EP1264156A1

EP1264156A1 - Rotary pivoting device for the probe of a coordinate measuring device

Info

Publication number: EP1264156A1
Application number: EP01953026A
Authority: EP
Inventors: Werner Lotze; Karl-Hermann Breyer; Peter Ebersbach
Original assignee: Carl Zeiss AG
Current assignee: Carl Zeiss AG
Priority date: 2000-02-16
Filing date: 2001-01-18
Publication date: 2002-12-11
Also published as: JP2003526090A; WO2001061273A1; DE10007062A1

Abstract

Disclosed is a rotary pivoting device for the probes (4) of coordinate measuring devices, comprising at least two hinges (14, 15) for angular orientation of said probes, in addition to a correction unit (22) wherein at least the position of the axes of rotation (aA, aB) of the hinges and deviations in the angular position of said hinges can be corrected. The correction process is simplified by correcting the above-mentioned errors with the aid of a common mathematical model.

Description

Rotary swivel device for the probe of a coordinate measuring machine

The invention relates to a rotary swivel device for the probe head of a coordinate measuring machine, with at least two pivot joints for angular alignment of the probe heads.

Turn-swivel devices have been known for a long time. On the one hand, there are the so-called continuously rotatable rotary swivel devices, in which the angle of rotation can be continuously adjusted via a corresponding motor and the exact angle of rotation is supplied by an encoder, and the so-called latching rotary swivel devices, in which only a limited number of angles of rotation can be adjusted. In a locking rotary swivel device marketed by the applicant with the type designation “RDS”, two interacting toothed rings of a so-called Hirth mink toothing are used for this purpose, which mesh in the locked state and thereby lock the respectively set angle of rotation.

A continuously rotatable rotary swivel device is disclosed in DE 37 400 70 AI and the corresponding US Patent 4,888,877. This shows a rotary swivel device which has two motor-driven, continuously pivotable swivel joints, the axes of rotation of which are perpendicular to one another. So that the stylus does not have to be calibrated each time the angle of rotation of the swivel joints changes, correction values are stored with regard to the different angles of rotation that describe the position of the axes of rotation relative to one another. In addition, correction values with respect to angular position deviations and the running deviations of the axes of rotation can alternatively or additionally also be provided, with the implementation of this last-mentioned correction not exactly describing how this can be used particularly advantageously. The peculiarity of the rotary-swivel unit shown in DE 37 400 70 AI is to be seen in the fact that a separate correction model is required for each of the individual errors, so that on the one hand the error parameters have to be determined for several correction models and in the correction of the measured ones Measured values must be carried out several correction calculations. In the past, this has resulted in the fact that only the position of the axes of rotation relative to one another and the angular position deviations of the rotary joints have been corrected due to the high demands on a short measuring time. However, these corrections were only applicable to continuously rotatable rotary swivel devices, as shown in DE 37 400 70 AI. The correction models were inadequate for latching rotary swivel devices in which the rotary swivel device can be locked in a variety of different positions. In principle, it is also desirable to further improve the measurement accuracy for continuously rotatable rotary swivel devices.

German Offenlegungsschrift DE 40 01 433 AI and the corresponding US Pat. No. 5,138,563 show a correction method for rotary swivel devices in which errors are corrected which result from the elastic deformation of the rotary swivel device and the pushbutton configuration, in particular if the Probe configuration includes a probe extension by which the probe is held. For this purpose, the method for correcting the position of the axes of rotation described in DE 37 400 70 AI, already cited above, was expanded by a term B (α, β) which indicates the deformation. The rotary swivel device with the probe configuration attached to it was considered here according to the model of a bending beam, so that to determine the term

B (α, ß) the maximum deflection of the rotary swivel device was determined and in

Depending on the respective angular position of the swivel joints was then interpolated to the corresponding deflection. The model has served excellently in the past. Due to the constantly increasing demands on the measuring accuracy of coordinate measuring machines, an attempt was made to make the probe extension in DE 40 01 433 AI designated by reference number 5 more rigid. It turned out, however, that the correction results did not improve as expected, because this caused the deformations occurring in the rotary-swivel device became larger than the deformations in the probe head extension and therefore the underlying model. s bending beam ^ ns no longer worked.

