produce an output showing all the surface defects such as scratches within the annular beam determined by the mirror 33 whether on the surface 10 oron mirror 33. As the surface 10 is rotated about the X axis the defects on mirror 33 will become immediately apparent as they will remain stationary and therefore can be ignored.

The size of the stop 76 must be sufficiently large to prevent all directly reflected light from the component under test from reaching the TV camera 73. The outside diameter of 76 is sufficiently large to prevent deviations of the reflected beam, due to errors in shape or tilt of 10 from reaching 13. When using the TV system as a means for quantifying surface defects, the central unused portion of the TV screen can be employed to display a pattern of standard defects for comparison purposes. A graticule of slits of varying width, back illuminated buy a portion of light from the laser can be projected by a lens from the reverse side of the mirror 71 on to camera 73.

The lines on the graticule can be calibrated in terms of standard scratches used in the working plane of the surface 10. In this way the instrument can be calibrated in terms of standard specifications such as might be employed in different countries orto satisfy the needs of particular instrument makers.

For automatic non-visual use of the instrument this comparison can be undertaken on the video signal produced byTV.

If the surface 10 is one of the surfaces of a lens both surfaces can be inspected at the same time by using similar systems on either side of the lens under test which must, of course, be mounted on a hollow spindle.

The output signal of one of the quadrants 53 of Figure 17 is shown in Figure 19. The trace A is the output signal for a good spherical surface and the trace B is the output signal for a trace with surface deviations. In Figure 20 there is shown a trace of the output signal from photocell 54 and various errors are illustrated therein.

A second embodiment of optical sensor apparatus suitable for simultaneous inspection and gauging of surfaces is shown in Figure 21 by way of example.

In Figure 21 a light source 101 is imaged onto a pinhole 104 by a lens 102, the light passing through a polarizer 117 and iris diaphragm 103 arranged between the lens 102 and pinhole 104. The iris diaphragm 103 restricts the area of spherical or slightly aspherical surface 108 which is illuminated for inspection and measurement. If the surface 108 is of short radius of curvature, only an area sufficiently small to be considered flat is illuminated. The light transmitted by pinhole 104, which is at the focus of a lens 106 is divided by a polarizing beam splitter 105 into two beams 123 and 121, the beam 123 passing to the surface 108 via lens 106. The ratio of intensity of the two beams 121 and 123 can be controlled by rotating the polarizer 117.The transmitted beam 123 is collimated by the lens 106 and passes through a quarter-wave retardation plate 107 before striking the surface 108 at normal incidence and the reflected beam passes through the quarter-wave retardation plate 107 again. At this point, when the beam returns to the polarizing beam splitter 105 its poiarization is changed by 1 8or and is reflected by the polarizing beam splitter 105 to form beam 122 and is brought to a spot focus 124 in the plane of a lens 112. The lens 112 is used to form an image of 108 on the image plane of the TV camera 116.

A focusing beam splitter 120 reflects a fraction of the incident beam 122 onto a concave reflective surface 113 and back down onto a flat reflective surface 115 and thereafter from the diagonal surface 1 to focus an image of the spot 124 onto TV camera 116. Thus the TV camera effectively receives two sets of images, one a direct image of the part of the surface 108 under inspection which will show the surface defects, and secondly a central spot whose position is determined by the tilt of the surface 108.

The first of these images, that is the image of the part of the surface under inspection is formed by the lenses 106 and 112 whilst the second, spot, image is formed by lenses 106 and 112 and the focusing surface 113 of the focusing beam splitter 120.

In addition to the position of the central spot indicating the tilt of the surface 108, the spread of the central spot is influenced by the scattering produced by the surface 108.

The portion of the beam from the light source 101 which is reflected at the polarizing beam splitter 105 and which forms the beam 121 is used to illuminate a standard or reference surface 111 which may carry a reference defect and/or measuring graticule. Before striking the surface 111, the beam 121 passes through lens 109 and quarter-wave retardation plate 110 and a beam from the surface 111 passes through the plate 110 and lens 109 again before returning to the beam splitter 105 where it passes directly through the beam splitter 105. Thus an image of the surface 111 is brought to a focus on the image plane of the TV ca mera 116 by means of the lenses 109 and 112. Thus in addition to the two images already referred to above, there is a third image relating to the reference defect.

