4.1. Tool Length Measurement
In the case of actual machining, as the spindle runs for a long time, some thermal deformation occurs in the spindle, which leads to tool length instability [
18]. To avoid the measurement imprecision of the tool length from spindle thermal deformation, the tool length measurement should be implemented under thermally stable conditions.
The thermal stable time region can be determined by monitoring the tool length deviation with a long-term spindle rotation. In this study, we designed three sets of experiments to monitor the tool length deviation, and the experimental conditions are listed in
Table 2. The objective of these experiments was to clarify the thermal stability time of the spindle rotation, which would contribute to the inaccuracy of the tool length value; accordingly, all experiments were operated with spindle idling rotation, without any cutting load or feed motions.
Figure 6 shows the measured results of the tool length for all three sets of experiments. During the experimental process, the tool and tool holder were kept fixed to the spindle, such that the error due to the tool changing accuracy could be assumed to be 0. The experimental value shown in
Figure 6 is the change in the tool length from the value measured at the initial time of each experiment. According to
Figure 6, after approximately 120 min of spindle running, the deviation in the tool length became stable. Thus, as the thermal deformation of the spindle stabilizes after 120 min, the tool length measurement and machining tests should be implemented after 120 min of spindle rotation.
4.2. Experiment with Tool Length Errors
In this study, the influence of the tool length error of the cubic-machining test and the validity of Equation (2) in actual machining were verified. For the machining tests, the tool center point (TCP) control mode with a tool length setting was adopted. This means that the controller controlled the position of the axes based on the tool length parameter. Therefore, the tool length eL was intentionally set to a certain value added to the measured tool length and set to the controller. The given values of eL were 0 and ±10 μm. In addition, to remove the tool length deviation caused during the machining process, the authors measured the tool length individually before machining each zone. It was also confirmed that the change in tool length was approximately 1 μm.
Figure 7 shows the measured machined surface of the cubic machining with different values of tool length error
eL, where the values (unit: mm) attached to each machining zone are the average relative height deviation from the standard machining zone (ZONE I), and the deviation appearing on the machining without tool length error (
eL = 0) was considered to be caused by geometric errors of the machine tool. According to
Figure 7, it can be seen that the average height of the surfaces became higher when the negative tool length error existed.
The influence of tool length error can be evaluated by considering the difference between the average heights of the surfaces with and without tool length errors, because the influence of other factors, such as geometric errors, can be assumed to be constant. Hence, the influence or tool length error can be qualified based on the differences between homologous zones with and without tool length errors, as shown in
Figure 7.
Table 3 shows the relative average height of the surfaces compared with the results without tool length errors. As the tool length error values were set to ±10 μm, according to Equation (2), Δ
h caused by tool length error was assumed to be ±1.3 μm. However, according to
Table 3, there are 1–3 μm differences between the deviations that appear in the actual machining test. The influence of the tool length error on the actual machined accuracy did not agree with the theoretical value. Thus, other factors, such as the tool path for machining, may have affected the results, which will be further evaluated. Nonetheless, it was confirmed that the tool length error directly affected the machined accuracy, and the tendency of the influence of the error was in accordance with the expected.
4.3. Simulation with Tool Length Error
To clarify the influence of the tool length error, the machined accuracy was simulated considering the geometric and tool length errors. It has been confirmed that the positional and angular commands of each axis can also cause machining inaccuracies [
19]. Therefore, simulations were performed based on the calculated position and angle of each axis obtained from the Numerical Control (NC) program used for the actual machining tests. Consequently, it was established that the simulation results were only influenced by geometric and tool length errors.
The simulation process used is shown in
Figure 8. The positions of the X-, Y-, and Z-axes represent the translational feed motion of the spindle, and those of the B- and C-axes indicate the orientation of the work table. According to [
14], the coordinate transformation from the machine coordinate system to the workpiece coordinate transformation was implemented using Equations (3) and (4), respectively:
where
X,
Y, and
Z are the positions of the X-, Y-, and Z-axes, respectively;
tl is the tool length;
PM,t and
PW,t are the homogeneous coordinates of the tool tip points under the machine and workpiece coordinate systems, respectively; and
MC and
MB represent the feed motion of the C-axis and B-axis, respectively, which are described through the D–H matrix as follows:
In addition,
MγBY,
MβBY,
MαBY,
MδzBY,
MδyBY, and
MδxBY are the impact matrices of geometric errors between the B-axis and machine bed, and
MαCB and
MδxCB are the impact matrices of geometric errors between the B- and C-axis, respectively. The definitions of each geometric error are presented in
Table 4 and illustrated in
Figure 9.
Based on the description in
Table 3, the impact matrix of each geometric error is defined by Equations (6)–(11). Geometric errors were identified for the simulations using a ball bar [
20]. The identified geometric errors are listed in
Table 5.
The tool tip position, considering the geometric and tool length errors, can be calculated as mentioned above. To simulate the machined accuracy, the position of the functional point is required. The relationship between the tool tip point and tool functional point in the cubic-machining test is illustrated in
Figure 10. Therefore, the tool functional point can be calculated using Equation (12), where the tool posture
v is calculated using Equation (13), where
r is the radius of the ball-end mill, and
PW,f and
PW,t are the coordinates of the tool functional point and tool tip point in
Figure 10.
The tool functional point indicates the geometry of the simulated machined surface. Hence, the
Z-axis coordinate
zf expresses the height of the machine zones. The relative height deviation due to errors can be calculated using Equation (14), where
is the average value of
zf for ZONE I, and
is the average value of
zf for ZONE n (n = II-1, II-2, etc.).
Figure 11 shows the simulated results of cases with a different tool length error
eL, and the value (unit: mm) attached to the machining zones suggests the height deviation from ZONE I. According to
Figure 11, the average height of the surfaces increased when a negative tool length error existed, similar to the real machined results.
Table 6 shows the influences caused by tool length errors compared with the results without tool length error. According to
Table 6, it can be indicated that the simulation results agree with the formulated value for each machining zone.