The European publication EP 07 59 534 A2 shows a method for correcting continuously rotatable rotary swivel devices or latching rotary swivel devices. It is proposed to bring the rotary swivel device into two different angles of rotation and to calibrate each one here. A third angle of rotation lying between the angles of rotation is then corrected by interpolating between the recorded calibration data. Nothing is explained here with regard to the correction model.

Based on this, the object of the present invention is to provide an improved rotary / pivoting device.

The object is solved by the features of independent claims 1 and 6.

The basic idea of the solution according to independent claim 1 provides for at least the position of the axes of rotation relative to one another and the angular position deviations to be corrected in a common mathematical model.

At this point it should be expressly mentioned that the term common mathematical model is to be understood in such a way that the individual errors do not have to be calculated separately.

As a result, both the computational effort for determining the error parameters can be considerably reduced, and also the time required to determine the individual errors.

The correction unit is a computer or a microprocessor in which the correspondingly recorded correction values are stored and which corrects the measured values according to the selected mathematical model. The rotary swivel device according to the invention can be developed particularly advantageously by additionally correcting the wobble errors and / or the radial running deviations and / or the axial displacements of the rotary joints in the common mathematical model.

This measure results in a manageable correction model which can now also be used to correct, in particular, latching rotary-pivoting devices, which was not yet possible with the previously known mathematical models.

A Hirth toothing should be used particularly advantageously for locking the rotary positions, since this enables all rotary angle positions of the rotary / swivel device to be set with high reproducibility.

The basic idea of the solution according to independent claim 6 is to be seen in the fact that a latching rotary swivel device is now created, in which it is no longer necessary to calibrate each rotary position individually for each probe configuration.

At this point it should again be expressly mentioned that the rotary swivel device according to claims 1, 2 and 5 can be a latching rotary swivel device as well as a continuously rotatable rotary swivel device.

Further advantages and developments of the invention result from the description of the figures.

Fig. 1 shows a perspective view of the rotary swivel device

FIG. 2 shows a purely schematic illustration of the rotary swivel device according to FIG. 1

Fig. 3 shows the rotation of the center (P) of the probe ball (12) around an ideal hinge

(14) without running deviations. FIG. 4 shows the rotation of the center point (P) of the probe ball (12) around a real swivel joint

(14) with run deviations 5 shows a schematic illustration of the errors in the rotation of the center point (P) of the probe ball (12) around a real swivel joint (14) with running deviations. FIG. 6 shows a measurement setup with which the correction parameters for the running deviations of the swivel joint (14) FIG. 7 shows a measurement setup with which the correction parameters for the running deviations of the swivel joint (15) can be determined. FIG. 8 shows a purely schematic illustration of the principle of the rotary swivel device

Fig. 1, wherein the probe configuration includes a probe extension (19) and

Calculation of the elastic deformation two finite elastic elements (17 and 18) were inserted. FIG. 9 shows a purely schematic basic illustration of the elastic center (K) of a finite element. FIG. 10 also shows a purely schematic basic illustration of the elastic center (K) of a finite element with the displacement of the center (P) of the probe ball (12) Fig. 1 1 shows the rotary swivel device shown in Figure 1 from the front with a

Star button (21).