To measure the severity of a particular defect on the surface 108 it is desired to vary the intensity of the image of the defect under consideration with respect to the intensity of the image of the reference defect. This can be simply done by varying the relative intensities of the beams 121 and 123. This can be simply effected by rotation of the polarizer 117 which thereby alters the angle of polarization of the incident beam 126 and therefore varys the proportion of that beam which is reflected or transmitted by the polarizing beam splitter 105.As the polarizer 117 is rotated the intensity of the image of the reference defect wiii ir,crease and the intensity of the image of the defect under examination will decrease and vice versa. "/ith very iittle experience it is a simple matter for the two images to be arranged to be of equal intensity. The angle of the polarizer can then be used as a non-subjective measure of the severity of the defect under examination.

We refer now to Figure 22. Figure 22 shows for each of three directions of the polarizer 117 (0 , 55", 90") the distribution of light intensity across, in the case of column (a) a defect from the surface 108 under test, in the case of column (b) the reference defect, and in the case of column (c) the combined intensity signal. The intensity of the defect on the surface under test has approximately twice the severity of the reference defect in this example.As the intensity in one beam varies according to the law of Malus as Cos2 O, the sum of both beams, the second of which is polarized at right angles, is constant but the visibility of the reference defect and defect under test can be varied continuously and the value of e where balance is achieved is a measure of the optical severity of the test defect compared with the standard.

According to Figure 22 if the peak intensity of the test defect (E) = 0 ) is A and the intensity at the point of maximum intensity of the defect under test is A (1 - a) and the peak intensity of the reference defect is B (O = 90") with the intensity at the point of minimum intensity of the reference defect as B (1 p), then the intensities in the combined beams again at the maximum intensity points is 11 = A (1 - a) cos2 e + B Sin2 e (i) i2 = A cos2 (3 + B (1- 3) Sin2 () If now e is adjusted to make 17 = 12, then from equations (i) and (ii) it may be shown that tango = Aoc/Bss (iii) If now either quarter-wave retardation plates 107 or 110 are rotated so as to make A = B, the value of a defect a can be calculated directly from the value of a reference defect ss. Thus if the test defect is twice as severe as the reference, i.e. or/ss = 2, then e = 55 or 10 higher than it would be if the test and reference defect were of equal severity when of course e = 45". The highest sensitivity of balance occurs when oc < css or the reference defect is more severe than the test defect.The practical operating range of defect severity measurement is about 75 to 1 for 13 in the range 20t to 70 .

It will be realised that this method can be used to quantify a whole range of surface defects in comparison with selected standard defects simply by adjusting e so that the two images appear to be of equal severity or contrast when viewed superimposed.

Instead of viewing the TV screen, automatic image analysis can be carried out on the video signal or a mechanical scanner and single photocell can be used to produce a time varying signal of the distribution of intensity across the image of the defects, for example, as described later.

In the apparatus thus far described the light source 101 comprises a tungsten source. However 3 laser could be used with suitable amendment of the optics. Also the surface 115 could be repiaced by a position-sensitive photocell to measure the tilt of 108 by means of a separate channel. If the tilt of 108 is measured at a number of adjacent points across the surface, the values obtained may be integrated to obtain the geometrical shape of the surface. Alternatively a laser source could be used with a reference flat between 106 and 107 so as to form a fringe pattern at 160. Traditional fringe counting methods could be used to record changes in the distance between 106 and 108.

If the surface defect produces only a phase change, such as a step in the level of the surface, it will be hardly visible at TV camera 160. By the use of a phase contrast filter placed over the pinhole image 120, such phase changes can be converted into amplitude changes as in the phase contrast microscope. Other standard techniques of illumination can also be employed, such as dark ground iliumination, by the use of an absorbing disk at pinhole 120.

The apparatus described with reference to Figures 21 and 22 can be used to detect errors in spherical or near spherical surfaces as follows. The beam 123 beyond the plate 107 can be scanned across the spherical or near spherical surface using the apparatus of Figure 23.

Thus in Figure 23 the apparatus of Figure 21 is illustrated in diagrammatic form at the top of the drawing and the beam 123 below the plate 107 is passed to a mirror 118.

In the following description we refer to the apparatus of Figure 21 as an optical head 200.

The principal axis 224 of the optical head 200, which coincides with the axis of beam 123, is defined by the centre line joining pinhole 104andthe optical axis of lens 106 and is directed towards the centre of curvature C of the surface under test 108. Mirrors 218,219,220 and 221 are arranged so that the beam 123 when projected at the surface 108 (shown as beam 224) is always approximately normal to surface 108. The mirrors 221 and 220 are fixed so as to rotate about an axis 222 which also passes through C. Rotation of mirrors 221 and 220 about axis 222 enables a circular path with centre 223 to be scanned over the surface of 108. At the same time the mirrors 218, 219,220 and 221 are slowly rotated about axis 224 so that after one rotation about the axis 224 the whole surface of 108 will have been inspected and measured for surface form if required. By coupling angular pick-offs to axes 222 and 224, the co-ordinate position of the area under inspection can be calculated. In this way, a record of the severity and position of all defects on the surface 108 can be produced.

