Quality Gear Inspection Part II
,Robert E. Smith ,R.E. Smith & Co., Inc. ,Rochester, NIY
Diagnostics This section will deal with the use of gear inspection for diagnostic purpos
es rather than quality determination. The proper evaluation of various characteristics in the data
but the gear may still be unsuitable for many applications, especially when noise is an issue. One must study the characteristics in order to
sort out the cause .. A good example is the difference between Examples 4 arid 5 in Fig, 2. If the waviness is the same on all teeth, the problem is probably within the hob or the accuracy of the mounting. [f the waviness is different. on every tooth, the cause is likely to be a machine looseness problem, Hob Feed Marks. When parts are bobbed at a high feed rate. uch as for pre-shaving, scallops in the surface characteristics are very evident. Generally, these scallops are so deep that involute charts are quite useless. Usually the only valid data are lead charts. The slope of the charts can show if the lead is correct. and the
Gea[ tlal Ia~(
Gca! t::m~ Lead ~iuiallgD ProlJable c.\lSe Gear Blank Has Fate fiunaul-Wobbl.
for the solution of qual.ity problems, It is imporrant to sort out whether the problem is corning from the machine, tooling and/or cutters, blanks, etc. An article by Robert Moderow in the May/June 1985 issue of Gear Technology is very useful for this purpose,
Machine Preblems, The following
lems Originate with the machines.
niques and devices win indicate when probLead and Involute Charts. Lead charts will often indicate problems with machine alignment as well as blank mounting (see Fig, I), A comparison of Examples I and 3 in this figure is instructive. Whether the charts of both sides of the teeth are parallel is a good clue to the source of the problem. Accumulated Pitch vs. Runout Data. If a gear comes directly off a bobber (no subsequent finishing operations) and has a very low runout, but a high accumulated index variation, pitch or total a problem generally there is probably
P,alJable c.",," Machi"" M,,,,lignmenl
r MiSilhgn""'n~ Setup
SEal t::Ia:iiLiliid Em:u: .PrDbaIJleC.USOl Mac~"le ErrorG.a's. Hob AI1lor End I'1<Iy
f::: f::: I-f-:
;i:: .... f:::
~ ~ :::-
t:.:: ;'~ t~~ ~
~ ~ Numtiefi\== ~ sa
with a worn-out table drive gear set.
Hob Problems, Hob problems
show up in involute charts. The e problems have several possible causes ..The hob could be made very accurately, but sharpened or mounted poorly; or it could be a hob that was made with excessive runout at the time of original manufacture, Even if it is sharpened and mounted correctly. it will not cut a good gear. This problem usually shows up as some form of waviness
Probable C.useHob Br.a i"!l OUI of Cut. Hob ArIxlr or Watt Arbor FI?? ng
P'obable c.useMachine MlOal'lllVllllnt. . Hob NOI Accu,ale Enough When O'.gonalllobtJ'"!I
slope error in. the involute charts of slope or Fig.1 -lead 4 and il1l1ollute chart. 5
(see Fig. 2). These characteristics
waviness may very well fit within the allowable bandwidth of the appropriate AGMA Kchart.
very useful for diagnosing
Probable Cause ~ Hob Sharpened With Positl'''' ~.k. Pmb~bl. Causa Hcb Sharpered Wit~ Neglliva Rake .. Probable Cause ~
gear noise and
? Flute Lead Error
transmission error problems. Other technique,
such as separate determination of "mean involute slope" and waviness in involute charts, should be used for these evaluations .. AGMA Q7 or Q8 gears can be made that are noisy, and Q]3 or Q14 gears can be made that are quiet. Fig. 4 shows involute traces from a gear that is
t"10UI ~ IMI,",TIPI
Probable Gause ~ Hob Mounting E~ors Hob Sharpening ErrDIS
about a Q7 in terms of runout, The effect of runout shows up in the traces as a variation up
IFlul. Spacing Off) ~HOb Not Accurate Eno~h ~, r<;:
P,obable Ma~h,"e Worn or Loose Worn Hob Spindle Sleeve Bearings &ce~,ve HobArbor End Play Worm Wheel Worn ;;;,Worm AU"cmg Qui
2 cau;. ~
ro .00066" in slope. However, the mean involute slope is only .00015", and it matches its mate properly, (See the middle example in Fig. 4.) Originally, the mean involute slopes for the gears were like the upper example in Fig. 5. These gears were noisy. Experimentally, characteristics of the bottom example, gears and were also made with the mean involute slope these also were quiet. In all cases, the runout would make them a Q7 set of gears, and they would have slope variation ..This points out the
fig. 2 -
inadequacy of the traditional methods of quali-
Lead and involute chart.
ty determination as far as noise is concerned. ISO Approach. The International Standards
Organization (ISO) uses a different approach to the evaluation of involute and lead charts. In the main body of the standard, they use a rectangular tolerance zone, much as AGMA uses the K chart. However, in an Annex, they have alternate methods of evaluation that are more appropriate for diagnostic involves evaluation doing and the a least-
A P Diom. Involute checks on !he sams loalh vary iflhe gear is hObbed wilh 3 heavy feed rate The involute d"ft! end one check may hal'8 high point of feed seallep at pitch line, and BI'lOlh.,<heel< may h....., high point "a""th., place .1009 Ille i""",ute.