Figure 1 shows a rotary swivel device, which is designed here as a so-called latching rotary swivel device. The rotary swivel device is fastened to a horizontally aligned measuring arm (8) of a column measuring device and has two swivel joints (14, 15) which rotate the components (1) and (2) and components (2) and (3) Connect with each other, whereby the swivel joints (15, 14) define the axes of rotation (a _A ) and (ae). The rotary swivel device has so-called Hirth serrations (6) and (7) for fixed locking of the set angle of rotation. These are gear rings that work in pairs and mesh with one another. In order to change the angles of rotation of the rotary joints (14, 15), there is a pneumatic cylinder inside the rotary swivel device, via which the component (2) relative to the component (1) and the component (3) relative to the component (2) can be lifted off. In addition, an electric motor is provided for each of the rotary joints, by means of which the angle of rotation of the respective rotary joint (14, 15) can be adjusted. After the desired angle of rotation is reached, the excavated components (1) and (2) or (2) and (3) are pulled together again by the pneumatics. In the illustration shown, a probe (4) of the switching type is attached to the receptacle of the rotary swivel device. A probe pin (11) with a probe ball (12) is in turn held interchangeably on the probe head (4), the probe head (4) triggering an electrical signal when a workpiece touches the probe ball (12). Of course, an optical probe or a measuring probe can also be used, for example. The probe (4) is attached to the holding plate (3) by means of an adapter part (5).

The measurement values recorded by the coordinate measuring machine during a measurement are corrected in the correction unit (22) shown here only in a purely schematic manner, according to a mathematical model. This correction unit (22) is usually the computer of the coordinate measuring machine. Of course, alternatively, a specially provided microprocessor can also be used, which is arranged, for example, in the control of the coordinate measuring machine or even in the rotary-swivel unit itself.

1. Basic model

In order to be able to carry out measurements with such a rotary swivel device with different angles of rotation of the rotary joints (14, 15), the exact location vector (x) of the center point (P) of the probe ball (12) must be related to the device coordinate system (XQ, Y _G , Z _G ) be known. This state of affairs will be explained below with reference to FIG. 2. Figure 2 shows only a purely schematic representation of the rotary-pivoting device according to Figure 1, in which the same reference numerals as in Figure 1 were used for the same components.

This location vector (x) of the probe ball (12) in the device coordinate system (X _G , Y _G , Z _G ) can be specified as a vector equation as follows:

Equation 1 x = T; ¹ R ^ (T ¹ R _B t + d) + c _/ The points (A) and (B) represent the intersection of a straight line with the axes of rotation (a _A ) and (a _ß ), the straight line running along the shortest distance between the two axes of rotation (a _A ) and (a _ß ) , The vector (x) here means the location vector of the center point (P) of the probe ball (12) based on the device coordinate system (X _G , Y _G , ZG). The vector (t) is the vector from the point (B) to the center (P) of the probe ball (12). The vector (c _A ) is the location vector of the point (A) based on the device coordinate system (X _G , Y _G , Z _G ). The vector (d) is the distance vector of point (B) from point (A). The rotation matrix (R _A ) describes the rotation of the swivel joint (15) about the axis of rotation (a _A ). The rotation matrix (RB) describes the rotation of the swivel joint (14) about the axis (a _ß ). The transformation matrix (T _A ) describes the transformation of the device coordinate system (XG, YG, ZG) into the joint coordinate system (XA, YA, Z _A ) at point (A). The transformation matrix (T _B ) describes the transformation of the joint coordinate system (X _A , Y _A , Z _A ) at point (A) into the joint coordinate system (X _B , Y _B , Z _B ) at point (B).

This equation 1 corresponds to the equation described in DE 37 400 70 AI and is only mathematically described somewhat differently.

In the case of the ideal, error-free swivel joints, their movement is a pure rotation and the rotation matrices (R _A ) and (R _ß ) apply to the rotation around the Z axis of the respective joint coordinate system (X _A , Y _A , Z _A or Xß, Yß, Z _ß ).

Equation 2 R _A

In this equation 2, (φ) denotes the angle of rotation about the respective swivel joint (14, 15). For the rotation matrix (R _A ) it is the angle of rotation (Ψ _A ) of the swivel joint (15), for the rotation matrix (R _ß ) it is the angle of rotation (c _ß ) of the swivel joint (14)

All other vectors and matrices are also unknown in the case of ideal swivel joints and must be determined experimentally, namely 9 components of vectors and 6 solid angles of the transformation matrices. The zero angles of the two angle measuring systems are also unknown. Accordingly, for the calibration of the rotary Swivel unit measurements for at least 17 independent condition equations. This is still the case if the correction of the individual swivel joints is known.

