NL1010894C2 - Mechanical detection system for use in co-ordinate measuring system, uses suspended probe elements to make very accurate measurements of geometric properties of workpiece - Google Patents

Abstract

The measuring probes in this system are electrical strain gauges, which are part of a resistive bridge circuit. Each probe housing is suspended in an intermediate frame assembly (5). The intermediate frame assembly is connected to the main frame by three elastic suspension elements (3).

Description

- 1 -

Mechanical touch probe equipped with a measuring system integrated in the suspension for use in coordinate measuring machines, suitable for determining geometric properties of workpieces with high precision.

5

Background of the invention

Mechanical touch probe systems are frequently used in the geometric control of all kinds of products that must meet certain requirements with regard to their dimensions. Mechanical touch probes serve as an interface between the coordinate measuring machine or machine tool on the one hand and the workpiece on the other. A coordinate measuring machine or machine tool has a number of rulers that are attached to the different axes of movement of the machine (usually X, Y, and Z) which, as it were, span a reference coordinate system to which the geometry of a workpiece can be related by means of mechanical affect. This attack is often done by means of a touch probe which is attached to the last link of the machine Js and which is often able to detect the surface of a workpiece with high accuracy. Upon detection 20, the touch probe generates a signal that instructs the machine to read out its rulers and calculate the touch point, that is, the contact point between probe and workpiece, based on this data. The workpiece is often attacked by means of an extremely accurately known ball (hereinafter referred to as the probe tip), but this is not a requirement. Other geometric elements are also used, provided they can be described mathematically accurately and can be accurately calibrated. The probe tip is connected by means of a probe to a detection mechanism located in a so-called probe housing. This probe housing is then connected to the machine.

30

Primary problems that arise in the development of mechanical probes are the following: 1 Π i S »O? - -2- 1. How can the spatial position of the probe tip be accurately determined; 2. how can the contact between probe tip and workpiece be determined; 3. How can damage to probe and workpiece be prevented when the probe tip touches the workpiece.

In principle, two approaches are generally accepted here, which are based either on the geometric recording of the probe tip until this position is disturbed by contact with a workpiece, or by measuring the displacements of the probe tip on contact with a workpiece.

10 Mechanical probes based on the first principle are also called switching probes, and are often known in Anglo-Saxon literature as "touch-trigger probes". The principle is based on a design in which in principle all 6 degrees of freedom of the probe are determined by the design. These 6 degrees of freedom are in principle described by 3 rotation components and 3 translation components. Considering a Cartesian X, Y, and Z coordinate system, this implies a rotation about each of the 3 axes and a translation in each axis direction, respectively. The mechanical construction on which this type of mechanical probe is based is often referred to as a kinematic coupling (see, inter alia, US 4,702,013) and is derived from the so-called "Kelvin clamp". A not unimportant advantage of this mechanical construction is that the masses and inertia of the probe parts involved in the attack can remain relatively low. In modern mechanical probes of this type, the detection task is often observed by sensitive sensors such as strain gauges or piezo elements that can detect contact between probe tip and workpiece at a very early stage of the attack, see for example EP 0 415 579 A1, EP 0 420 416 A2, US 4,702,013 and US 5,018,280. As soon as an attack is detected by the probe, a "trigger" signal is generated and the machine reads out its measuring rulers which together determine the measuring point. Then, the machine initiates a deceleration movement to stop the machine. The advantage of the commonly used 'kinematic coupling1 is that it allows a large deflection of the probe tip compared to the .... / -¾ i -3- reference position, without damage to the probe, workpiece or machine. This deflection capability is often referred to as "overtravel" and is a functional requirement for mechanical probes to prevent damage. All measuring points are related to this reference position, so the accuracy of the measured points depends in part on the accuracy with which the probe tip returns to this reference position. To overcome or correct measurement deviations of this type, systems have been designed with which these deviations can be measured or largely avoided, see for instance EP 0 415 579 A1, EP 0 307 782 B1, EP 0 423 307, EP 0 445 945 A1, DE 27 12 181 A1, DE 35 06 892 A1, US 5,018,280 and US 5,319,858.

The mechanical probe described so far only reliably generates a measuring point when attack is effected with a certain minimum speed. The "trigger" signal must reach a certain threshold value. This operation immediately implies that scanning of workpiece surfaces, where a large number of measuring points are collected from a workpiece surface to get an impression of the shape (accuracy) of the workpiece, is not possible without a new contact between workpiece for every 20 new measuring point and probe tip is realized.

