Testing Rotary Axes on NC Machine Tools W. Knapp - Submitted by A. Wirtz (1) Received on January 15,1990
AbCtrP_Et Tho pnper denls with tho tooting of rotnry axoa on NC machine tools (machlning contros). Tho rotnry axia is chocked in combination with other aachlne axos no that not only tho individual poomotric errorm but alno tho porfornance of tho numorical control can bo checked. Tho now tost proceduro is baaed on tho circular teat for tho linoar axon of NC machino tools. Tho now procedure in doacribod in dotail. First monnuremont results aro preoentod and discussod In the pnpor.
KgX- WSSdl machino tool, machining contro. NC rotary table, rccoptance test, circular tost for fourth axia. 1
On rotnry nxos thore sro six goometrical errors: nngular positioning orror, two radial movononta, one axinl movomont and two rooling movoments. A rotary nxis on a machino tool hna a norinal ponition nnd a nominal oriontation, thus givinp five additionnl error sources: two orrors of position and throo errors of oriontation. Tho third oriontation is tho orientntion of tho anglo that oquals zoro; if tho zero anglo is dofinod 0.8. by tho workpioco this orror haa no influonco on tho porforamnco of tho machino. Thin in truo for many machines if tho rotary axia is usod as a fourth axia. Tho third orror of orientntion ray not be noploctod on machine. with fivo and more nxis.
Rotary 1x0s of NC machino tools nnd machining contros pro normally chocked with tho help of din1 paugon, LVDTs and procinion cylindorn (for chocking radinl and reeling aovomontn), optical polygona and autocollimators (for checking nngular positioning orrorn). Somotiron tho nngular poaitioninp error is checkod by comparative moasuremontn with a procision indoxinp tablo. Error8 of position and oriontation of tho axis aro chocked normally with dial paupoa, LVDTs, ond bnra and precision cylindors.
At tho bepinning of the moasuronent the block with tho aurfaco of contnct is mochanically ndjunted to nchiovo n defloction of the 2D probe. Thon tho circular and tho rotational aovoaentm are startod by tho NC program. If tho machino han no orrorn tho size of tho doflection koepa constant, only the diroction of dofloction changes by 3600 during tho circular roverents. This producon a porfect circlo on nn analogous XY plottor thnt plot5 tho doflection u g n i f i o d and simultanoOUBly. If tho machine door have orrors the dofloction chnnges and tho plotted dovintions from a circlo.
sizo of the line shows
At the bepinning of tho monnuremont tho 2D probo ia contrad to tho rotary axis by contacting En ondbar in four positionn. Tho ondbnr is rotnted by tho rotnry axis t o the four positions (Be, 9 0 0 , 1800, 2700).
A0 NC rotnry nxos aro ofton usod in contouring rod., i.e. in combination with tho linoar axos of the anchining centre, on. should chock tho influonco Of difforent faodraton. This csn bo onsily done for tho axinl and radial movoronts nnd for the roeling DOVOaonts, but not for tho anpular positioninp orrors ns a11 tho coamon mothods for chocking tho angular positioninp orror are static and non-continuous methods.
Anothor source of orrors on machining centrea 1s a misratching of combinod rovomonts of linoar axoa and rotnry nxin 0 . 8 . whon rilling in contouring rod.. Tho chock of combinod movomontn of two linonr axon with tho circular toat nhows a larpo varioty of orrors ( 1 ) so that a continoun chock of tho combinntion of linonr and rotary ax08 mipht rover1 anothor group of errros. 2
Tho now method is baaod on tho circular toat for tosting linenr 1x0s of NC rAchino; ( 1 ) . Thoro aro rany possiblo not-ups with tho now method (2). Wo only want to doscribo a net-up that is quit. suitablo for analyning tho rosulta, i.0. for aoparatinp individual error corpononts. A 2D probo (2D 1 two dironaional) is mountod on tho non-rotatinp part of the machino, a block with s aurfnce of contact is rountod on that part of tho machino that in rotatod by tho rotary axis ( F i g . 1). If tho 2D probe is mountod in tho tool npindlo the spindlo rust bo fixed as tho spindle should not be chocked but any other rotnry axis of tho machino. The nurfnce of contnct 1s 0.8. n plane nurfnce.
Tho mschining contro i n programrod in such n way thnt tho 2D probo moves on a circlo rolativ t o tho rotary axis of tho archino. This circular rovoront is a pure trannlatory rovoaont producod by two linoar nxos of the n c h i n e .
