AGMA 02FTM9-2002 Gear RollScan for High Speed Gear Measurement《高速齿轮测量用齿轮滚动扫描》.pdf

1、02FTM9Gear RollScan forHigh Speed Gear Measurementby: A. Pommer, FRENCO GmbHTECHNICAL PAPERAmerican Gear Manufacturers AssociationGear RollScan for High Speed Gear Measurement ARevolutionary New Method for a Complete TopographicalInspection of GearsA. Pommer, FRENCO GmbHThestatementsandopinionsconta

2、inedhereinarethoseoftheauthorandshouldnotbeconstruedasanofficialactionoropinion of the American Gear Manufacturers Association.Inmoderngearproduction,simpleinspectionsystemsforquickandclosebymeasurementshelptoproducehighquality.The Gear RollScan method inspects all flanks of helical gears within abo

3、ut 1 minute.ThegearRollScansystemissimilartotheone-flankgearrolling inspection. Themaster gear,however, hasmeasuringtracks on selected flanks. All other areas of the master gear are clear to create a contact ratio below 1. Thus, a clearallocationbetweentheangleofrotationandthecontactpointcanbemade.

4、Asuitablenumberofteethratiobetweentheworkpieceandthemasterwheelischosen. Everymeasuringtrackonthemasterwheelischosen. Everymeasuringtrackonthemasterwheelwillcontacteveryflankofthespecimenafteraspecifiednumberofrotations. Twomastergearsarein roll contact with the specimen simultaneously. One master g

5、ear inspects all left hand flanks, the other all righthandflanks.Thetopographicalmeasurementofallteethrequiresthedeploymentofstatisticalmethodsforevaluatingthedeviations.New parameters have to be derived, describing features of the gear which are critical for function, such as the noiseexcitation, a

6、s function of the topography.With regard to the tolerances, this measuring method has wide-ranging consequences. Whereasa standard evaluationconsiders 4 teeth, with the effect that a gears worst tooth is measured more or less by coincidence, the RollScanmeasuring device always will find the worst to

7、othCopyright 2002American Gear Manufacturers Association1500 King Street, Suite 201Alexandria, Virginia, 22314October, 2002ISBN: 1-55589-809-21Gear RollScan for high speed gear measurement.A revolutionary new method for a complete topographical inspection of gearsAndreas Pommer , FRENCO GmbHRolltest

8、er with component between two master gears1. History Of Development:The idea for a newgear testingprocedure wasborninCologne.Dr.Guido Hopffrom Fordmade thefirstcontact with FRENCO about the development ofsuch a gear tester in 1995. In the following yearsFRENCO built the first fixtures and suitableel

9、ectronics for fast capturing of measuring points.The software was written in Cologne. In 1999 thefirst gauges were installed on the shop floor. Theprocess is protected by two patents 1. The Fordteam was honored with the ,Henry Ford technicalachievement award 1999.Since 2001 FRENCO is holding the lic

10、ense to bringthis technique to the market. FRENCO is nowresponsible for development, design andproduction of the complete system. The largepotentialofthissystemstillcreatesdemandfornewideas and solutions.2. Existing Gear Testing ProcessesThis article is about a new gear testing process,called RollSc

11、anning. It is useful to considercommon methods first. Existing gear testers couldbe divided into two groups:2Classification of gear testersexisting existing NEWtype of contact point contact line contact short line contactexample coordinate measuringmachinessingle or double flank rolltesterRollScanni

12、ngelements of contact ball probe gears contact by smalltracks on gearsprecision very high limited by master gearqualityhigh, due to discreetcontact areasdensity low, normally only 4teeth high, complete surface iscoveredhigh, all areas and allteeth are coveredspeed slow very high high, 12minutesconta

13、ct theoretically idealsurface contactrealistic, gear-to-gearcontactrealistic, gear-to-gearcontactflexibility very flexible individual master gearsrequiredindividual master gearsrequiredresults independent inspectionof many gear errorssum error inspectionwithout access toindependent errorsindependent

14、 inspectionof many gear errors3. The Process of RollScanning:Single Flank Roll TestRollScanningisbasedontheknownsingleflankrolltest. Main difference to the well known double flankroll test is the use of a fixed center distance. Withone driving and the other gear being driven a singleflank contact is

15、 achieved. The deviations of thesample will be found as an angular error betweenboth axes. So two rotary encoders are necessaryone for each gear. As this method is a realistic copyof the real transmission a large contact ratio istypical.Thismeanstheresultsshowaninterferenceof a number of tooth flank