Mechanical probes based on the second principle are also called measuring probes, often known in Anglo-Saxon literature as "measuring 25 probe" or "analog probe". With this type of probe, any movement of the probe tip that is not structurally recorded is measured. A measuring point is usually only "realized" when a certain force, the preset measuring force, is applied to the workpiece. Many measuring mechanical probes are based on mechanisms that allow three translation movements (X, Y, and Z movement), but do not allow rotary tips of the probe tip. Examples of this are given in DT 2 207.270 and DT 2 242 355. Systems have also been developed in which degrees of freedom other than just rotations are recorded -4. Examples of this are given in DE 28 35 615 C2, which describes a probe with two rotary axes, and in DT 1 184 972 and DT 2 019 895, which describe mechanical touch probe systems with very limited overtravel in a specific direction. Probing systems based on the measurement of 5 displacements are often large and the probe parts involved in the attack have significant masses and moments of inertia, in particular because construction elements are required to guarantee precise probe tip movements or to suppress unwanted movements, as well as to to assemble parts of measuring systems.

10

Previous research conducted by us has shown that two effects play a major role in the accurate, mechanical, measurement of workpieces using mechanical probes. First of all, it has been found that in particular the masses and mass moments of inertia of probe components involved in the attacking process play a major role in whether or not to damage workpieces. We are aware that measuring mechanical probes, in particular, can damage workpiece surfaces in such a way that functional damage can occur. Previously conducted experiments with mechanical probes of both types, ie both switching and measuring probes, in which steel and aluminum specimens were attacked at different measuring speeds, resulted in damage to the specimens under usual attacking conditions. In all cases, the damage was attributable on the one hand to the relatively high masses and moments of inertia of the 25 probe parts involved in the attack, on the other hand, the damage was attributable to the directions of movement ("overtravel" movements) that the probe suspension allowed. The dynamic attack forces, which are the cause of this test piece damage, varied between 1 N and 15 N. These are impact forces that rise above the desired measuring force (usually <1 N).

30

We have shown that the impact forces, which are the inevitable consequence of mechanical attack, can be limited by the masses and the 4 t s · - r. O V 4 '<

"I

-5- limit very much inertia of the probe parts involved in the attack. It was also shown that impact forces on workpieces can be limited by rotating the probe during the attack, instead of translating it.

5 Based on these data, one of the inventors has developed an opto-mechanical touch probe that has been previously patented (NL 1003175). Below, a new concept touch probe is described that is suitable for measurements with nanometer resolution.

The invention The invention relates to a measuring touch probe intended for use in coordinate measuring machines (CMMs) with high accuracy.

The design of the measuring touch probe consists of a number of functional parts, namely the probe (1) with which the measuring object is attacked, the probe housing (2) that forms the connection to the CMM, the suspension (3) that elastically connects the probe with the probe body and the measuring system (including R1, R2, R3 and R4) that measures the displacement of the probe relative to the probe body. The composition as well as all parts are shown in the overview drawings (top view in figure 1, section on the yz plane in figure 2 and detail A in figure 3).

20 The probe

The probe configuration is not essential. As an example, a probe (1) with a single measuring ball is used as the probe tip (4j ”), but other configurations, such as star probes, are also possible.

The probe is connected to the suspension by an intermediate body (5). The probe (1) and the intermediate body (5) must be at least rigid such that the deformation of these parts is small relative to the deformation of the elastic suspension (3). This is necessary because otherwise a substantial part of the deflection by the probe and / or the intermediate body will be captured. The elastic suspension (3) then deforms less, so that the strain gauges integrated in the suspension measure less displacement, so that the sensitivity decreases. The embodiment of the intermediate body (5) is not i Ü Ί U · ... .

-6- essential. Due to the probing forces and natural frequencies of the probe, the mass of the intermediate body must be as small as possible. Because of this mass limitation, a provision has been dispensed with which makes it possible to exchange the probe. This makes it possible to limit the effective mass rigidly connected to the probe tip 5 to a few milligrams. If another probe configuration is required, the suspension must also be changed.

