CH620291A5 - Method for optical calibration and means for implementing the method - Google Patents

Description

The present invention relates to an optical calibration method and to a means for implementing the method.

The most obvious use of the present invention is in dimensional control at 100% of a part in a production line. It applies best but not exclusively to the inspection of small metal parts, the tolerances of which are on the order of a micron and the dimensions of a few millimeters to centimeters. The invention lends itself as well to the measurement of absolute dimensions as to the comparison of a dimension with respect to a reference.

A known optical control method calls for the appearance of Moiré figures by the superimposition of light fringes projected onto the measurement object and fringes of a mask obtained from a reference piece. Another known optical control method is described in patent CH No. 585 874. Here the difference with respect to a reference dimension is obtained by determining the quantity of light passing through a mask and coming from the measurement object, l object being lit through a grid.

These methods suffer from the disadvantages that they require the projection of fine fringes of very high contrast and that the mask must be very precise.

The control is carried out on a surface element, or even on the entire surface of the measurement object. Therefore, the signal to noise ratio is low when the error to be detected is small relative to the surface element scanned. Moiré fringes are generally difficult to interpret quantitatively, and the detection of light having passed through a mask provides only an “average” value of the defects present. In addition, both methods are sensitive to the surface condition of the measurement part.

The object of the present invention is to remedy these drawbacks and to propose an optical control method which, as with the traditional method using mechanical probes, performs measurements at selected points on the surface. It is characterized in that a set of regular fringes in uniform linear translation is projected onto an object to be measured; in that one observes by means of an optic, in a different direction and not perpendicular to the direction of projection, the image of the illuminated object; in that at least two points of an image plane, one of which is the image of a reference point, are detected, the light signal reflected or diffused from the image of the projected fringes; and in that the phase shift between the signal at the reference image point and the signal at any other measurement image point is determined.

The measurement method in accordance with the invention is only slightly sensitive to the surface condition of the object checked, provided that the surface defects are small compared to the pitch of the projected fringes. The measurement of dimensions is only made at the chosen points of the controlled object, which makes it possible to obtain a much more significant signal-to-noise ratio. Other advantages inherent in the calibration method according to the invention relate to the fact that a phase shift measurement is carried out. Thus, the requirements on the precision of a mask and on the sharpness of the projected fringes are not so high, while maintaining a very high measurement accuracy. Indeed, the phase measurements are easily made with a relative error close to 10 ~ 3. Other advantages of the method will be explained in the course of the description which follows.

The appended figures illustrate by way of example the calibration method according to the invention and the means necessary for its implementation.

Figure 1 is a schematic view of a first measurement variant in accordance with the method.

Figure 2 illustrates a second variant implementation of the method.

In Figure 1, the light fringes projected on an object are denoted by (F). The fringes (F) have a regular pitch (in) and move linearly at a constant speed (vx). The fringes (F) are projected onto a three-dimensional object (3D) characterized by two curves of the Cartesian axis system (Oxyz). Point (B) is a point on the surface of the object (3D).

An optic shown schematically by a lens (L) collects a

10

1S

20

25

30

35

40

45

50

55

60

"5

3

620291

part of the light rays scattered or reflected by the object (3D) and makes the image of it on the plane (M). The image of point (B) is therefore (B7) - The image of a point (Bo), chosen as the reference point, is (Bo '). By way of simplification, it will be assumed that the magnification ratio of the optics is unitary.

At least two point detectors are arranged on the plane (M), or as shown in FIG. 1, optical fibers (Co) and (Ci) connect points on the image plane (M) to the detectors (Do) and (Di ). A phase detector (<î>) measures the phase of the signal between the point (Bo) and another chosen point (B ').

The calibration process in accordance with the invention takes place as follows. The phase difference apo between the signals at points (Bo ') and (B') is determined for a masterpiece. The phase shift between (Bo ') and (B') is the same ipo for any other room identical to the master piece, but it presents another value (i | j) for any room whose elevation according to Oz in (B) compared to (Bo) is different than for the centerpiece. The value of the phase shift (ip-ipo) is proportional to the deviation Az relative to the centerpiece.

