Confrerence proceedings of 1st international conference and general meeting of the european society for precision engineering and nanotechnology, bremen, germany, may 31st - june 4th 1999.

New coordinate measuring machine featuring a parallel mechanism

Takaaki OIWA 1

1 Shizuoka University , Johoku, Hamamatsu, 432-8561, JAPAN


This study proposes a new coordinate measuring machine (CMM) based on a parallel mechanism. The use of this new mechanism will potentially improve the stiffness, accuracy, and efficiency of the CMM. This paper describes the fundamentals, an experimental CMM in detail. The proposed CMM is composed of an octahedral truss frame, specially made revolutionary and spherical joints, three actuated struts and a touch trigger probe. Moreover, simple calibration and accuracy test of the CMM are performed. The results of the test show that measuring repeatability is less than 0.15 m, and measured deflections are less than 10 micrometer when measuring the gauge blocks in XYZ directions.


In recent years, coordinate measuring machines (CMMs) have been widely used for precision measurement in various fields. Such the conventional CMM uses an XYZ mechanism consisting of three mutually orthogonal slide mechanisms. It appears, however, the machine's accuracy and efficiency are already at their limit due to several of its characteristics: (1) violation of the Abbe's principle, which is the basis of precision measurement; (2) a weak (cantilever) beam structure in which deflection is often generated by minimal bending force; (3) accumulation of measurement errors led from each axis; and (4) low traverse speed due to a mass accumulation. In short, these problems, which limit the precision of the CMM, are inherent in its stacked architecture or serial mechanism.
This study proposes a new coordinate measuring machine[1][2][3] based on a parallel mechanism that avoids the problems above. The use of the parallel mechanism constructed of closed-loop links will potentially improve the stiffness, accuracy, and efficiency of the CMM. This paper discusses the fundamentals, construction, calibration and accuracy test of an experimental CMM.


Figure 1 depicts the proposed CMM. The touch trigger probe attached to the stage is connected to three struts through the revolutionary joints. Each strut is connected to the overhead base through three spherical joints and contains within it a prismatic joint, the length measuring instruments (scales) and actuators to expand and contract itself. Variations in the length of the three struts move the stage in three-dimensional space. When the probe touches the measuring object, the probe's position (coordinate) and attitude can be derived absolutely from the strut lengths. The proposed CMM has a number of advantages over the conventional CMM; (1) because the probe is in sensitive directions of the scales, the motion error of the stage has little effect on the measured value; (2) the truss structure has a high stiffness because its members are subject to very few bending forces; (3) the systematic error produced by each of the scales is averaged with the other two; (4) the small inertial mass enables high moving speed.

Outline of manufactured CMM

The experimental CMM with 3 DOF has been constructed as shown in figure 2 and figure 3. An octahedral truss frame and a granite surface plate are mounted on a vibration-isolation table. The frame supports three spherical joints connected three struts. The struts with the prismatic joints are expanded and contracted by three individual AC servomotors and ball screws. The prismatic joint is guided by four linear ball bearings. Each length variation of the struts is measured by three linear encoders with a nominal accuracy of 0.47 micromenter(p-p) and a resolution of 50nm (Sony, Laser Scale). Figure 4 shows the detail view around the stage. The stage mounting the touch trigger probe (Renishaw TP200, Repeatability: 2s=0.18 micrometer) is connected to the strut ends through three rotational joints. A stroke of the struts is 220mm, then measurement work space is 150X150X150mm approximately. Figure 5 shows the block diagram of the control and measuring system.


This CMM needs the spherical joints and rotational joints with high rotating accuracy within 1 micrometer. The spherical joint employs 1" steel ball for ball bearing (nominal sphericity: 0.7 micrometer), which is friction-welded together shank as shown in figure 6. After welding, the measured sphericity of the ball is less than 0.75 micrometer. A holder made from self-lubrication engineering plastics supports the ball in line contact. Rotating angle range of the spherical joint is about +-20 degrees. Figure 7 shows the detail of the rotational joint connecting the strut and the stage. The rotational joint employs the ball center system using 1/4" steel balls for ball bearings (nominal sphericity: 0.5 micrometer). It is estimated from previous work[4] that runout is less than 0.25 micrometer, and contact stiffness of the ball center system is more than 100 N/micrometer.

Software for measurement

A personal computer holds and reads counter's values when the touch trigger probe contacts the workpiece. The computer, moreover, calculates XYZ coordinates of the probe tip from the above values by solving the simultaneous nonlinear equations. Furthermore, the alignment program defines a work coordinate system in the machine coordinate system when the workpiece is located for any position and any attitude on the surface plate.

Simple calibration

In conventional CMM, the calibration means checking the motion errors and squareness of the XYZ axes. In proposed CMM, identification of kinematic parameters is important to ensure measurement accuracy because disagreement between design parameters and actual parameters causes the measurement error. Dimensions of the stage and the probe, and location of the rotational joints are measured in advance by conventional CMM, because these dimensions are relatively small. The parameters about the location of the spherical joints and the initial length of the struts are adjusted gradually by repetitive operation so that the lengths of the gauge blocks placed on the center of the surface plate are measured correctly in XYZ directions.

Results of accuracy test

Figure 8 shows deflections and standard deviations of measured values before the calibration when measuring the length of the block gauges with several sizes in Z direction. The deflections increase in proportion to the gauge sizes because of measurement space distortion caused by the disagreement before calibration. The standard deviations, however, are less than 0.15 micrometer. This results show that this CMM has high repeatability of the mechanism. Figure 9 shows deflections measured in XYZ directions after calibration. The deflections are independent of the gauge block size, and are less than 10 micrometer. Figure 10 and figure 11 show measurement results of an optical flat with flatness of quarter wavelength (Measured area: 100mmX100mm). Measured profile before calibration is saddle-shaped surface with maximum deflection of 60 micrometer. After calibration, deflection of the measured surface decreases to 5 micrometer. I expect that measurement error will be decreased by more exact calibration because the measured surface is simple and continuous.


Using parallel architecture with 3 DOF has been proposed to develop a more precise CMM. The fundamentals and the detail of the experimental CMM have been described. The calibration and the accuracy test have been performed to obtain the kinematic parameters. Measured deflections were less than 10 micrometer when measuring the gauge blocks in XYZ directions.


[1] T. OIWA: "New coordinate measuring machine featuring a parallel mechanism", Int. J. Japan Soc. Prec. Eng., Vol. 31, No. 3 (Sept. 1997) 232.
[2] T. OIWA: "New coordinate measuring machine using parallel mechanism - fundamentals and kinematics -", J. Japan Soc. Prec. Eng., Vol. 64, No. 12 (1998) 1791 (in Japanese).
[3] T. OIWA, N. KURI and S. BABA: "New coordinate measuring machine using parallel mechanism - link layout design and error analysis -", J. Japan Soc. Prec. Eng., Vol. 65, No. 2 (1999) 288 (in Japanese).
[4] T. OIWA and A. KYUSOJIN: "Development of precise cylindrical grinding by ball centres -contact stiffness between ball and centre hole-", Precision Engineering, Vol. 12, No. 1 (Jan.1990) 37.