The suspension 10 The suspension allows translation of the probe tip in all directions relative to the probe housing. This is necessary to absorb the overtravel of the CMM without the measuring force becoming unacceptably high. The suspension consists of one or more elastic elements. Due to the applicability of etching techniques for the manufacture of the suspension and the strain gauges vapor-deposited thereon (see also under measuring system), it is essential that all elastic elements are in one plane.

A possible suspension consists of three elongated elastic elements (3), hereinafter referred to as "blades". They connect the probe - via the intermediate body (5) - to the probe housing (2) according to figure 1. Each blade, in a good approximation, only records the movement parallel to the longest side of the blade. The arrangement drawn in figure 1 allows three movements of the intermediate body relative to the probe body, the so-called degrees of freedom, namely: translation in z-direction and rotation about x and y-axis (hereinafter referred to as Tz, Rx, respectively). and Ry), expressed according to a coordinate system as shown in figure 1 and figure 2. The other degrees of freedom, namely translation in x and y direction and rotation about the z axis, hereinafter referred to as Tx, Ty and Rz, are recorded particularly well in relation to Tz, Rx and Ry. The Tz, Rx, Ry freedom of movement gives the probe tip, provided it is not in the plane of the blades, a freedom of movement in the x, y and z direction. The use of three blades is not essential, but it is optimal. The use of more blades leads to over-determination and thus to mechanical and thermal behavior that is difficult to predict. Using fewer blades results in more than three freedoms of movement. Since every freedom of movement has to be measured, this leads to a more complicated measuring system.

In the prototype as shown in figure 1, the blades are practically perpendicular to an imaginary line between the free end of the blades and the origin. Should the blades point to one point, for example the origin, then over-determination occurs. This over-determination would lead to a stiffness in the z-direction which depends on the displacement in the z-direction applied. This is due to the fact that the length of the 10 blades projected in an xy plane decreases as they are bent out of the xy plane. Due to this decrease in the effective length of the blades, when the probe tip is moved in the z direction, rapidly increasing forces in the longitudinal direction of the blades occur, as a result of which the z component of the stiffness increases rapidly. Moreover, the mechanical and thermal behavior becomes difficult to predict because of the over-determination. Furthermore, blades pointing to one point hardly capture the Rz's freedom of movement, so that an additional measuring axis is required, especially if star probes or variations thereof are used. When using the configuration shown in Figure 1, the stiffness in the z direction remains approximately constant. In that case a rotation around the z-axis does occur in displacement in the z-direction, because of the reduction of the length of the blades projected in the xy-plane. This rotation is calculated from the z displacement of the probe, after which a software correction for this rotation is applied.

The measuring system 25 The measuring system accurately determines the orientation and the position of the probe relative to the probe housing. Since some movements are excluded by the suspension, it is only necessary to measure the remaining movements. In the described invention, the orientation and position of the intermediate body, and thus the position of the tip, is read by strain gauges on the elastic elements, such that at least all degrees of freedom released can be measured. A strain gauge is defined as an electromechanical sensor that has an electrical resistance that depends on the mechanical strain at the sensor's location. Because all elastic elements lie in one plane, it is possible to produce the suspension, the strain gauges and the electrical connections in a number of lithography and etching steps. For this, use is made of the so-called "microsystem technology" (MST). The substrate then generally consists of crystalline silicon, the strain gauges of polycrystalline or crystalline doped silicon and the electrical connections of a conductor, such as gold or aluminum. It is also possible to integrate electronic components with the suspension. In this way, a preamplifier can be integrated with the suspension, thereby limiting the reception of interference signals in the lines. It is even possible to convert the signal directly to a digital signal, which further reduces the reception of interference signals.

Since the elastic elements may be etched from the substrate material, they will often be silicon in that case. This makes the suspension sensitive to shock loads because of the brittleness of silicon. It is also possible to manufacture the elastic elements from a metallic material, but then the strain gauges must be glued to the blades and the electrical connections must be made manually. Extremely thin wires must be used for this, in order to disturb the elastic suspension as little as possible.