2. Kinematic correction

In a zero approximation, a swivel joint has only a single kinematic degree of freedom, which can be described with the rotation matrix according to equation 2 as a pure rotation around the axis of rotation, as is shown in FIG. 3 purely by way of example for the rotation about the axis (aβ). As can be seen here, in this case the center of the probe ball (P) is rotated by an angle (ψ _ß ) around the axis of rotation (a _ß ) to a point (P ') so that both lie on a circle, its plane is perpendicular to the relevant axis of rotation and its center (M) lies on the axis of rotation (a _ß ).

On the other hand, due to the manufacturing inaccuracies, a real swivel joint has movements in all six degrees of freedom, as can be seen well from the purely schematic FIG. 4. As can be seen from this, the center of the probe ball (P) is rotated this time around the axis of rotation (a _ß ) on a point (P ") that is not on the said circle around the axis of rotation (a _ß ). The six degrees of freedom in that the movements can take place at the same time correspond to the six error components that have to be added to the theoretical point (P ^ς ) in order to arrive at the actual point (P "), as shown in FIG. 5. These error components are:

- Angular position deviations (δφ = δ _z ) of the locking system or angle measuring system

- radial run deviations (y _. , v _y ) in the x and y direction,

- axial displacements v _z

- tilts δ _x , δ _y around the x and y axes as wobble errors.

Instead of the rotational movement, the guided part thus performs a general rigid body movement in space, which consists of the displacement vector (v) and the vector (δ) of the spatial rotations. Accordingly, based on the rotary swivel device according to FIG. 2, the spatial displacement of the center point (P) of the probe ball (12) is made up of seven components during a rotation about one of the two rotary joints (14 or 15) from (P) to (P ") together, namely the nominal rotation by the angle of rotation (ψ _A or cp _ß ) and the six error components belonging to the respective rotation, ie three shifts (v _x , v _y , v _z ) and three rotations (δx, δy, δφ).

Under the above considerations, the basic model mentioned above can therefore be expanded as follows:

Equation 3 x = T _A - ^I R _A [D _A (τ- ^, R _B (D _B t + v _B ) + d) + v _A + c.

The components expanded compared to the basic model according to equation 1 have the following meaning. The vector (v _A ) here means the displacement error caused by the swivel joint (15). The rotation matrix (D _A ) represents the rotation error that arises around the swivel joint (15). The vector (v _B ) means the displacement error caused by the

Swivel joint (14) is created. The rotation matrix (D _B ) represents the rotation error that arises around the swivel joint (14).

The vectors (v _A ) and (v _B ), which are each composed of the displacement errors in the x, y and z direction, are therefore defined as follows:

Equation 4 ^V _A

The spatial rotation matrices (D _A ) and (D _B ), which each consist of the individual rotations D _x , D _y and D _z about the coordinate axes x, y and z with Euler's angles δ ₂ , δ _y , δ _x consequently as follows:

1 0 0 cosδ "0 - cosδ. cosδ .. sinδ _z 0

Equation 5 D _A = D _B 0 cos δ _x cos δ, 0 1 0 - sin δ, cos δ. 0 0 - cos δ. cosδ. sinδ. 0 cos δ. 0 0 V The determination of the vector (v _B ) for the displacement error in the swivel joint (14) and the rotation matrix (D _β ) for the rotation error of the swivel joint (14) can be carried out relatively easily, as will be explained in connection with FIG. 6. For this purpose, the rotary swivel device is clamped on a highly precise coordinate measuring machine and a ball test body (9) is attached to the rotary swivel device, which has at least three balls (16a, 16b, 16c). Now the swivel joint (14) is brought into any of its possible rotation angles and the respective position of the balls (16a, 16b, 16c) is measured with a centering button, which is constructed similar to a thimble. It has been shown here that the errors are only sufficiently small if measurement is carried out with a sufficient measuring force. However, this leads to a relatively large bending of the ball test piece (9). Therefore, the exact position of the balls (16a, 16b, 16c) can be determined particularly advantageously by measuring with two different measuring forces and then extrapolating to the position which the balls have at the measuring force 0 N. For each angle of rotation of the swivel joint (14), a plane is then spanned from the position of the balls (16a, 16b, 16c) and the center of gravity is determined from the measured ball positions. The vector (v _B ) for the displacement errors then results as a vector from the center of gravity in the reference angle of rotation of the rotary-pivoting device to the center of gravity in the current angle of rotation. The error angles for the rotation matrix (D _ß ) result from the rotations of the calculated plane in the reference rotation angle to the calculated plane in the current rotational position.