The advantage of using a periscopic mirror system for scanning the beam over the surface 108 is that the angular deviation of the beams are independent of tilts of the mirrors, providing they are rigidly clamped together. The separation and tilts of the mirrors have to be adjusted depending on the radius of curvature of the component undertest and its diameter. When surface 108 is flat axis 224 is parallel to axis 222.

The TV camera 116 which is useful for slow visual inspection over the whole surface 108 can be replaced by a pinhole and photomultiplierto produce a time varying signal of reflectivity over surface 108 as the system is rotated about axes 222 and 224.

if a pinhole of diameter equal to the spot image is placed at lens 112, relatively slow changes in mean intensity level due to the tilts of the surface 108 can be recorded as the spot moves away from the pinhole, whilst rapid changes in output will be due to surface defects. The use of electronic filters tuned to different frequency bands can be used to separate the errors due to surface shape and those due to surface damage or surface texture.

Figure 24 shows a side view of an alternative arrangement similar to Figure 23. Similar parts are numbered similarly and we shall restrict ourselves to description of the differences between the two Figures.

The reference scratches on the surface 111 are arranged horizontal which is more convenient, and lens 106 is arranged to image the pinhole 104 onto the surface 108 at O. The advantage of this is that the surface 108 does not expand the image. The reflection of the image at O is imaged onto a pinhole 150 where it is relayed back to the TV camera 116 by a lens 151. The television camera 116 will pick up defects in the surface, for example, scratches and the like. As described with respect to Figure 21, these defects can be compared with the images of the reference scratches on surface 111 which are also transmitted by the beam splitter 105 to the TV camera 116. They are compared in the same way as described with respect to Figures 21 and 23. Tilt of the surface 108 from a spherical surface results in movement of the beam 122 away from the axis of the lens 151.It will be seen from Figure 24 that the beam 122 passes through a mirror 154 which comprises a mirror with a central non-reflective hole. The beam 122 is circular in section and the dimensions of the hole are chosen so that the beam, when on the axis 122, passes through the hole (the hole is, of course, eccentric because the mirror 154 is at an angle to the normal to the beam 122). If the beam is displaced from the axis of lens 117 then some of the beam will strike the reflective part of the mirror 154 and will be reflected to a diode 156. In this way, tilt of the surface 108 can be detected.

As the apparatus is rotated the beam 122 will move eccentrically if the part of the surface under examination is tilted and the direction in which the beam 122 is deflected will depend upon the direction of tilt. This can be detected by comparing the phase of the signals received by diode 156 with respect to the rotation of mirror 219 and 220.

In all cases described, surfaces of different radii can be dealt with by moving the optical components so that the axis 12 passes through the centre of curvature of the surface.

The techniques described here reduce the cost of inspection and measurement of a wide variety of surfaces. The principal features are: rapid inspection, no-contact, area information provided, surface defects can be quantified in terms of their optical severity, surface shape can be measured, a wide variety of surfaces can be inspected including optically polished surfaces, diamond machined surfaces, painted or plated surfaces and the surface can be virtually of any radius of curvature and of any material. The technology can be applied in the following areas: optics, electronics, engineering and surface properties research.

Savings result from the greater degree of automation made possible by replacing the traditional slow process of visual and stylus inspection by fast operating, computer controlled inspection using the scanners and sensors described.

Claims (33)

1. A method for detecting deviations in a spherical or substantially spherical surface having an axis of symmetry comprising passing radiation to the surface to define at the surface a circle or part circle of a radius less than the radius of curvature of the surface, receiving radiation influenced by the surface and from said received radiation detecting any positional differences between the circle or part circle and the part of the surface adjacent said circle or part circle and comparing said detected difference with a predetermined difference.

2. A method as claimed in claim 1 in which the surface is non-spherical in which case the predetermined difference is a calculated value.

3. A method as claimed in claim 1 or 2 in which the surface has a radius of curvature of infinity.

4. A method as claimed in claim 1 or 2 in which the surface has a radius of curvature of a finite value.

5. A method as claimed in any of claims 1 to 4 in which the circle or part circle and the surface are moved relative to one another so that the circle or part circle is scanned across the surface.

6. A method as claimed in claim 5 in which the surface is rotated about its axis of symmetry.

7. A method as claimed in any of claims 1 to 5 in which the axis of symmetry of the surface passes through the circle or part circle.

