ney Available online at www.sciencedirect.com ™ ae ScienceDirect Procedia CIRP 14 2014) 323 ~328 www else comoeateroreds 6th CIRP International Conference on High Performance Cutting, HPC2014 A machining test to evaluate geometric errors of five-axis machine tools with its application to thermal deformation test Soichi Ibaraki"*, Yusuke Ota* “Dep aera Eimer, Kot Unbersty,Kar, Kye 615850, Jpn + compending wb, Tels 81-26 98-26%6; fax 8-1-9836, mal ade anki hye jp Abstract ‘This paper proposes « machining ett calibrate positon pendent geometric eon, of “ere map” of rary axes of Sven machine tool At given ets of angular posidons of roar ste, a simple stat sde~cating Using a stigh end ml performed. BY measuring scone cor ofthe machined test pice, positon and oricalton of war axis average ses actin rs, swell as psitin-dependcat somtric error of rtry axes, cn be numerically identified based onthe machine's Kinematic mode, Puthemare, by epeating the proposed ‘machnig tet consequently, one can quately observe how the pstion andthe onetation of rary axes change with respect the oa! spindle duc to thenmal deformation indaced mainly by tol spindle rotation, Experimental demonstration is presented, (© 2014 Published by Elsevier B.V. Open acess under CC BY-NCND license Selection and peer-review unde responsibilty ofthe Intemational Scieafc Commitee ofthe 6th CIRP Intemational Conference fon High Performance Cuting 1. Introduction Machine toolt with (wo rotary axes to flvrotate a tool andlor a workpiece, i addition o three orthogonal linear exe, ave collectively called fve-axis machine tools. On five-axis ‘machine tools, error motions of each liearrotary axis, as wel, 28 its aligoment (assembly) errs, are accumulated in the positioning error of too relative ta workpiece. For efficient, ‘measurement of these erors, many "indirect" measurement rethodologes, i. schemes fo separately identify each error ‘component from a set of measured tool cence position (TCP) profiles, have been studied (1, 2]. ISO/DIS 10791-1 (3), curently under a revision process in ISOTC 39/SC 2, contains quati-sttic tests with a main interest in calibrating position and orientation errors of rotary axie average lines, ‘The application of the ball bar to roiary axis dynamic measurements hasbeen sted by many researchers [4,5] and ‘snow included in ISO/DIS 107916 [6], also eutently under revision process in ISO/TC 39/SC 2. The Rest [7,8] i also in ISOWDIS 10791-6 [6]. A touchtigger probe ean be applied snalogously to the ‘shase-the-bal! tet (9,10) and commersi probe ‘ased calibration systems are now avaiable. Although itis important 10 evaluate geometric enors of rotary axes by such a non-cutting measurement, ‘ypical Iachine tool users consider more the machine's accttacy when it performs actual machining, Non-cuting tests are sometimes performed wien the machine is "cold" In the ‘machine's normal operating conditions, the spindle rotation, as well as environmental change, may potentially cause significant thermal deformation, In suck» normal operting condition, the machine's geometric errors may be significantly ferent fom those in ‘cleanin, NAS (National Aerospace Standard) 979 [11], Clause 4338.1, deseribes @ cone fastum five-auis machining tes, Since it is only standard well known describing a five-axis machining test, itis widely accepted by many machine tool builders as one of final performance tests. ISOITC 391SC 2 is, currendy discussing the inclusion of ‘the cone frustum machining ttn the revision of ISO 0791-7 (12). Although it gives a good demonstration of the machine's overall machining performance, itis generally difficult to diagnose error sources ffom the measured geometry ofthe finished test, pieve (13.14) ‘The objective of this pepe it to propose new five-ais ‘machining test such that geometric errs of olay axes ean be 2212-8271 © 2014 Published by BseverB.V. Open access under CC BY-NC-ND lcs, Selstion and peer-evew der responsiblity of the International Scent Commitee of the 6 CIRP Intrational oo Mich Poevivonence Cattineau Soick Ibaraki and Yazuke Ota Procedia CIRP 14 (2014) 323-528 separately identified by evaluating the geometric eror of the ‘machined test piece. In (15), apart ofthe authors presented a ‘machining test to identify position and orientation erors ofthe axis average line of two rotaty axes. This paper presents is ‘extension fo the calibration of position-dependent geometric {erors, of “ror map,” of rotary axes. Furthermore, the paper will present its application to the ‘observation of thermal influence on geometri errors of rotary axes, Experimental demonstration will be presented, 2. Proposed machining test ‘This paper considers a five-axis machine configuration with a titling rotary table (riven by B- and C-axes) depicred in Fig. 1. In principle, the basic idea of this paper can be