Bi Arcs
Bi Arcs
1007/s00170-005-0290-8
Received: 8 November 2004 / Accepted: 4 July 2005 / Published online: 5 July 2006 # Springer-Verlag London Limited 2006
Abstract With the rapid development of the information/ image system and aero-space industries, high quality optic aspheric surface lenses play an increasingly important role for completion of the functionalities. Aspheric lenses are non-spherical surfaces having rotation symmetry about the lens axis. The aspheric lens has various shapes according to its application and often requires tens nanometer order form accuracy since surface roughness and form accuracy play essential roles in the functional performance of the optical products. Interpolation of the aspherical surface path must precisely meet the allowable tolerance. Linear interpolation of the aspheric surface path for CNC machining generates an enormous amount of NC code to satisfy the extremely small tolerance, and produces scallops on the machined surface due to the acceleration and deceleration of the tool during every linear motion. Alternatively, interpolations with bi-arcs are used. In this paper, in order to minimize the error induced by the cutting tool path and to shorten the calculation time of interpolation, a precise -arc interpolation method is proposed. The developed algorithm of bi-arc interpolation meets the given tolerance precisely. This is guaranteed by an analytical proof and error maps. Another advantage is its ability to
calculate about five times faster than the existing arc interpolation, since iterative calculations for the maximum error can be omitted. The developed algorithm has been used for the precise aspheric machining. Keywords Aspheric surface . Interpolation . Bi-arc . Tool-path
1 Introduction
Nowadays, with the rapid development of consumer electronics, information/image systems and the aero-space industry, the demand for aspheric lenses increases drastically. With the development of technology, high quality optic aspheric surface lenses play an increasingly important role for completion of the functionalities in many instruments [1]; especially applicable are parts applied in spaceflight and military equipment. Aspheric lenses are non-spherical surfaces having rotation symmetry about the lens axis. Manufacturing of aspheric parts involves shaping the surface by generating a closest fit aspherical curve by diamond cutting. For manufacturing aspheric parts, high accuracy is required for machining. The aspheric lens has various shapes according to its application and often requires tens nanometer order form accuracy since surface roughness and form accuracy play essential roles in the functional performance of the optical products. Traditionally, these surfaces are generated by grinding followed by polishing. The quality of the lens surface generated relies largely on craftsmanship and is assured by the trial-and-error approach. In recent years, the manufacture of aspheric lens surfaces by diamond turning is most economical for low quantity production. However, it is still very difficult to machine aspheric lenses with high accuracy. To obtain precision products, the error caused by the path of the cutting tool, influence of the cutting force, inaccuracy of the feed drive system, wear of the tool, tool decentering, etc. must be minimized [2]. In ultra-precision
T.-M. Lee Korea Institute of Machinery and Materials (KIMM), P.O. Box 10, 1 Yusong-gu, Daejon, 305-343, South Korea e-mail: taikmin@kimm.re.kr E.-k. Lee (*) Division of Mechanical & Automotive Engineering, Kongju National University, 182 Shingwan-dong, Kongju, Chungnam, 314-701, South Korea e-mail: eklee@kongju.ac.kr M.-y. Yang Korea Advanced Institute of Science and Technology, 373-1 Kusong-dong, Yusong-gu, Daejon, 305-701, South Korea e-mail: myyang@kaist.ac.kr
1192
machining, the form accuracy of the machined surface is seriously affected by the residual form errors [3, 4]. The control modules of numerically controlled manufacturing systems usually have a straight-line and circulararc interpolation processor. In general, CNC machine tools carry out linear, arc, and spline interpolating tool paths, such that the path for aspheric surfaces, expressed by a complicated equation, has to be interpolated. Interpolation of the aspherical surface path must precisely meet the allowable tolerance. Linear interpolation of the aspheric surface path for CNC machining generates an enormous amount of NC code to satisfy the extremely small tolerance, and produces cutting marks on the machined surface due to the acceleration and deceleration of the tool during every linear motion. Some NC processors support spline interpolation. The usual method of spline interpolation is referred to as a cubic spline. In this method, the curve segments between successive data points are represented by cubic equations, and the smoothness of the curve is obtained by matching the first and second derivatives of each segment across all the intermediate data points. This method of curve interpolation evolved naturally because it is almost the mathematical analogue of the practical method of drawing