In the suspension with blades, four strain gauges are applied to each blade according to Figure 3. The resistors are then connected in a complete Wheatstone bridge according to Figure 4. Because the blades will usually be bent in an S-shape, according to Figure 5, the resistance change of the strain gauges R1 and R4 opposite to the resistance change of the strain gauges R2 and R3. Therefore, it is sufficient to have only strain gauges at the top, which greatly simplifies the etching process. The use of four strain gauges per measuring axis is not essential, but it is optimal. It is known from the literature that the resistance of strain gauges depends not only on the elongation but also on the temperature. When using a half (two strain gauges) or a whole bridge (four strain gauges), this is corrected for because all resistances change by the same amount. Compared to a half bridge, a whole bridge has a greater sensitivity and thus a higher signal-to-noise ratio.

The measurement data of all elastic elements is converted into tip displacements. If the tip displacement is small, the bandage can be linearized and written in matrix form: (5) m-m0 = ATlip (5)

In this m is a measuring vector with the measuring data of all elastic elements, m0 is the measuring vector at rest, Tlip is a vector of tip displacements and A is an nx3 matrix with n the number of measuring axes. The components ai}, i = 1 = 1..3 of the matrix A are obtained by calibration.

The calibration arrangement consists of an accurate one dimensional displacement generator (cross section shown in Figure 6) and a holder (three views shown in Figure 7) that positions the touch probe against the displacement generator 15 in at least three mutually independent directions. The displacement generator realized by us consists of a plateau (6) with a reference mirror (7) directly connected to it, a flat mirror (9) which can be moved by an actuator (8) and a laser interferometer with a flat mirror optic (10) that measures the distance between the plateau and the flat mirror measure differential. The laser beams (11) are mirrored by a deflecting mirror (12) 20 such that they incident perpendicular to the two flat mirrors (7 and 9). The holder (shown in three views in Figure 7) can position the touch probe in 7 independent directions against the displacement generator. To this end, the holder has three sets of three V-grooves (13) at 120 °, on which an adapter (shown in figure 8) can be placed via three balls (14), analogous to the so-called "Kelvin vapor". The touch probe is attached to the holder via the adapter. Because the Kelvin Clamp has triple symmetry, the adapter can be positioned in each clamp in three ways. This leads to 9 different orientations, but 3 depend on it, leaving 7 independent orientations.

From the sensitivity to displacement in the different directions, the matrix A can be determined via a least squared estimate.

-10-

Matrix A can also be calculated, although it is based on some approximations. An explanation of the calculation can be found in Appendix A.

5 The technology described has made it possible to realize probing systems with the following properties: • displacements of the probe tip can be measured in three orthogonal directions (3D) • the resolution with which these displacements can be measured 10 is a few nanometers • the measuring forces, including including the dynamic measuring forces that occur during normal use are limited to a few millinewtons.

The results were obtained with an elastic suspension with integrated measuring system manufactured by means of MST. The blades have a length of 1.6 mm and a diameter of 160 * 40 µm. The probe length is 10 mm.

- 11 -

Appendix

In this appendix, the model is described that the mechanics of the touch probe as shown in figure 1 to figure 3 and the sensitivity of the strain gauges (including R1, R2, R3 and R4 in figure 3) on the elastic 5 elements (3 ) describes. The model has also been used to choose optimal values for the dimensions and position of the blades.

Stiffness of the suspension

Consider the suspension with dimensions as shown in Figure 1, Figure 2, Figure 10 and a probe tip located at the point {0,0,1st) below the origin. For the stiffness that is felt when applying force against the probe tip, it can be calculated:

Esbsds3 Ci = 3 Is3 3 Esbsds3 1 ys 1 ys2 xs2 '

Cxy = iisist2 Uii + v) 0 '"7T + 3 + Ί? ~ + Ί? ~ /

Where c, the said stiffness in the z-direction, cv the said stiffness in an arbitrary direction in the xy-plane, Es and v is the modulus of elasticity and the cross-contraction coefficient of the material used for the elastic elements, respectively, the thickness of the elastic elements and a, a torsion "constant" which depends on ds / bs and approaches ^ when ds / bs is small.

20 Mechanical stresses

The mechanical stresses cr, i = x, y, zj = x, y, z in the blades are calculated to determine at which deflection the stress in the blades exceeds the yield point and to determine the strain at the strain gauges from which the sensitivity of the measuring system to an applied displacement can be determined. The said stresses are calculated by calculating the displacement of the blades at the location of the attachment with the intermediate body. This displacement is converted to a local coordinate system of the relevant boom, after which the stresses in the relevant boom can be calculated. The phrases will be published in a thesis on probing systems in 1999.