straightforward extended to any five-axis configurations ‘The proposed machining testis illustrated in Fig, 2. At 0°, a square-shaped step is machined by simple side cutting using a staight end mill by driving X- and Y-axes only. The square step is machined at different heights at each ‘combination of C=0, 90, 180, 270° (=1~4) and B=, 90, 90° (1-3), Total 43-12 finish cuts are made. Figure 3 shows an example of the nominal geometry of the finished tes piece (adopted in the experiment in Section 4). The finishing condition must be properly chosen such thatthe influence of tool deflection or surface finish on the geometric measurement ‘becomes sufficiently small. It is recommended to repeat the Finishing with zero radial depth of eut (“zero cu ‘Then, the finished test pece’s peometry is measured. Figure 4 shows an example of probed points. The measurement coordinate system is set up based on the position and the ‘orientation aFthe uppermost step, machined at 3. Identifiation of rotary axes geometric errors 3.1. Geometric error parameters to be identified In ISO 230-1 (16), the axis average line of a rotary axis is defined as "the straight line representing the mean locaton and orientation of iS axis of rotation.” Position and crientation errors ofa rotary axis average line, called location errors in ISO 230-7 [17], are clearly among. the most fundamental error factors inthe five-anis kinematics, Table 1 shows location errors sufficient to describe the kinematics for ‘he configuration in Fig, 1 [16,18]. isto be noted that they only represent ‘average’ position or orientation of a rotay axis. The axis of rotation may change is postion and orientation with its rotation, Such an ‘error mation ean be parameterized by position dependent geometric errors [13], Table 2 shows positionsdependent ‘geometric errors for Beans. 1 js important to note that this paper assumes geometric rors of linear axes are negligibly small compared to those of rotary axes, Many five-ani error ealbration methodotogies, briefly reviewed in Section 1, are based on the measurement ‘of the TCP relative tothe table, and its therefore not possible spetdaeroy me 1g. 2 Papas machines posse i 5 Macind epee geommary (xg) Machined eute — Probe po Fig 4 Probes pins on machines ees exp)Soich Ihara and Yasuhe Ota/Proceda CIRP 14 (2014) 323328, in principle to separate error motions of rotary axes and linear axes, Ertor mations of linear axes must be separately pre- calibrated by conventional measurement (e.g. ISO 10791-1 3p. 3.2, Calewation of posiion/orientaion of each step For each square-shaped step machined at 8, and C), denote the F-th measured postion in the measurement. coordinate system by plij)e%?. Suppose that its nominal position is given by p'UjAeW. Denote the displacement of the ‘machined step from its nominal position by (Ax), AM), Ax(iy)) in X, ¥, and Z. directions. Denote its orientation error by (Aa(i), APG), Acti) around X, Y, and Z axes. They ean be calculated by solving the following minimization problem: coe Rigo COCO o where n'(jA}e%" js unit veetor representing the norma direction to the target surface. This term is needed since a touch-tigger probe is sensitive only to this direction, cesta commen om et EE orem om a oxrecimnececomerces to the one presented in our previous work on probing-based ‘geomettic error calibration [19] 3.3, Caleutaion of geometric errr parameters When the nominal point in the workpiece coordinate system is given by "4" eM, its actual postion, “ge, when ‘here exist location erors in Table 1, is given by: orton ocr 08 o where "thesis the homogencous transformation matix (HTM) representing. the transformation from the workpiece ‘coordinate system tothe machine eoondinae system: So OC "Oy 0 OpChaa Te Choo aT pCLhon (ahem He CLhas CaCO °O)0 Optom DoCheoe MOCO oo where TYTN denotes the HTM representing the translation in X, ¥ o Zor the rotation around X, Y, or Z. ns Sema) Oe) Ee er Potion cr of aap a Fre ‘ca Poston erro Cans vege Y Ene ey Poston eno of ass average ine Foe oy il angler postionng enor of axis Table? Posi epee geomet ewes of Brats Sane Feld Fes Es) Ral enor nas of ae tioond Yad lB) BB an) in, Such a fveaxis kinematic model cane found in numerous references, eg. (2818), "Bbc represents the nominal transformation "go o.coOGo 6 As is presented in [8], the kinematic model (4) can be reuriten as Son nycmoon,cenco conn, conan cooceycenod where 6 (COO Otay ELT, Chgg EGO Chemo TO Ca (COO Oop LDC Chg HIE, Chg TG CLLR (COO Chee EAD Che, (CO ae EO Cpe, Chee Ch CE, (DO Dag CLD} Chg CIE, Chop CHT Che CELTS (COO Chap EO Chae CELT, The relationship of postionorienation errors of each machined step, Ax(,/} to Acti), to locaton errors, can be derived from the formulation above, Recall that each step's position and orientation, Ax(y) to Ady), are measured in refereace to thote of