The operation according to Figure 1 is as follows. The fringes (F) parallel to the plane (Oyz) are projected with a step (po) in a direction parallel to the plane (Oxz) at an angle (a) with (Oz) on an object (3D). The tangent to the surface of the object (3D) along a plane parallel to (Oxz) forms an angle ô with a normal to (Oyz). Thus the image of the fringes (F) will have a step (p) at the image point (B '), the step (p) being given by the relation:

p = p0 cos ô / cos (a-ô) (1)

and this on the assumption of unit magnification. A variation zz of the part in (B) with respect to the masterpiece then causes a displacement Ax of the fringe in M given by:

A X = Azsina-w / in (2)

A detector placed at the image point (B ') sees the fringes in uniform translation of speed vx and gives a signal of the form:

S (B ') = A (B') cos {tot + (ijj-apo)} (3)

where the pulse co = 2it vX / p0,

the phase shift (i |> —ijto) = 2ji A X / p and A (B ') = amplitude of the signal in (B').

The phase shift ipo of the masterpiece can be arbitrarily chosen from value O by a simple adjustment of the phase detector.

The phase shift oj) is proportional to the displacement Ax of the fringes on the image plane, Ax being itself proportional to the variation in dimension Az of the measurement part. The phase shift ■ »p is then well proportional to the deviation Az.

From relation (2) above, we see that the projection angle a must be different from O or jt, otherwise Ax is always zero. In addition, a is preferably also different from jr / 2 without which the control of plane parts is impossible (in relation (1), p is indeterminate for a = jt / 2 and to = O).

By choosing the reference point (Bo) on the workpiece support, the calibration process in accordance with the invention allows absolute measurements to be made. On the contrary, if the reference point (Bo) is on the measurement piece (3D) itself, the calibration process makes it possible to detect relative elevation errors between two points (B) and (Bo) of the piece (3D).

The optical calibration process, as described in principle so far, is very sensitive to the location of point detectors Do, Di ... placed on the image plane (M). A very small displacement in a direction perpendicular to the fringes, of the order of 1/1000 of the fringe pitch, is sufficient to introduce a measurement error. This severe requirement for detector positioning accuracy is incompatible with an industrial environment. A means of making the process largely insensitive to the positioning accuracy of the detectors is described below. A mask is provided that merges with the image plane, the mask itself being the image of the fringes projected onto the centerpiece. The mask can be either drawn or photographed and will present transparent zones alternating with opaque zones, the zones preferably being clearly defined. The detectors, placed behind the transparent zones, no longer punctual, but at least as wide as the transparent zones, will provide an electrical signal corresponding to the average of the light signal over the entire width of the transparent zone. Thus the detectors can be positioned with a certain tolerance, limited in that the detector collects the light signal over the entire width of the transparent area.

Another advantage of the mask is the fact of permanently recording the phase shifts (tyo) of the centerpiece. This avoids the measurement of the reference phase shift i |> o of any other point (B) whose dimension is to be checked on the series parts. The positioning of the detectors is thus greatly facilitated.

There are several ways to produce fringes scrolling at a constant vx speed. A first method consists in scrolling a grid in front of any light source. A second is to obtain interference fringes from two sources of coherent light. The scrolling of the fringes can be done by an offset A co of their respective pulsations. Different methods are described in the journal Applied Optics, volume 2, page 839 (1963), volume 8, page 538 (1969), volume 9, page 649 (1970) or in IEEE Journal of Quantium Electronics QE - 7, page 450 ( 1971) for example. Interference fringes have the advantage of being able to be projected over a greater depth of field than the fringes obtained with incoherent light. The optical calibration process according to the invention easily lends itself to industrial automation and makes it possible to control with apidity, without creating any delay in the production chain, the 100% of the parts produced. The control can be carried out on one or more critical points of a part with respect to the reference point. The requirements for precision lateral positioning of the part for inspection depend on the fineness of the details machined on the part. It is sufficient that the detectors always identify the same element of the surface of the room. The accuracy of depth positioning is not critical for relative measurements.