The vector (v _A ) for the displacement errors in the swivel joint is also completely analogous

(15) and the rotation matrix (D _A ) for the rotation errors of the swivel joint (15) are determined, in which case the ball test piece (9) is attached to the rotary swivel device via an angle piece (10).

3. Correction of elastic bending errors

Under real conditions, due to the weight forces, in particular the probe configuration, that is to say the probe (4), the stylus (11) and in particular Probe head extensions (19) ∑τι Deformation of both the rotary swivel device, as well as the probe configuration seber (see Fig. 8).

For the calculation of this elastic deformation, finite elastic elements are introduced according to the invention, with which the deformation of elastic systems under external static loading can be described. On this basis, manageable analytical model equations for the correction of the deformation can be derived, the coefficients of which can be determined by means of best fit calculations from a number of positions as well as deformation and load conditions. The replacement model is to be explained here with reference to FIGS. 8 to 10. As can be seen from the purely schematic schematic diagram according to FIG. 8, the finite elements (17) and (18) are placed here in this particularly advantageous embodiment at the points at which the components (1) and (2) or which are rotatably connected to one another (2) and (3) abut each other, that is, at the point where the Hirth serrations (6,7) are located (cf. FIGS. 1 and 2). These finite elements (17, 18) can be presented in simplified form as rubber disks that elastically connect parts (1) and (2) or (2) and (3) to one another.

The model is based on the assumption that the deformation is a spatial displacement and rotation between the articulated components (1) and (2) with respect to the swivel joint (15) and the components (2) and (3) with respect to the swivel joint (14) can be described, while the other components, such as the components (1), (2) and (3) of the rotary swivel device, the stylus (11), the probe (4) and the probe extension (19) as completely rigid be accepted. The total error can then be calculated from a superposition of the corrections of the rigid model, as described above in equation 1 or equation 3, and the correction of the elastic bending errors, as described here. If one describes the deformation correction vector by which the joint (15) deforms as (ü _A ) and the deformation correction vector by which the joint (14) deforms as (ö _B ), the following equation results when combined with equation 3:

Equation 6 x = T _A -'R _A [D _A (τ _B ^I R _B (D _B t + v _B ) + d) + v _A ] + c _A + Ü _A + Ü _B It goes without saying that equation 6 is particularly advantageous because, in addition to the deformation, angular position deviations, radial running deviations, axial displacements and wobble errors of the rotary joints (14) and (15) are also corrected, as stated above. Of course, the deformation correction vectors (ö _A ) and (G _B ) can be combined with Equation 1 just as well, for example if a continuously rotatable rotary-swivel unit is used, the rotary joints of which have very few errors of the type described. Likewise, the deformation correction vectors (ü _A ) and (ü _B ) can also be calculated completely separately.

Thus, as FIG. 9 shows, a finite element can be treated mathematically as if only a force vector (f) and a moment vector (m) were to attack in the center (K) of such a finite element (17), where the force vector (f) and the

Torque vector (m) are generated by the external load, ie the weight forces of the probe configuration and possibly by the measuring forces. This model presupposes that the elastic center (K) of the finite element with its position and its orientation in space as well as with its elastic parameters contains the elastic properties of the deformed components. In addition, the deformation must be linearly dependent on the loads and proportional to the forces and moments acting in the elastic center (K). The superposition principle must continue to apply. The finite element reacts to the force vector (f) and the moment vector (m) with a