curves by using ducks and a physical spline as a thin elastic beam. A cubic equation is the lowest order which permits continuity to the second derivative between each segment. Higher order equations can be chosen to represent each curve segment but these are more complex to solve and more likely to produce unwanted inflections. Thus, the spline interpolation is also not suitable to ultra-precision, because the error of interpolation is calculated not in an analytical manner but in an iterative manner and thus requires more computation time. A further disadvantage is that the intersection of a spline with a line, a circle or another spline is non-analytic and has to be obtained by iteration. A circular arc interpolation is an alternative solution for machining a precise surface. The simple circular arc fitting curve, without consideration of the tangents of the original shape, may result in large changes of tangent direction and in discontinuities in curvature. Additionally, it makes the first differential values continuous at the connection points of each arc, which eliminates scallops. Also, arc interpolation of an aspheric surface path can be solved analytically, and thus requires much less calculation time than spline interpolation. Also, the arc interpolation method generates less NC codes than those of linear interpolation. Nevertheless, the existing arc interpolation still requires a large amount of calculation to meet the tolerance exactly. Alternatively, numerically expensive approximation methods or interpolations with bi-arcs are used. The word bi means the number two. In the bi-arc curve fitting method, two circular arcs are fitted between any two points along the given curve. To form the two arc circles, six unknown variables have to be determined. To generate two mutually tangential arcs at the arcs-connecting point, the five conditions should be fulfilled by the two circles. Thus, the bi-arc interpolation curve consists of mutually tangential circular arcs constructed to ensure continuity of the first
derivative; but, the second and higher derivatives are discontinuous. The final requirement in addition to the original five is to have influence on the characteristic of arc interpolation. In this paper, for precise and rapid calculation of the error between the interpolated arc and the aspheric surface, a new bi-arc interpolation method was developed. The novel bi-arc interpolation method, in which a segment span is composed of two circular arcs and the location of the maximum interpolation error is fixed at an optimal position, is proposed to minimize the error induced by the path of the cutting tool and to shorten the calculation time of interpolation. As a result of this requirement, the developed new bi-arc meets the given tolerance precisely. The developed bi-arc curve interpolating algorithm has been used within the CAM system for the ultra-precision aspheric machining. Consequently, the new bi-arc interpolation method can be expected to improve the precision and productivity of aspheric surfaces. With the developed interpolation method, machining experiments of the aspheric surfaces have been performed by the diamond turning system.
1193
t1
P1
P2 t2
In the present paper, a precise bi-arc interpolation method is proposed to minimize the error induced by the path interpolation and to shorten the calculating time of interpolation. The new method fixes the location of the maximum interpolation error at a point on a real curve for being able to calculate the exact error.
In the new bi-arc interpolation method, the maximum fitting error is to be located at an optimal position for accurate and efficient bi-arc interpolation. Generally, in a bi-arc interpolation, the interpolation error between the original curve and the generated arc at an arbitrary position of the real curve is analyzed as follows. On the assumption that the origin is the point O1 in Fig. 2, the error between Pa and Po at an arbitrary position can be calculated by Eq. 1. error O1 Pa O1 Po (1)
t1 t2
P1
P2
Considering the arc radius and coordinates of the points, Eq. 1 can be converted into Eq. 2 p (2) jerrorj r1 x2 f 2 x The maximum error should be the extreme value of the error in Eq. 2. Thus, the extreme value satisfies Eq. 4. x f xf 0 x jerrorj0 p x2 f 2 x x f xf 0 x 0 (3)
(4)
Generally, other existing bi-arc interpolation methods have the following drawbacks. One is that the location of maximum error is not considered for getting the exact interpolation error (Fig. 1a), thus a large amount of calculation is required. To overcome the weak point, the new algorithm of bi-arc interpolation begins with locating the maximum error on a known position and results in satisfying the allowable tolerance exactly. The other drawback is inefficiency of the algorithm. If the radius of curvature of the real curve increases along the path, as shown in Fig. 1, the radius of the first arc should be sufficiently small and the second sufficiently large, as shown in Fig. 1b, to decrease the maximum error value. However, existing bi-arc methods do not consider this as resulting in a greater maximum error.