8. A method as claimed in any of claims 1 to 7 in which the circle or part circle is defined by the path mapped out by a beam of radiation.

9. A method as claimed in claim 8 in which the beam of radiation incident on the surface is provided by a beam which is coaxial with the axis of the circle or circle portion and is deflected therefrom by a rotating plane mirror to form an incident beam which rotates about the axis of the circle and scans, at said surface, along the circle.

10. A method as claimed in claim 8 or 9 in which the beam is a parallel beam.

11. A method as claimed in claim 8 or 9 in which the beam is focussed onto said surface.

12. A method as claimed in any of claims 1 to 7 in which the circle or part circle is defined by a circle of radiation.

'3. A method as claimed in claim 12 in which the circle of radiation is provided by the periphery of a hollow cone of radiation.

14. A method as claimed in any of claims 8 to 13 in which the beam or circle of radiation is reflected from the surface and collected and changes in the reflected radiation are used to determine the said positional differences between the circle or part circle and the adjacent part of the surface.

15. A method as claimed in claim 14 in which the changes are positional changes of the reflected radiation.

16. A method as claimed in claim 14 or 15 in which the changes are positional changes of the reflected radiation.

17. A method as claimed in any of claims 8 to 16 including detecting scatter of radiation from the surface.

18. A method as claimed in claim 17 in which radiation from the surface is displayed to provide an image of the surface, a standard surface fault is also provided and the image of a standard surface fault superimposed on the image of the surface so that any faults on the surface can be compared with a standard fault.

19. A method as claimed in claim 18 in which one or both of the image of the surface and the image of the standard fault is or are passed through a or two polarisers, the polariser(s) being adjustable so as to vary the intensity of one image with respect to the other, whereby the polariser is adjusted so that the two images are of the same intensity to provide a measure of the degree of severity of the fault on the surface.

20. Methods for detecting defects in a spherical or substantially spherical surface having an axis of symmetry substantially as hereinbefore described.

21. Apparatus for detecting defects in a spherical or substantially spherical surface having an axis of symmetry comprising radiation producing means, means for mounting said surface, means for passing said radiation to the surface to define at the surface a circle or part circle of a radius less than the radius of curvature of the surface, means for receiving radiation influenced by the surface, and means for detecting any positional differences between the circle or part circle and the part of the surface adjacent said circle or part circle and means for comparing said detected difference with a predetermined difference.

22. Apparatus as claimed in claim 21 in which means is provided to move the surface and the circle or part circle with respect to one another so that the circle or part circle is scanned across the surface.

23. Apparatus as claimed in claim 22 in which means is provided to rotate the surface about its axis of symmetry.

24. Apparatus as claimed in any of claims 21 to 23 in which the means for passing said radiation to the surface comprises an optical system to provide a beam of radiation.

25. Apparatus as claimed in claim 24 in which said optical system includes means to provide a beam which is coaxial with the axis of the circle or circle portion, said optical system including a rotating plane mirror to deflect the beam from the axis of the circle or circle portion to form an incident beam which rotates about the axis of the circle and scans, at said surface along the circle.

26. Apparatus as claimed in claim 24 or 25 in which the optical means includes means to provide a parallel beam.

27. Apparatus as claimed in claim 24 or 25 in which the optical means includes means for focussing the beam onto the surface.

28. Apparatus as claimed in claim 22 or 23 in which optical means is provided to define said circle or part circle by means of a continuous circle of radiation.

29. Apparatus as claimed in any of claims 21 to 28 in which means is provided to collect radiation reflected from the surface to determine said positional differences between the circle or part circle and the adjacent part of the surface.

30. Apparatus as claimed in any of claims 22 to 29 in which means is provided to detect scatter of radiation from the surface.

31. Apparatus as claimed in claim 30 in which display means is provided to display an image of the surface, a standard fault is provided and the display means is arranged to superimpose an image of the standard fault and the image of the surface so that any faults on the surface can be compared with the standard fault.

32. Apparatus as claimed in claim 31 in which radiation from the surface and radiation from the standard fault are passed through one or respective polarisers, the polariser(s) being adjustable so as to vary the intensity of one image with respect to the other, whereby the polariser may be adjusted so that the two images are of the same intensity to provide a measure of the degree of severity of the fault on the surface.

33. Apparatus as claimed in claim 21 substantially as hereinbefore described with reference to Figures 1 to 6, and 12, 13, 19,20 or as modified with reference to Figures 7 to 9, or Figures 10 and 11, or Figures 14 and 15, or Figure 16, or Figure 17, or Figure 18, or Figure 21,22 and 23, or Figure 24 of the accompanying drawings.