Deformation correction vector (ü), which is composed of a translation vector (w) and a rotation vector (γ). The deformation correction vector (ü) can be determined as follows:

Equation 7

Where (N) is the compliance matrix that contains the compliance matrices (Nu) to (N22) as a hypermatrix. Written clearly, this means the following for the translation vector () and the rotation vector (γ): Equation 8 w = N ,, f + N ₁₂ m

Equation 9 γ = N ₂₁ f + N ₂₂ m

In this equation, the compliance matrices (Nu) through (N ₂₂ ) mean the following:

Nj ι = translation due to the force vector (f) acting in the elastic center (K)

N ₁₂ = translation due to the moment vector (m) acting in the elastic center (K) N ₂ ι = rotation due to the force vector (f) acting in the elastic center (K)

N ₂₂ = rotation due to the moment vector (m) acting in the elastic center (K)

The compliance matrix is defined in the coordinates ( _K , yi, Z _K ) of the elastic center (K) and must be transformed into the current device coordinate system (X _G , Y _G , Z _G ). In the real system, as can be seen from the purely schematic schematic diagram according to FIG. 10, the dead weight of the probe configuration, such as the probe (4), acts as an external load according to the well-known formula f = m * g in the center (S). The dead weight consequently causes the moment vector (m) from the force vector (f) and the distance vector (s) between the elastic center (K) and the center of gravity (S) according to

Equation 10 m = f x s

The deformation correction vector (ü) by which the center point (P) of the probe ball (12) is displaced as a result of the deformation then results from the superposition of spatial displacement according to the translation vector (w) and the rotation according to the rotation vector (γ) as follows:

Equation 11 ü = w + γ x p

= N _u f + N ₁₂ (f χ s) + (N ₂₁ f + N ₂₂ (f χ s)) xp wherein the vector (p) is the distance vector between the elastic center (K) and the center (P) of the probe ball (12). How this is composed is shown in FIG. 10. The vector (γ xp) is shown here as vector (i).

If you choose the coordinate system so that the force vector (f) is elastic

Center (K) alone causes a translation and the moment vector (m) in the elastic center (K) alone causes rotation, the sub-matrices (Nι) and (N ₂ ι) can be replaced by null matrices. With long lengths of the distance vector (p) and thus also the distance vector (s) and high translational rigidity of the elastic replacement element, the translation vector () and consequently also the compliance matrix (Nu) can be neglected. With a corresponding choice of the coordinate system (X, Yκ, Zκ) of the finite elastic elements, the coefficients outside the main diagonals of the compliance matrices (N, _j ) become zero. With these simplifications, the following equation now applies:

Equation 12 ü «γ xp = (N ₂₂ (fx sJJ xp

The original 36 coefficients of the hypermatrix (N) have thus been reduced to three rotary compliance coefficients and the remaining matrix is then:

Φ, 0 0

Equation 13 N ₂₂ = N _φ = φ ₂ 0

0 ^φ j

In this matrix N _2, (Φ,, Φ ₂ ) mean the flexibility for the tilt about the x and y axis (Xκ, Yκ) of the coordinate system of the finite elements and (Φ ₃ ) the rotation about the z axis (Z ).

The deformation correction correctors (Q _A ) and (ü _B ) are therefore calculated separately for each of the finite elements (17, 18) according to equation 12. As already mentioned at the beginning The choice of the number of finite elements (17, 18) and their position here is particularly advantageous in relation to the direction of rotation and rotation shown here. In principle, both the number and the location of the finite elements can be freely selected. For example, a single finite element is sufficient. The location is also variable. If, for example, the elastic deformation of the horizontally aligned measuring arm of a column measuring device and / or the connection of the rotary-pivoting device to the measuring arm is to be recorded, the finite element (18) should be moved further towards the measuring arm.