Pa
E max s1
p3 p
t3 s2
p4 s3
t1
Po
p1 pc r1(r)2
p2 t2
1 2
o1(o)2 o3
3
r3
Generated arc Real curve
Fig. 2 Geometric configuration of the new bi-arc divided into three arcs
1194
(5)
p l p
cv
23
In Eq. 5, f xx means the slope of O1Pa. The extreme exists at a position where the tangent vector at a point on the original curve is perpendicular to the radial direction from the corresponding point on the generated arc, in other words, where the normal vector at a point on the original curve passes through the center of the generated arc. Therefore, at junction P3, the interpolation error is the maximum value of the error and knot P1 has the minimum value of that in Fig. 2. These results are the same as those that will be generated on the right side. From the maximum error analysis, to ensure the maximum error point at the optimal error position, the arc of the new bi-arc, which contains the maximum error point, is divided into two arcs at the maximum error point. Thus, the new bi-arc temporarily consists of three arcs (the first S1, the second S2, and the third S3) to consider the location of the maximum error (Fig. 2). To support this hypothesis, the below nine constraints should be fulfilled to generate three arcs for interpolating a given curve segment. 1. The first arc (S1) must pass through the point P1. 2. The first arc (S1) must have tangent vector t1 at the point P1. 3. The second arc (S2) must have tangent vector parallel with the chord P1P2 at point P3. 4. The junction point P3 between the first arc (S1) and the second arc (S2) must be on the line of the normal direction of the point Pv, which is the farthest point of the given curve from the chord P1P2. 5. The two arcs of the first (S1) and the second (S2) must have a point of common tangency at the point P3. 6. The two arcs of the second (S2) and the third (S3) must have a point of common tangency at the point P4. 7. The third arc (S3) must pass through the point P2. 8. The third arc (S3) must have tangent vector t2 at the point P2. 9. The radius (r2) of the second arc (S2) is equal to the radius (r1) of the first arc (S1). As shown in the above nine conditions, the first arc (S1) is generated from the start knot point (P1) to the position (P3) where the maximum error occurs. If a junction point P3 exists on the normal vector at the point Pv , which is the farthest point from the chord P1P2, and its tangential direction runs parallel with the chord P1P2, then the maximum error (Emax) exists at the junction point P3. For generating three arcs with the nine conditions, a geometric configuration is introduced for using the sine law (Fig. 3). In Fig. 3, there are some variables which include l12, lcv, 1, and c2v that are known, thus r1, r2, r3, l23, 2, 3, c23
c2v
c23
12
should be calculated for the new bi-arc interpolation. By applying the sine law to P2P3Pv, we get Eq. 6. l23 lcv tan c2v cos c23 (6)
Equation 7 is calculated considering O1P1Pc and P2P3Pc, and Eq. 8 considering P1P3Pc and P2P3Pc. l12 r1 sin 1 l23 cos c23 l23 sin c23 1 cos 1 (7)
r1
(8)
With Eqs. 6, 7, 8, Eq. 9 is calculated. 1 cos 1 tan c2v l12 1 tan c23 sin 1 lcv
(9)
With Eq. 9, the maximum error can be calculated as shown in Eq. 10. Emax l23 sin c23 lcv (10)
The end point (p2) of the new bi-arc interpolation will be positioned for the maximum error of machining allowance. After positioning the point (p2), a and b can be calculated by considering geometrical relationships (Fig. 4). With a and b known, the unknown r3, 2, and 3 can be calculated as shown in the following procedure. In Fig. 4, the relationship among P3P2, P3QC and QCP2 is P3 P2 P3 Qc Qc P2 . Therefore, l23 r2 sin 2 r3 sin 3 cos a 2 cos 3 b (11)
1195
t2
p3
2 a
s2 p 4 Qb Qc Qa s3
Emax
s 1+ s 2
pv
3 b
t1 p1 r1
s3
p2 t2
r2 pc
p2
1+ 2
o1
3
r3
t3
Generated arc
3
r3
o3
Real curve
Fig. 5 Geometric configurations for the new bi-arc (three arcs are combined into two arcs)
o2
Generated arc Real curve
And, in Fig. 4, O3 P4 O3 Qa Qa Qc Qc Qb Qb P4 . Therefore r3 r3 cos 3 r3 sin 3 tan 3 b r2 sin 2 tan a 2 r2 1 cos 2 (13)
o3
Fig. 4 Detailed diagram of the second and third arcs in the precise bi-arc
Substituting Eqs. 11 and 12 into Eq. 13 will result in Eq. 14. r3 r2 1 cos 2 l23 sin a 2 1 cos 3 (14)
With Eqs. 11, 12 and 14, Eq. 15 is formed. r2 fsin a b cos 2 cos a b sin 2 sin 2 sin a b g l23 fcos a cos 2 sin a sin 2 cos b g (15)
Equation 16 is then the simplified form of Eq. 15. K1 sin 2 K2 cos 2 K (16)
K1 r2 f1 cos a b g l23 sin a K2 r2 sin a b l23 cos a where, K r2 sin a b l23 cos b K In Eq. 16, 2 sin1 p where tan1 2 2
K1 K2
by iteration
K2 K1
According to the above process, three arcs are generated between two knots as shown in Fig. 2, and the first and second arcs of identical radii are combined into one to complete the new bi-arc interpolation (Fig. 5). The advantage of the new bi-arc method is that the maximum error can be calculated without iterative error calculations of general arc interpolation methods. Therefore, it is easy to make the error meet the allowable tolerance by controlling the terminal knot P2 (Fig. 6).