5 Sensitivity and resolution of the measuring system

The displacement of the probe tip is measured by four strain gauges on each boom, as shown in Figure 3. The change in strain gauge resistance is measured in a so-called Wheatstone bridge, as shown in Figure 4. If all resistances are approximate being the same size 10 holds approximately: M = y

0 4R

where ΔΛ, represents the resistance change of strain gauge /, R is the nominal resistance of the strain gauges, V0 is the supply voltage of the bridge and M is the unbalance of the bridge in question in volts. For the relative resistance change of a strain gauge holds:

ΔΛ G, G

- = - σ + - (4) R Es> y Es a where G, and G are the longitudinal and the transverse gauge factor, respectively. The imbalances of the bridges can now be calculated from the calculated mechanical stresses. If the tip displacement is small, the bandage 20 can be linearized and written in matrix form, as previously done in equation (1): m-m0 = Lip (5)

The sensitivity of m for tip displacements depends on the direction. For example, the sensitivity to z-direction displacement will usually be greater than the sensitivity to x-direction displacement. The minimum sensitivity s3 is the smallest singular value of A, which can be found by means of a so-called "singular value decomposition" (SVD, known in Anglo-Saxon literature as "singular value decomposition"). If the minimum 1 01 CGS4-1 i -13 signal size that can be measured Mmin is known, the resolution may be written: M-res = —— (6).

*

Overtravel 5 In case the so-called comparison stress σ (in the blades exceeds the yield point amax, the blades will either deform plastic (in case of plastic material) or break (in case of brittle material). The distance the probe tip can travel before a (in the blades the stretch limit exceeds is called the 'overtravel'. There are several accepted ways to calculate a, 10. A common one is that of 'von Mises': af = σ2 + 3τ2 = af, + af, + a2 + i (af2 + af, + af,) (7).

It turns out that af contains only terms that are quadratic in the components of Ttj, so that it can be written as: =% Ν, φ (8) 15 where B is a symmetric matrix. The overtravel depends on the direction in which the probe tip is moved. It can be shown that the overtravel in the direction in which the overtravel is smallest (ov / min) holds: ovt. (9) where λ is the greatest eigenvalue of B.

20 J ^ '1

Claims (8)

1 4 1. Probe system, characterized in that it consists of a probe, including a probe strip, a probe housing, a suspension which rigidly links some translation or rotation directions and others elastically to said probe housing such that the probe strip can move in three independent directions, whereby use is made of thin elastic elements which all lie in one and the same plane and a measuring system that unambiguously records the displacement of said probe strip relative to said probe housing by means of strain gauges 10 on said elastic elements and that the suspension and the measuring system be integrated by means of a series of lithography; and etching steps (MST). Touch probe system according to claim 1, characterized in that said suspension and said measuring system are manufactured separately, wherein said strain gauges are applied separately to said elastic elements. R Touch probe system according to claims 1-2, characterized in that it is manufactured with said suspension technique and said measuring system by means of the manufacturing technique stated in claim 1 and integrated with an electrical pre-amplification of the measuring signals originating from said 20 strain gauges. of an extra series of lithography and etching steps (MST). 4. Probe system according to claims 1-3, characterized in that, with the manufacturing technique stated in claim 1, an electrical pre-amplification of the measuring signals originating from said strain gauges as well as a circuit converting these signals into digital values is manufactured integrally with said suspension and said measuring system by means of an additional series of lithography and etching steps (MST). Probing system according to Claims 1 to 4, characterized in that the conversion from a vector, with the individual measuring signals from said measuring system as components, to a vector Ttlp, with the components x-, y- and z- respectively displacement of the probe tip, is done using a matrix A according to m-niQ * = ATjjp 35 where iïi = rüo for Tjjp = O. The scanning system according to claim 5, characterized in that the matrix A is determined by a calibration using a calibration arrangement consisting of a one-dimensional displacement generator, the displacement of which is measured with a special differential interferometer so that said displacement is known with sufficient accuracy and from a holder which positions the touch probe in at least three mutually independent directions against the displacement generator, the components of said matrix A being calculated by means of a least squared estimate based on the calibration set-up of measured data. The scanning system according to claims 1-6, characterized in that the displacements are measured with a resolution of less than 10 nanometers and with an estimated three-way uncertainty of less than 25 nanometers and with a measuring force of less than 1 millinwton.