1、09FTM02AGMA Technical PaperImplementing ISO18653, Gears -Evaluation ofInstruments for theMeasurement ofIndividual GearsBy R.C. Frazer and S.J. Wilson,Newcastle UniversityImplementing ISO 18653, Gears - Evaluation of Instrumentsfor the Measurement of Individual GearsRobert C. Frazer and Steven J. Wil
2、son, Newcastle UniversityThe statements and opinions contained herein are those of the author and should not be construed as anofficial action or opinion of the American Gear Manufacturers Association.AbstractA trial test of the calibration procedures outlined in ISO 18653, Gears - Evaluation of Ins
3、truments for theMeasurementofIndividualGearsshowthattheresultsarereasonable,butaminorchangetotheuncertaintyformula is recommended. Gear measuring machine calibration methods are reviewed. The benefits fromusing workpiece like are artifacts are discussed and a procedure for implementing the standard
4、in theworkplace is presented. Problems with applying the standard to large gear measuring machines areconsidered and some recommendations offered.Copyright 2009American Gear Manufacturers Association500 Montgomery Street, Suite 350Alexandria, Virginia, 22314September 2009ISBN: 978-1-55589-955-43Impl
5、ementing ISO 18653, Gears - Evaluation of Instruments for theMeasurement of Individual GearsRobert C. Frazer and Steven J. Wilson, Newcastle UniversityIntroductionCylindricalinvolutegearsareprecisioncomponentswith a relatively complex geometry which must bemade accurately to fulfil their specificati
6、on in termsof noise, power density and reliability. It is commonfor gears to specify profile, helix and pitch toler-ances in the 5-10 mm regionand manyapplicationsdemand tighter tolerances. Modern machine tools,operated in a carefully controlled environment andcorrectly managed, can achieve these to
7、lerancesprovided thereis anappropriate independentmeth-od of measuring the geometrical accuracy of thegears and thus control the process.The traditional “Golden Rule” for metrology is thatthe uncertainty of a measurement process shouldbe 10% of the tolerance inspected. Measurementuncertaintyistheter
8、musedtoquantifytheunknownrandom and systematic errors that occur in anymeasurement process. With tolerances of5-10 mm, our measurement uncertainty should be0.5-1.0 mm, on the shop floor, which is still too diffi-cult to achieve and even National MeasurementInstitutes (NMIs) around the world can only
9、 justachieve these levels. Thus the shop floormeasuring instrument capability is an importantconsideration when interpreting measurementresult conformance with specification.In recent years the of range gear measuringequipment available to the gear manufacturer hasexpanded. There is greater choice o
10、f dedicated4-axis CNC gear measuring machines (GMMs),with three linear axes, a rotary table and tailstock.General purpose co-ordinate measuring machines(CMMs) are now equipped with gear measurementsoftware and previously where only the highestquality machines were considered for gearmeasurementappli
11、cations,recentimprovementsinerror mapping to improve measurementperformance and the introduction of scanningprobes systems has meant that even relativelymodest costing CMMs can now be considered forgear measurement applications. The gearmanufacturer has a wider choice of measurementsolutions than ev
12、er before, but how should theappropriate solution be selected?Figure 1. CMM used for gear measurementFigure 2. Example GMM - the UK primarygear measuring machineIt is surprising therefore, that when ISO publishedISO 18653 in 2003, Gears - Evaluation of instru-ments for the measurement of individual
13、gears and4a supporting technical report (guidance document)ISO/TR 10064-5 that the gear industry has notadopted the recommendations and applied thestandard more widely.The proposal to develop the ISO document camefrom AGMA using ANSI/AGMA 2010-A94, Mea-sUring Instrument Calibration, Part 1 - Involut
14、eMeasurement as the working document. Otherdocuments are also used extensively throughoutthe gear industry. The VDI/VDE guidelines 2612and 2613 1, 2 propose limits on measurementuncertainty depending on the DIN 3962 qualitygrade. They were first published in the 1980s butwererevisedin2000. Theguidel
15、inesalsoprescribelimits on runout of centers, machine alignment andinstrument repeatability and importantly theuncertainty of the calibration data artifacts used toprove machine capability. The VDI/VDEmeasurement uncertainty limits are used to definethe measurement capability of the instrumentsworld
16、wide.In the UK in the early 1990s there was generalacceptance of the philosophy of the VDI/VDEguidelinesbutitwasconsideredthatmoreguidanceon the procedure to assess measurementuncertainty was required. Also more guidance onthe routine testing of measurement instrumentswas required. The result of thi
17、s was a series ofcodesofpracticepreparedbytheUKNationalGearMetrologyLaboratory(NGML)andpublishedbytheBritish Gear Association (BGA) 3.One of the reasons that the guidance in ISO 18653is not more widely adopted is that measurementuncertainty is seldom considered unless a disputeoccurs, usually betwee