the reference step, machined at By-C 0" In other words, Ax(i/) to c(i) aust be zer0 at 0° This must be taken info account (see (19), ach location error can be identified from Axi) t0 Ae) by applying the least square fit to this model. The algorithin can be straightforwardly extended to the identification of positionsdependent geomet erors in Table 2 (see (19,8)a6 Soich baraki and Yauke Ota Pocedia CIRP 14 2014) 323 — 528 4. Experiment 4d, Test sep ‘The proposed machining testis conducted on a machining center of the configuration in Fig, I. Table 3 shows major ‘machining condition, Figure S shows the test setup. Although itis preferable to measure the finished test pcos’ geometry by a coordinate measuring machine (CMM), on- ‘machine measurement using & touch-tFgger probe was used in this experiment, assuming sufficiently high volumetic aceuraey of machine too’ linear axes, OMP-400 by Renishaw is used (unidirectional repeatability: 0.35 pm (max 2c value ‘with 100-mm stylus), probe sphere: §6 mm (ruby)). The machining center’ positioning resoution is 1 jum on all the Tinear axes; the probe's measurement resolution is also 1 um, With the machine's B- and C-anes fixed at B~C-O, points showm in Fig. 4 are probed. Note that error motions of B- and axes do not influence this measurement 42 Test results Figure 6 shows the measured geomet of four steps on the test pieee's top face, each of which is machined at Bj-0", and C0, 90, 180, 270°, respectively. Figures 6 (a) and (6) respectively show the projection onto the XY and XZ planes 1m Fig, 6, the dots («) represent the nominal probed position, 'Cij.8), and the citcles (0) epresent the measured position, ‘PlijA). The error between measured and nominal positions is ‘magnified 1,000 times. The position and the orientation of each step, represented by Ax(ij)~Ac(), is calculated as shown in Seetion 3.2. In Fig. 6, the painted bold-line square Fepresents calculated position and orientation of each step (those in Fig, 6(a) arecalculated from probed points om side faces of each step, and those in Fi. 6(b) ate ealeulted from ‘probed points onthe bottom fe). From Fig. 6, many intuitive abservations can be made on location errors or eror motions of C-axis (at BO"), 4+ tn Fig. 6(a), the square machined at C--180" is shifted by about 8 jum in X-direction, and +3 ym in Y-direetion ‘This is mostly caused by the postion error of the C-axis average line, Eycand Ey + In Fig. 6(6), the bottom surface of each step is tilted toward both X- and Y-directions, This is mostly caused by the orienaton enor of Caxis to X- and Y-directons, ‘The measured goometty of steps on the test pieces side faces, machined at B=-90° and 90°, can be plvted analogously, but omitted here. ‘Then, location erors, shown in Table 1, are identified as shown in Fig. 7, Figure 8 shows position-dependent geometric errors of Beaxs, shown in Table 2, identified from the measured geometty. While location errors in Fig. 7 represent ‘average’ postion and orientation of rotary axes, Fig. 8 shows how the Beaxis position and orientation vary from their average values with the B-axis angular positon, B, ae as a am, Sa orkiesematriat—Asmiumalloy A852 Rati dephofeue Om ae) ‘cing Bid Airbow Ming dete Downeut 43. Uncertainty analysis The calculation in Section 3 assumes negligibly small cerror motions of linear axes. It is, therefore, practically important to assess the uncertain in identified geometric ‘ror pararscters due to, especially, linear axis error motions Inthe present analysis, other potential contributors, eg. the uneertinty associated with probing and machining process, are regarded relatively small Table 4 shows assessed uncertainty contributors in linear axis error mations, Statistical uncertainty analysis based on the Monte Carlo simslaton, analogous tothe one presented in (20), is applied. The modelling of uncertainty consibution of Tinear axis error motions in [20] is adopted. In Fig. 7, error bers represent the standard uncertainty (=) calculated from ‘uncertainty contibuters in Table 4, 5. Application to thermal test 5.1. Objective For five-axis machines, even when thermal deformation, typically caused by heat generation fom spindle rotation (possibly also from linea/otary axes or environment), causes simple trinslational errs it offen changes the postion of rotary exes with respect to the machine coordinate system defined by linear axes. As a result, thermal deformation more likely causes the machined workpiece’s geometric erors in ‘ive-axis machining. Thermal deformation may also influence rury axis position orientation directly, ISO 230-3:2007 [21] only desenibes tess to investigate thermal inuence of spindleSoich Ihara and Yasuhe Ota/Proceda CIRP 14 (2014) 323328, an rotation, linear drives, and environment. The importance of thermal tess for rotary axes as been discussed only lately in the literature. Recent works include the application of R-test (22,28, 24], By repeating the proposed machining est consequently on the same machine, one can evaluate thermal influence on positon and orientation of rotary axes. Ths section presents such an application of the proposed machining test. 5.2 Test procedure ‘The test procedure is described as follows: |, Mount the unmachined workpiece on the machine's work table. Perform spindle warm-up (e.g. 30 min in our tes) 2. Rough cutting (about 25 min in our test). 