In order to determine the parameters of the compliance matrix (N ₂₂ ) both for the joint (14) and for the joint (15) as well as the vector (t) from the center point (P) of the probe ball in a probe configuration newly recorded on the rotary swivel device (12) to determine point (B), at least 8 calibrations in different rotary positions of the rotary joints (14) and (15) must be carried out on the calibration standard of the coordinate measuring machine. Often, however, probe configurations, such as the stylus or the probe extension, have to be replaced in everyday measurement. If the same button configuration is used again at a later point in time, the above parameters must be determined again, which is relatively time-consuming. If one assumes that when the same probe configuration is replaced, the elastic properties of the probe configuration and thus the parameters of the compliance matrix (N ₂₂ ) remain almost unchanged, in this case it is sufficient to determine only the vector (t), so that in principle, only a single calibration on the calibration standard would suffice.

This method can be used particularly advantageously when calibrating star styli. In this regard, FIG. 11 shows a schematic diagram in which only the component (3) of the rotary-swivel unit shown in FIG. 1 can be seen, to which a probe extension (19) with probe (4) and a star stylus (21) is connected. If one assumes that approximately the same bending parameters are present for all probe balls (20a, 20b, 20c) of the star stylus (21), then it is sufficient for only one of the probe balls (20a, 20b, 20c) the parameters of the compliance matrix (N ₂ ) to determine. Only the vector (t) then needs to be determined for the other probe balls (20a, 20b, 20c).

Claims

claims:

1. Rotary swivel device for probe heads (4) of coordinate measuring machines, with at least two swivel joints (14, 15) for angular alignment of the probe heads, in which the rotary swivel device is assigned a correction unit (22) in which at least the position of the axes of rotation ( aA, a _B ) of the rotary joints (14, 15) to one another and the angular position deviations (δ _φ = δ _z ) of the rotary joints are corrected, characterized in that these errors are corrected using a common mathematical model.

2. Rotary swivel device according to claim 1, characterized in that the correction unit (22) is designed such that in addition to the common mathematical model also the wobble (δ _x , δ _y ) of the rotary joints and / or the radial run deviations (v _x , v _y ) and / or the axial displacements (v _z ) of the rotary joints (14, 15) are corrected.

3. rotary swivel device according to claim 1 or 2, characterized in that a latching rotary swivel device is used.

4. Rotary swivel device according to one of claims 1 to 3, characterized in that a Hirth toothing (6,7) is used in the rotary swivel device for locking the rotational positions.

5. Rotary swivel device according to one of claims 1 to 4, characterized in that the following mathematical model is used:

= T _A - ¹ R _A [D _A (τ _B ¹ R _B (D _B t + v _B ) + d) + v ₄ + c.

where mean: x location vector of the probe center in the device coordinate system (X _G , Y _G , Z _G ) t distance vector from point (B) to the center (P) of the probe ball (12) c _A location vector of point (A) in the device coordinate system (X _G , Y _G , Z _G ) d distance vector between point (A) and point (B)

R _A rotation matrix for the rotation of the swivel joint (15)

R _B rotation matrix for the rotation of the swivel joint (14)

T _A transformation matrix for the spatial position of the system in point (A) related to the

Device coordinate system (XG, YG.ZG) T _B transformation matrix for the for the spatial position of the coordinate system in the point

(B) based on the coordinate system in point (A) v _A vector for the displacement error of the swivel joint (15) v _B vector for the displacement error of the swivel joint (14) D _A rotation matrix for the rotation error around the swivel joint (15) D _ß rotation matrix for the rotation error around the swivel joint (14)

6. rotary swivel device for probes (4) of coordinate measuring machines, with at least two rotary joints (14, 15) for angular alignment of the probes, a Hirth toothing (6, 7) being used for latching the rotational positions, characterized in that the A rotary unit is assigned a correction unit (22) in which the measured values are corrected.

7. rotary swivel device according to claim 6, characterized in that the position of the axes of rotation (a _A , a _B ) of the rotary joints to each other and / or

Angular position deviations (δ _φ = δ _z ) of the rotary joints and / or the wobble errors (δ _x , δ _y ) of the rotary joints and / or the radial run deviations (v _x , v _y ) and / or the axial displacements (v _z ) of the rotary joints (14 , 15) are corrected.