18、n customer and supplier.The suppliers measuring machine shows the gearsare within tolerance and customers machineindicates the gears are outside tolerance and thusrejects them. Sometimes the cause of thedisagreement is simply the interpretation of thespecification, a gear mounting error or a mistake
19、 inthe measurement process but at other times thecause of the differences are more subtle. Allmeasurement processes contain error, includingNMI and shop floor machines. The only certainty isthat the measurement result is wrong.ISO 18653 addresses traceability, calibrationintervals, sources of measur
20、ement uncertainty orerrors, basic instrument checks, environmentalconditions,calibrationartifactdesignandprovidesamethod for estimating measurement uncertainty. Itcontains sound guidance on how to estimate gearmeasurement uncertainty using simple robustmethods. It allows users to assess the differen
21、cesin measurement instrument capability and thusmake informed choices. It minimises the riskassociated with high accuracy gearing operating insafety critical situations and allow manufactures tofocus on manufacturing gears rather thanmeasuring them.Calibration methods - micrometerexampleIt is the ex
22、perience of the authors that many gearmanufacturersconsiderthatthey carefullymaintainand calibrate their gear measuring machines.Compared to the care taken to calibrate a simpleinstrumentsuchasamicrometershownisFigure 3,we do very few tests.Figure 3. Micrometer and M8 gauge block setused for functio
23、nal micrometer calibrationA typical micrometer calibration procedure is asfollows:S Check the micrometer spindle is free through itsrange of operation and the lock functionscorrectly.S Verify the fixed anvil is flat within defined limitsand free from damage with a calibrated opticalflat.S Verify the
24、 moving anvil is free from damage withan optical flat and then verify that the two anvilsareparallelwithinlimitswithanopticalparallel. It5is usual to use 5 optical parallels with differentthicknesses arranged to set the spindle atdifferent angles to verify for spindle runout.S Check the zero point i
25、s within acceptable limitsand adjust if necessary.S Use a range of traceably calibrated gaugeblocks to verify the measurement performancethroughthe25mmrangeofoperation. Itisusualto use 8 gauge blocks (M8) set as a functionalverification of the performance of themicrometer.S If all the results are wi
26、thin acceptable limits, thecalibration is complete and the micrometer isreturned to the shop for use.In addition, before use every competent operatorchecks the zero point setting before use and makesurethatitiswithinitscalibrationinterval. Weapplythese thorough checks to a simple single axis mea-sur
27、ing instrument used to inspect simple lengthswith tolerances of 15-100 mm.Many users of gear measuring instruments do notcalibrate them with this rigor. Most rely on themachine service engineer to perform a calibrationwith the gear artifacts supplied with the machinewhen it was originally installed.
28、 They may use agear artifact to verify the machine at 3 or 6 monthintervals and then use a mandrel to checkalignment, but in general the measurementuncertainty is only considered when there isproblem, flagged by manufacturing machineoperators or the customer.How ISO 18653 worksThe key concepts in th
29、e ISO 18653 standard aresummarized below:S Measurement uncertainty is assessed byperforming a series of measurements on a gearor gear artifact that has been calibrated in anaccredited calibration laboratory.S It is a comparison process: the results from thecalibration laboratory are compared with th
30、eresults from a series of measurements on thesubject measuring machine.S All parameters that the machine will measureand evaluate (profile, helix, pitch and tooththickness parameters) are analysed.S The gear or gear artifact should be of similargeometry to product gears inspected by themeasuring mac
31、hine (geometrical similarityimplies the same size and weight, module helixangle, facewidth and where possible, the samemeasurement position and locatingarrangement on the measuring machine).Artifact design is discussed in detail inISO/TR 10064-5.S It is preferable that data for the series ofmeasurem
32、ents is gathered over a long periodoftime so that affects from temperature variation,machine alignment and different operators aretaken into account (reproducibility data). TheISO procedure uses the mean and standarddeviation from these tests to estimatemeasurement uncertainty. The minimumnumber of
33、tests is 10, but 30 is recommended.S Guidance on other factors that are known toeffect measurement results is given such astemperature and instrument alignment(ISO/TR 10064-5 covers these in detail).S It recommends minimum re-calibrationintervals for gear artifacts.S The methods are consistent with