Fig. 1 Measuring set-up for checking NC rotnry ax08 on machining centrea (from (2)) I...anchine bed, 2. ..workpioco table, 3...tool holdor, 4...rotary axis, S...probe tip (sphere), 6...2D probo, 7...block, 8...surface of contnct
Tho movomont of the rotary axis is p r o g r a u e d so that the surfaco of contact movos nirultanoounly with the circular movemont of tho 2D probo.
Annak of the ClRP Vd. S/l/lCW
If the aurface of contact is a plan. and is adjunted tangontial to the nominal circular path (Fig. 2 ) radinl deviations can be chocked. Tangential deviations as caused by angular positioning errros do not influence the defloction of the 2D probe. If the moasuremonts according to Fig. 2 are carried out at a radius equal zero only the radial movements of the rotary axis are chocked am the linoar axen do not move. If the moasuremont is ropoated at an offnet position (offset parallel to the rotary axis) the radial movements may change due t o reeling movements. A change of the contering of the moasured circle Indicatos an oriontational error of the rotary axis (3).
In a second step the moasuremonts - still according to Fig. 2 - are carried out at a rndius larger than zero. This leads to a combined chock of the radial movement of the rotary axis and tho movomont of tho linear axes. As tho radial error. of tho rotary axis are known from the first aot-up (at radiua equal zero) the change of the plotted path shows tho orrora of the linear axes. Those chanpes are evaluated aa common circulnr tosta ( 1 ) with one excoption: there Is no nominal diameter of the circular path 80 that only differencos betwoen the positioning errors of tho two linear errors can bo evaluated. In a third step the plane of contact is arranged radially t o the nominal circlo as shown in Fig. 3. Radial movomonts of the rotary axim have no influence on the deflection of the 2D probe. Angular position orrorn influenco the deflection directly. Tho doflections of the 2D probe are drawn radially on tho XY plotter and thus change the radius of the plottod path. This set-up correnponds mathematically a tangontial to radial convorsion without any cnlculationa. The set-up according to Fig. 3 is alno influenced by goomotric errors of the linear axos. Firat calculations show that different positioning errors of tho linonr axon produce an ollipae with ita principal axes at 4 5 O t o the machine axes, a perpendicularity orror between tho two linear axes produces an ellipse with it0 principal axen para1101 to the mnchino axes, and PO on (soe also (3)).
2 Arrangement o f surface of radial deviations (from ( 2 ) ) .
B...centre of circular movement and of rotation, 7. ..block, 8,EB contact surface, 5,CB ...p rob. tip, C1,Cl’ ...p ath of point of contact (Tb), C2,CZ’ ...p ath of centre of probe tip ( 2 1 , A,B axes of menaurement of 2D probo, &...radial deviation, oituation at different angles ( a ) , PB,Pl,PZ r...radius of circle
The geometric errors of tho linear axes are known from 2 at a radiun the second set-up (according to F i g . larpor than zero), their influonce t o the third set-up
Fig,--i nensuremont of circle equal %Pro.
radial movements at
1. ..counter-clockwino contouring, two meaauromonts, P...clockwise contouring, two reanuroments (accordinp to Fig. 3) can bo calculated and subtrncted. With this procedure the puro influence of angular positioning errors can be calculated.
Elg.-3 Arrangement of contact surface to check tangential ieviations (from (2)). @...centre of circular movomont and of rotation, 8,EB contact surface, 5,CB probe tip, 7...block, C1,Cl’ ...p ath of point of contact (TB), C2,CZ’ path ( 2 ) A,B.. .axon of measurement of centre of probe tip of 2D prob8, o...trngential deviation, PB,Pl,PZ situation at different angles ( a ) , r...radius of circle
Fig. 4 ahows the radial movements of tho rotary axin (reasureront set-up according t o Fig. 2, radius oquals zoro). Betweon clockwise and counter-clockwise contouring ( 1 and 2) a difference of up t o 5 ,’a can bo seen. The difference between two successive rotations in the same diroction (clockwise or counter-clockwise) is smaller thAn 1 p u . The deviations from a circle aro alao small PO that these errorn can be negloctod whon chocking at a radius larger than zero. By remounting the plane of contact at a new orientation that differs by 9 0 O to the first oriontation tho influenco of radial moverents can he checkod coupletely. Tho results on that machine a practically tho adme (3).
Fin. 5 Mennurerent of radial moverents and of linear axes at circle radius equal 120 mi, clockwise. l...contouring 2...contouring
speed 500 mm/min, two measurements, speed 1700 mm/mln
Fig. 5 shows the reault with the second set-up (acThis corcording to Fig. 2, radius equals 120 m m ) . responds to a pure circular test for the two linear axes sa the radial movements of the rotary axin are neglectable in comparison with the errors of the linear axea. The two linear axea show backlash (15 and 10 p i ) and have different following errors as the influence of a changed contouring speed shows (l...at 500 m d m i n , 2...at 1700 mm/min contouring speed). Geometric errors as different linear positioning errors and non-perpendicularity are relatively small with smaller than 5 p m and, respectively smaller than 10 pm/240 mm (description of evaluation of circular tents see (1)).