16、s. For helical gears acontact ratio up to 3 is typical.plane of actionTypical single flank test.The section in the ,plane of action shows contact between two gears along 3 lines.3SolutionAs the main solution the contact ratio needs to bereduced. This is achieved with so called tracks onthegearsurfac

17、e.Toavoidinterferencebetweentwotracks, they are separated by a number of,transport teeth between every track-flank. Twodifferent types of tracks are used:Lead TrackProfile TrackThe surface beside the tracks is lowered about0.2mm to always keep contact on the tracks. Alsoall ,separation teeth are low

18、ered by the sameamount on their complete surface. As a result wehave now only a short part of the complete ,line ofcontact left over. It is nearly a point contact now.plane of action with leadtrack:contact is reduced to a short line at one flankThe output signal of a single flank test is usually the

19、angular transmission error. If we set contact ratiobelow 1.0 in some areas tooth flank contact is lost.These zones are part of the design and will beignored by the software.transmissionerrorrotationRaw signal for calibration and for a component;every track takes 12 points for each flank, anactive zo

20、ne is separated by an unused garbagezone because contact ratio is smaller than 1.0.Because we know the location of the tracks we cancalculate the exact position of the contact point allthetime.Everydetecteddeviationcouldbelocated.So it is possible toscan thegear surface.Thereforewe call this process

21、 RollScanning. To cover a flanknot only in two sections many different tracks areused. A typical rollscan master is designed withabout 7 lead tracks covering the whole flank frommajordia.downtoformdiameter.Inaddition,upto3profile tracks are used to check involute profile inthree sections. After one

22、rotation every tooth ischecked by a different track, so only one section isavailable. The master gear is usually designed witha different number of teeth than the component. Sodifferent flanks will touch for every new rotation.Increasingthenumberofrotationswillresultinmoretracks per flank. If the nu

23、mber of rotations is largerthan the number of teeth of the component, allflanks are covered by all tracks. The result is acomplete grid of points for all flanks.4lead trackpointsprofile trackpointsTo check both sides of a gear the same time withsingle flank testing twomaster gearsare used.Oneforleft

24、andoneforrightflank2.Allgearsaredrivenby one motor . Brakes force the gears to stay indirect contact all the time. At the moment thenumber of points is limited, finally about 370 pointsfor each flank are stored. The whole process takesless than two minutes, for a small gear about oneminute. Remember

25、 this will include the check for allteeth.4. Multiple Inspection ReportsThe storage of all points allows the visualization ofmany different types of inspection reports. First ofall the classic report with involute and lead for 4teethispossible.Nearlyallgearmakersare usedtowork with these reports.Tor

26、espect habitsof ourusers weshow layoutsverysimilartothosefromCMMs.Ofcoursetheusercanchoose which teeth he wants to see in his report.But this will use only 1% of the stored informationsimultaneously.So FRENCO, together with our customers, havecreated totally different layouts for reports.5Themainobj

27、ectivewastobringmoreinformationtothe user. It is possible to show all profile or all leadgraphs beside each other to find the mostinterestingflanksasfastaspossible.Itispossibletocalculate a single flank test result. It is possible tochoose lead and involute graphs by ,walking downto index error grap

28、hics.Statistical ReviewThe increased number of data creates a problem.The operator will need more time to check allresults. As an example, for a gear with 35 teeth wehave 70 profiles, each with up to four criteria likesumerror(F),formerror(ff),directionerror(fH)and crowning (c). The same amount is a

29、vailablefor lead. Also for spacing we find some values pertooth. Our solution is to use statistical methods tosupport the operators. For every value thedistribution is shown. Now the attention is reducedto the mean and the maximum values.We call this sheet TOLANY report. The mean valueis shown by th

30、e vertical line. If some values for acriteria are outside their tolerance the graph willbecomeyellow.Ifthemeanvalueisoutsideitwillbered. This distribution will also form the base for anautomatic assessment. If the gauges are part of aconnected production chain, no operator could beasked to have a lo

31、ok at the results every minute.The system has to decide by itself. From othersystems with a high density we learned to seetolerance limits under a new light. If you take verymore and more points it is no question to findpointsoutside the limit. Statistical methods must be usedto come to a decision.F

32、rom this sheet we can learn something else. Tohave values from all tooth is a very big advantage.Because now we can be sure to find the extremevalues.By checking only 4 teeth we create an uncertainty.Uncertainties are not what we want in a measuringsystem. Wefind asignificant variationfor themeanval