34、those used fortask specific calibration in general metrologywith CMMs.S If the calibration artifact is significantly differentto the product gear geometry, additional timeconsuming tests are needed to establish anuncertainty budget for the product gear. This iswhy the standard recommends that thecal
35、ibrated gear is similar to the product gears.S The subject of fitness for purpose of theinstrument is complex and is not covered in thestandard but is discussed in detail inISO/TR 10064-5.S An accredited calibration laboratory is one thatcomplies with the requirements of ISO 17025,that is laboratori
36、es accredited by A2LA, DKD,UKAS etc. The calibrationcertificate thatstateshow the gear was measured, calibration dataand its measurement uncertainty.Care has beentaken whenpreparing thedocumentto make it applicable to dedicated GMMs andCMMs.TraceabilityThe requirement that calibration data is suppli
37、ed byan ISO 17025 accredited laboratory implies mea-6surement traceability. Traceability implies thatthere is an unbroken chain of calibrations betweenthe subject measurement result and the primarystandards (of length, angle and temperature for thedimensional measurement of gears) at the NMI.Traceab
38、ility is usually established or transferred bycalibrated artifact and is illustrated in Figure 4.Thus data from a properly accredited calibrationlaboratory is required to establish measurementuncertainty.Estimating measurement uncertaintyIt has long been recognized that measurement pro-cesses are su
39、bject to errors that are not known andtherefore can not be corrected. The results fromany measurement process are thus incompletewithout the statement of its associated measure-ment uncertainty. It is common practice is to definea measurement uncertainty (U95) with specific con-fidenceintervalof95%,
40、meaningthatthereisa95%chance that the actual result lies within upper andlower limits stated. There is, of course, a 5%chance the actual result is outside the upper andlower limits stated. This is illustrated in Figure 5.Figure 4. Example traceability chain for gears using artifacts as transfer stan
41、dardsFigure 5. Definition of measurement uncertaintyThemeasurementuncertaintystatementisastatis-tical definition of how wequantify measurementun-certainty. The general calculation is defined in ISO18653 as:7U95= ku2m+ u2n+ u2g+ u2w+ |E| (1)Where :K a coverage factor is set to 2 to give anapproximate
42、 95% confidence interval,assuming the distribution is a normaldistribution.umstandard deviation of the series ofreproducibility tests of the subject machine(10-30 test results are required to complywith the standard).uncalibration artifact standard deviation.Assumed to U95/2 where U95is calibrationc
43、ertificate measurement uncertainty.uggeometrical similarity uncertainty toaccountfordifferenceingeometrybetweenthecalibratedartifactandtheproductgearsmeasured.uwworkpiece similarity uncertainty, account-ing for uncertainty due to the workpiece -e.g., it could be due excessive workpiecedeflection dur
44、ing measurement or poordatum surface quality, etc.E bias or difference between the mean mea-sureddata(xmean)andthe calibrationvalue(xcal).This relatively simple formula (1) is very difficult toapply in practice without suitable experience inmodelling measurement uncertainty, but ISO/TR10064-5 provid
45、es information on applying it to thecommon situations.The easiest situation is to estimate the uncertaintyof measurement taken on the calibrated gear arti-fact used to establish traceability. In this situation,ugand uware zero, because the product gear weare measuring is the calibration artifact (or
46、 a nearidentical copy of it). Thus the resulting formula issimplifiedtothestandarddeviationofthecalibrationdata, standard deviation of the measurements onthe subject measuring machine and the bias(differ-ence) between the mean of the measurements andthe calibration data values, see formula 2.U95= ku
47、2m+ u2n+ |E| (2)The procedures for estimating values of ugand uware more complicated and discussed briefly inISO/TR 10064-5 but the details, particularly for ugwhen there are significant differences in geargeometry between the calibrated artifact, arebeyond the scope of that document. Somemethods to
48、 overcome this are discussed in thefollowing sections, but the recommendation thatusers obtain work-piece like artifacts to establishtraceabilityavoidsthedifficultyofestablishingtheuguncertainty contributions.Artifacts and master gearsISO 18653 provides examples of different artifactdesigns and ISO/
49、TR 10064-5 provides furtherinformation on the design and specification ofartifacts. Users should ensure they have artifactsthat cover all the features that are measured on themeasuring machine, including profile, helix, pitch,tooth thickness and other features such as datumaxis runout correction.Traditional artifacts are illustrated in Figures 6, 7and8. Thesewereoriginallydevelopedtoprovetheperformance of manual gear measuring machines,wherebasediscswereusedwitha mechanicalsinebar to set the base helix angle. In these cases therange of helix angles w