The optical calibration process of the invention not only makes it possible to carry out part elevation checks in a production chain but also to establish the profile of any reduced part. This case is illustrated in Figure 2. The object whose profile is to be determined is located on a plane reference surface, the two objects being scanned by the same set of parallel fringes. Two detectors in the image plane, parallel to the fringes, follow the scanning of the fringes of the planar surface. The phase shift between the two detectors provides a measurement of the profile. Instead of providing two mobile detectors, it is possible to envisage a succession of fixed detectors.

The optical calibration process in accordance with the invention is sufficiently precise for industrial application in micro-mechanics. Indeed, the depth resolution is given by: Rz = e • po / sin a.

s

10

15

20

25

30

35

40

45

50

55

60

65

620,291

4

With a precise phase detector at e = 10 ~ 3, with po = 10 lines / mm and with lighting at an angle a of 15 °, an axial resolution of 0.4 microns is obtained. The lateral resolution, or the minimum distance between two measurement points is given by the distance between fringes which is 100 microns in this example.

The optical calibration method according to the invention finds its limit of application in the depth of field of measurement. The maximum difference in phase shift allowed at a point is between - jt and + tz.

The maximum elevation error, or the maximum allowable elevation variation is therefore

Az max = p0 / sin a

In the example above, Az max s 0.39 mm, which is quite sufficient for most micro-mechanical applications. In the application of Figure 2, the depth of the measurement field is extended with a greater distance between fringes.

10

B

1 sheet of drawings

Claims (10)

620,291 2 Optical calibration process, characterized in that a set of fringes (F) are projected onto a uniform object in uniform linear translation on an object to be measured (3D); in that one observes by means of an optic, in a direction (a) different and not perpendicular to the direction of projection, the image of the illuminated object; in that at least two points of an image plane (M) are detected, one of which is the image (Bo ') of a reference point (Bo), the reflected or scattered light signal of the image of projected fringes; and in that the phase shift (ij) —tyo) is determined between the signal at the reference image point (Bo ') and the signal at any other measurement image point (B'). 2. Optical calibration method according to claim 1, characterized in that the light signal of the image of the projected fringes is detected by means of detectors fixed on the image plane. 3. Optical calibration method according to claim 1, characterized in that one detects by means of detectors movable on the image plane and rigidly linked together, which follow the uniform movement of a reference fringe, the light signal of the image of the projected fringes. 4. Optical calibration method according to claims 2 and 3, characterized in that one brings, by means of optical fibers, the light signal of the image of the fringes projected from the image plane to the detectors. 5. Calibration method according to claim 2, characterized in that one detects by means of detectors placed behind transparent areas of a mask coincident with the image plane, the mask being the image of the fringes projected onto an object of reference, and in that the detectors provide an electrical signal corresponding to the average of the light signal over the entire width of the transparent zone. 6. Means for implementing the optical calibration method of claim 1, characterized in that it comprises a device for projecting regular light fringes (F) in uniform rectilinear translation onto a measurement object (3D), a optical system (L) which projects the image of the object traversed by the light fringes on an image plane (M), at least two detectors (Do, Di) of light signals, one of which is the reference detector, on the image plane (M) and which detect the scrolling of the image of the light fringes and at least one phase detector (0) which measures the phase shift between the signals emitted by the detectors (Do, Di). 7. Means according to claim 6 for the implementation of the optical calibration method of claim 5, characterized in that it further comprises a mask in the image plane (M) consisting of the static image of a centerpiece lit by the fringes and having transparent zones alternating with opaque zones. 8. Means according to claim 6, characterized in that the device for projecting regular fringes is constituted by a light source preceded by a grid traveling at constant speed. 9. Means according to claim 6, characterized in that the device for projecting regular fringes is constituted by the superposition of two beams of coherent light. 10. Application of the optical calibration method of claim 1 for the control of critical dimensions of parts in a production chain.