5. Finishing as described in Setion 2 (about 25 min), 44 Dismount he finished test pices, and repeat the procedure {to 3 (otal thre times in ou et) 5. After the machine suffieiently evo's down, the three finished test pieces are measured on the machine as presented in Section 2 5.3 Test results Figure 9 shows postion and orientation errors of B- and CC-axis average lines (location erors, Teble 1) identified by the geometry of each of three finished test pieces. Notable gradual change can be observed in the Y-posiion of the C- axis average Tine, Ec, and the Z-position of the B-axis average line, Egg. The variation in other eror parameters is smaller. This likely indicates the thermal deformation of the spindleside- machine structure mainly toward Z- and Y- directions (see the machine configuration in Fig.1). Such a simple linear deformation (at ol tip) causes geometric errs of fnished test piece infive-axis machining. “To validate location erors estimated ‘om the machined test pieces, the position errors of C-axis average line, Eric (Ene Enonc) and Eyye, were directly measured by using a dial gauge, atached to the spindle, and a test picce fixed on the (able. This measurement was conducted right afler the machining of each test. piece was finished. Figure 10 ‘compares the estimates and direct measurements, The validity of the estimates can be observed. 6. Conclusion In the proposed machining test, a simple square-shaped stop is finshed by a straight (eadius) end mill at piven sets of 1B and C-axis angular positions. By examining the position and the orientation of each step on te finished test piece, peometric error parameters of B- and C-anes, both location ‘rors and positon-dependent eros, can be identified. ‘By repeating the proposed machining ts, ime-dependent variation in geometric error parameters ean be observed, ‘which is mostiy caused by thermal deformation induced by spindle rotation or environmental change. In the machine configuration in Fig. the heat from tool spindle rotation mainly displaces the too spindle. This deformation changes the relative position (and possibly the orientation) of rotary faxes tothe tool spindle, In fve-axis machining, from our 8 Machined at 8-0", -2705 10, 39 “hoa 0h | 80120 Ximm Vo:prbed pont Rejoronce step machined nominal point Erorseale: 5 um_ Reloroncesepmachined a0-¢-0" i Paste i 6 Mead esmty ofthe machin! et piss (or apace t B07 abd 9, 180,270) rojo eto XY ple, projston nt 2 plane ig, Poston an renin of 8- a Cas avenge ies ocaion evr see Tablet y test ooh et. rr bts Teese ‘Me sadn unceaty(er) elated Hom nce consti experience, thermally-induced variation in rotary axis location eors is often among the most critical factors for potentially significant geometric errors of finished workpiece. ‘The ‘proposed machining test is effective to evaluate the machine's ‘thermal stability ‘The machine's geometric errors are sometimes evaluated when the machine is ‘cold’ (eg, static tests [3] at assembly. inspection sites in machine too) builders). Such a test isae Soich baraki and Yauke Ota Pocedia CIRP 14 2014) 323 — 528 clearly not good when the machine may be under significant ‘thermal i‘luence. The proposed machining testis effective to evaluate geometric errors when the machine i in normal ‘operating conditions. References ‘zachiss "Ab ute CIRP Anas ManTech 2008372 07S, Sade renal Ze) . Tsuisumi Mt, Sa A. Meiicwion ahd compeesation of systematic Gemeie Uo 847, ete fr tine t= Pr 7 Osa 8}inasat | Kishi K,Sokamot S, Sugimura N, Takeuchi T, Tana Sloper genres tery of rachel he bs sod appa, so: Yoke, 1997 Go opanese) iar 8, aT, Masha Cabin ofan omar f rata tsecbenaper pre I Machine Tele an! Manat 22,3 2u)biggmane B, Bouche JP Robe, Sytentc eration of eratin 1) 60 280-5007 Tet coder uci te Pat 3 Deron of (22} Hong C, thor S, Observation of Thera Influence on Ear Maton of tary Aacs oi Fives Masog Ta! by Se Ren a ‘Aeation Tecnology 2012 62. 196-208 23 Geant Mt, Cate. Kopp, Wegener, Meseent ehs ane r2Gebhade M,Cappael'S Ex Kapp W, Wegener K, Phys and See nc Fai) Tale 4. Asomed snes contele near ai err motos Nensyetemaiccree | (somal tsar) Ta 20 ms Ave EROS ECS EAOBE Fig 9. Pxtion an orton of and axis vege ines Hetifeby € [EWersiccer & —-EaWoreiece g werisiece| 54 J. Ba0 Td | | EXOG EVO pines a) thn ily tarred yong