Fig,-& Measurement of tangential deviations, radius of mrl clockwiae contouring, two circle equals 120 measurements, contouring speed 500 nm/min. upper: result a s obtained on the analogous XY plotter lower: indication of periodicity superimposed on original plot
Fig. 6 shows the result with the third set-up (according to Fig. 3). In large contrast to Fig. 5 the circular deviations are no longer smooth but change rapidly within about 7 pm. These changes are caused by angular positioning errors of the rotary axis. These 7 pn correspond t o 7 pm/120 mm u 12 seconds! of arc (120 ma is the rsdius of the circular movement). Theme deviations are repeatable, all measurerents are taken and drawn twice.
Looking once more at Fig. 6 one recognizes a periodicity in the radlal deviations. If the periods are marked as done in the lower picture of Fig. 6 w e recognize 30 periods. These thirty periods correspond to the number of teeth of the worm-wheel of the rotary axin, in this case of a rotary table. The deviations within one period show how smooth the worm gear works. Figs. 7 and 8 show the influence of contouring speeds to angular positioning errors. By increasing the contouring speed from 500 to 1700 mm/min the difference between the circular and the rotative moveaent increases, counter-clockwine t o amaller dinmeters (Fig. 7), clockwise to larger diameters (Fig. 8 ) . The difference at these two speeds is about 15 pn which corresponda to 25 seconds of arc (radius of circle = 120 mm), 1.e. the mean sngular positioning error changes by 25". This error may be due to a following error of the rotary axis, s following error that in not adjusted to the following errors of the linoar axes. This adjustment is especially complicated as the correct adjusted following error of the rotary axis depends on the radius of contouring. Nevertheless, when milling in contouring mode using these three axes (two linear, one rotary axis) these errors will be transferred to the workpiece.
Fig, 7 Measurement of tangential deviations, radius of circle equals 120 mm, counter-clockwise contouring. l...contouring speed 500 nm/min 2. ..contouring speed 1700 am/min
The difference between the mean diameters of clockwiae and counter-clockwise contoured tests at the name contouring apeed shows the "dynamic" backlash of the rotary axis. We see that the dynamic backlaah depend. on the contouring speed (comparison betweurr F i g s . 7 and 8 ) . At 500 mm/min the radial difference i n about 10 pa (or 17 seconds of arc)1 at 1700 mm/mln about 35 pa (or 58 ( ! ) seconds of arc). This error, too, will influence the workpiece.
The firnt results nhow tho possibilitio8 of the new method, the circulnr tort for rotnry axes, t o show the behaviour of angulnr ponitioning errorn in a continuous form. This nakon it ponniblo t o realize tho smoothnosn of tho rotntional rovemont and to ronlize effects of tho numerical control nn influenco of contouring speed and dynnmic backlash. The noxt stepn will bo to measure difforent types of mnchinen ospecinlly with different typos of rotary 8x0s. E.g.: With this rothod a rotary axis that just movon + 6 5 O hnn nlready beon checkod (3). Anothor ntep will be to define a systematic procedure to evaluato tho single error compononts, and porhnps to include this evaluation in the exintinp noftwnre pnckago for the circulnr tost (I).
Acknowlednesgnng I would like t o thnnk D1pl.Ing. ETH A. Bucher and Dipl.Ing. ETH X. Yu at the Inntitute for Hnchine Tools Construction nnd Production Engineerlng (IWF) nt the SWIEB Fedoral Institute of Technology (ETH) in Zurich for carrying out syntomatic sensurmentn according to tho new method and compnring the renultn with convent ional moaaureronts.
Fin. 8 Hoasureront of tnngentinl doviationn, rndlus of circle equals 120 mm, clockwiso contouring. l...contourinp 2...contourinp
Knnpp, W., Hrovat, S., 1987, The Circular Tont for Testinn NC Hachino Tools, S. Hrovat, Trottenstr.79 CH-8037 Zurich Schwoizerische Patontanneldung 4238/89-0,27.11.89, Vorfahron zur nensung der Abweichungen zwischon Bewogungen 01n.r nnnchino, dio urn eine Drohachne und mit z w e i Trnnnlntionrachnon erzeugt wordon.
YU, X . , 1989, Abnahmo von NC-Bearbeitungszontron, Diplonarboit am IWF dor ETH-Z, WS 1989/90.
npoed 500 mm/min speed 1700 nr/min