33、ue and for the range if we use only 4 of 33measurements to calculate our results. Even if wetake these 4 measurements with the highestprecision, this uncertainty can not be avoided.Systems performing complete checks will showlower uncertainty.UncertaintiesRollScanning is a highly precise inspection

34、systemfor the shop floor. The inspection of the involuteprofile depends on a successful compensation ofthe master gear errors. Larger deviations of thecomponent can lead to decreasing precision. Tocompare absolute results you need a reference likea CMM. This can easily create differences of indexand

35、 runout due to different clamping conditions. Ifyou ensure comparable clamping, uncertainty is 3microns or better. Repeating the samemeasurement is normally better than 1 micron. Thereason is a rigid design and the simple kinematicprocess with only three rotating axes.6-0,02-0,0100,010,021 4 7 10131

36、61922252831valuemean (t4)4 teeth check vs all teeth checkexample T=331234Measurement: -0.006 to +0.020mean value complete : 0.0068 mean value (T=4): 0.0033 to 0.009range complete : 0.026 range (T=4): 0.014 to 0.0245. TopographyThemostinterestingandmostdifferentreportistheso called topography. Only h

37、ere it is possible toshow 100% of the information. And only here youcould be sure to find all the truth about your gearssurface.7The picture contains a 3-dimensional grid tovisualize the surface. Colored dots mark all pointswithina5micronrangetothehighestpointtoformacontactpattern.Acoloredlineshowst

38、hepathofthecontact line along the flank, highlighted sections onthis line accentuate the part where this tooth willhave contact, while for the rest of the line the toothbeside is in contact. The vertical position of thewhole grid shows the influence of spacing andrunout on the displayed tooth. This

39、threedimensional graphic is very easy to understandevenit containssomuchinformation.Both,expertsand beginners have learned to work with it soon.Experiences With Topographic Inspections.A topographic inspection is very useful which welearned from our customer Volkswagon in Kassel.,Whydidthisgearfail?

40、Askthisquestiontoyourselfif you look at the classic report below.You will not find the reason here, because the errorwas not detected by the CMM machines. But theassembled gearbox did not pass the final noisecheck.Thereasonisobviousifyoutakealookatthetopography. The final grinding process left overm

41、aterialinonecorneroftheflank. RollScanningwillfind such errors always, also if they are at one toothonly.6. ConclusionApplications for RollScanning:The system depends on special master gears.These gears could be replaced fast to adjust thegauge for different components. But they aredesigned for one

42、component only. For economicalreasonsaproductionprocesswithlargenumbersofidentical items is a typical field of application. Thissystem is designed mainly to support production ofgears for cars. It could be used in this manner:- Automatic inspection up to 100%, damage(nick) detection, classification

43、advantage: veryreliable inspection due to complete topography- Manual inspection of samples (i.e. 1 of 50), in-creasing inspection after tool changes advan-tage: very fast reaction inside the production- Process control advantages:very stable measurements with good repeat-abilityandprecisemeasuremen

44、tbycheckingofall teethLikeeverynewtechnology RollScanninghas tofindthe right place in the modern gear manufacturing.Thisversatiletechnologyoffersmanyopportunities.8Combined advantages- very fast inspection, 2 to 3 seconds per tooth- simple operation- very compact machine design- no air conditioning-

45、 all tooth check- complete surface covered by topography- shows usual CMM inspection reports- export of data for further analyzing possible- check of oily parts- process control within the production line7. OutviewFRENCO will release a complete new family ofelectronic units soon. For RollScanning th

46、is willresult in increased measuring resolution up to0.0001(equal to 0.1 Micron at100mm). Also thenumber of points will climb up from 32000 per gearto 320000 per gear (equal to 150 points per track).To create a very fast 100% inspection system acombination of the normal single flank testand RollScan

47、ning is under development. All tracksare usually applied to one flank. So the backside ofourtrackmastersisstillanormalgear.Ifwereverserotation the system performs a conventional singleflank test within some seconds. With a mix ofsingleflank for most items and topographic inspection foronly some item

48、s a very fast system is created. Allflanks of all parts could be covered. The completeinspection of the topolographic samples forms thebaseforasafeprocesscontrol. Wehavepresentedthis solution at the Control Exhibition in Germanyto the public.1 Patent DE 19519429C1 (2.10.96) Topographicinspection with track masters2 Patent DE 19719249.1 (24.9.98) Combined in-spection of both flanks with two mastergearsROLLSCANNING Single flank roll testleft and right flank of a track master:right side: activ flank with tracks (red) and separation teeth(white)left side: backside without tracks