Calculate
r1 E max N
Calculate
1196
Fig. 7 Error 2 map 2 between the new bi-arc and various aspheric X Y X2 Y2 profiles. a 502 402 1; tolerance 1; 000 nm: b 502 402 1; 2 2 X Y tolerance 10 nm: c 502 302 1; tolerance X2 Y2 10 nm: d 502 302 1; tolerance 1 nm
images in computer graphics [7, 8]. In order to display a graphic image on a raster device, a frame buffer is used [9]. A frame buffer is organized as an X-Y matrix of memory locations, and each memory location corresponding to a pixel of display screen contains color data to be displayed. Instead of color data, z-map data contains a corresponding height (z) value of the surface at each frame buffer location. For verification of the new bi-arc interpolation method, the z-map method is used in this paper. Since the shape of an optical lens is symmetric, a 2-dimensional z-map is suitable for simulation of aseptic surface machining. For geometric simulation using the z-map method, at each end of tool motion, the tool swept-area is generated and converted into z-map data, which is compared to the original z-map data. The comparison is used to update the geometric change of the workpiece after the tool motion. Zmap is an efficient method in that once it is constructed the cross-sectional display, transformation and point evaluation are easily and rapidly achieved. As shown in Fig. 7, the z-axis of the z-map is perpendicular to the tool axis direction and the origin is the workpiece center. The boundary encloses the workpiece boundary. The size of each mesh () is determined from the z-map boundary area and the number of elements meshed. The z-value of the workpiece shape is determined by calculating the intersection point between the workpiece surface and the line passing through the mesh center point. The method for calculating intersection between a line and any kind of surface appears in some texts [10, 11]. The zvalue is used to verify the CL data and check the geometric information after the tool motion. The interpolation error maps are presented in Fig. 7. The interpolation errors are calculated by interpolating elliptic curves (Fig. 7a,b) and hyperbolic curves (Fig. 7c,d). The error calculations are specified so as to verify the developed method to be able to meet the given machining tolerance. Visual verification of the interpolation path is done by displaying the z-map data of the machining error. As shown in Fig. 7, the error maps for various aspheric profiles show that new bi-arcs meet the given tolerance precisely.
Fig. 8 Machining examples of aspheric surfaces using the new biarc interpolation
1197
The new bi-arc interpolation technique is applied in NC machining for experiments. Figure 8 is an example of convex and concave aspheric surfaces machined by an ultra-precision machine (surface roughness: Ra 11 nm).
References
1. Chen MJ, Li D, Dong S (2002) Research on a large depth-todiameter ratio ultra-precision aspheric griding system. J Mater Process Technol 129:9195 2. Minaev AI (1991) Methods of increasing the finishing accuracy of specular surfaces of optical elements by diamond turning. Optical Society of America, Washington, DC, pp 501504 3. Kim SW, Walker D, Brooks D (2003) Active profiling and polishing for efficient control of material removal from large precision surfaces with moderate asphericity. Mechatronics 13:295312 4. Lee WB, Cheung CF, Chiu WM, Leung TP (2000) An investigation of residual form error compensation in the ultra precision machining of aspheric surfaces. J Mater Process Technol 99:129134 5. Bolton KM (1975) Bi-arc curves. Comput Aided Design 7:8992 6. Parkinson DB, Moreton DN (1991) Optimal bi-arc-curve fitting. Comput Aided Design 23:411419 7. Takata S, Tsai MD, Inui M, Sata TA (1989) Cutting simulation system for machinability evaluation using a workpiece model. Annals CIRP 38:417420 8. Takeuchi Y, Sakamoto M, Abe Y, Orita R (1989) Development of a personal CAD/CAM system for mold manufacturing based on solid modeling technique. Annals CIRP 38:429432 9. Newman WM, Sproull RF (1979) Principles of interactive computer graphics, 2nd edn. McGraw-Hill, New York 10. Mortenson ME (1985) Geometric modeling. Wiley, New York 11. Faux ID, Pratt MJ (1981) Computational geometry for design and manufacture. Ellis Horwood, UK
5 Conclusion
In this paper, the more accurate and efficient bi-arc algorithm is developed. The main advantage of the new bi-arc method is that the maximum fitting error can be calculated without error-finding iterations as used in other arc interpolation methods. The new bi-arc interpolation method has the following characteristics. The new bi-arc in which the location of the maximum interpolation error is fixed meets the allowable tolerance precisely and fast. The new bi-arc generates less NC code blocks (about 60%) in comparison with existing bi-arcs due to the efficient location of the maximum error. For verification of machining accuracy, the interpolation errors are calculated for various aspheric profiles. The error calculation result shows that new bi-arc meets the given tolerance precisely. The proposed new bi-arc can be expected to improve the precision and productivity of aspheric surfaces.