control of noise and transmission error. This process squares-best-fit line (usually a straight line) to
the involute or lead trace (see Fig .. 6). This line is laid on the trace and the various parameters, such as slope, average slope, waviThis is much more useful for ness and total error, are read and quantified independently. many applications and diagnostics. In addition, the ISO standards give recommended tolerances for accumulated pitch error
21 Hob hu
Po'Sible causa 11 Hob huD:s) wl1ning M weave error in lead out of phasa from thread to tI1read
error. These are characterisEquipment There
ticsthat are much more direct and functional. Good Measuring achieving are machines: Good measuring equipment is essential for quality gear measurements. types of gear the older, traditional two basic measuring mechanical
Fig. 3 -lead
and involute chart,
depth of the waves win tell if there is stock for cleanup in the shaving operation. However, the
IRobert IE. Smith
is the principal ill the gear consulting firm of R. E. Smith & Co., Inc .? and one o/Gear Technology's technical editors.
in the involute
invalid for the determina-
tion of profile shape (see Fig. 3).
generative machines and the newer Computer Numerical Control (CNC) generative or Coordinate Measuring Machines (CMM).
Conjugacy-NQise alld Transmission Error. A mentioned earlier. K charts are not
Mec.lumical Machines. The mechanical
machines an example of which is shown in Fig .. 7. use means such as base disksv sine bars, levers, formers, etc., to generate the theoretical motion such as involute or helix (lead). A probe on the tooth surface then measures the variation between the actual tooth surface and the theoretical shape being generated. Usually
its mounting, work-holding tor, etc. If you don't equipment and practices,
equipment, operahad product can
have good inspection
become very expen ive in terms of lost time, poor quality and bad customer relations. One should look hard at the true cost of not having good equipment compared to. thecost of buying it. A Look At The Competition A good example of the value of good inspection equipment and practices was found at the 1991 AGMA Hear Expo, A Japanese plastics Qll. gear manufacturer was displaying gears made to AGMA quality levels of QW to. Our American plastics and sintered gear typically claim the ability to measuring equipment is manufacturers
of some type makes the resulting charts.
CNC or CMM Machines. The newer CNC
machines come in two types. On some. the machines did CNC i used to generate the theoretical motion. much as the older mechanical (see Fig. 8). On the other hand. CMMs measure the tooth forms by moving the probe to a series of X, Y, Z, and
High resoluare in
tion and accuracy is necessary.
make only Q6 to Q7 gears. What is the difference? Good diagnostic and practices. regularly Even though this company measuring
mechanical or CNC, alignment and calibration in order to have confidence the results .. Any machine should be calibrated on a regular basis. A new standard, AGMA 2010, Part I, is in the final stage of preparation, Part 1 deals with involute, but a standard for other parameters will follow, and the same principles apply. The first step is to check the alignment of the machine for squareness, parallelism, runout of centers and spindle, etc. Then mount a certified artifact such as an involute master. Many measurements are then made and plotted on X and MR bar charts for a statistical evaluation of capability. This establishes bia: and vari-
making fine-pitch gears (20 DP and finer), it uses CNC elemental machines and even single-flank. testers to mea40°
5.24" Results Left Flank , /1 0,000· Right Flank
4 3 lHa 1.9 16 fHam (MIS)
2 ·3 ·1
Fig. 4 -Involute GearMIT
traces with runoul. Gear MIT
ability (accuracy and confidence levels) for the machine. Remember that resolution is not the same as accuracy. PinionMn
Tip +.000'" Tip +.0006"
National Institute of Standards and Technology (N[ST) has not calibrated and certified masters for about ten years. The recent work by the calibration committee of AGMA ha stimulated in artiTip + .0006" -
Tip - .0001"
intere t by NlST ill resuming it effort set up a calibration service again.
fact calibration. A project is now under way to Gear MIT
Cost of Inspection .. Many people seem to find it hard to justify the cost of good in spectionequipment .. After cutting or grinding all, the inspection in the world machines don't make chips. However, the best machine doesn't automatically make good gears. There
, Root Fig. 5 - Mean involute traces (MIT).
are still a lot of other variables. such a tile cutting tool and its mounting, the gear blank and
sure gear quality.
good symmetrical gear blank design that promotes uniformity of teeth around the gear.
American gear manufacturers only use double-flanktesters.
in this field
They can't reaUy
tell what tooth form they are making or even
know what is wrong when the product doesn't
Tooth Tip or Start of Chamfer Start of Active Profile Start of Usable Prali Ie Design Profile 2 Actual Profile 3 = Mean Profile A = E = F = 1 =
meet the specifications.
Profile Evaluation Range = Tota I Profile Deviation 'Ha = Profile Slope Deviation ffa = Profile Form Deviation fa = Pressure Angle Deviation fwa= Profile Undulation ft.a = Profile Undulation Wavelength
are not good measuring devices for diagnosing problems: iterations
necessary to develop
don't do the good gears.
One of the major reasons that American plastic gear quality is as low as Q6 or Q7 is that the gears have too much
tering the hole in the center of the gear. This hould be an easy problem to solve with proper
mold and cavity design. Better measuring point manufacturers technique would soon
III the right direction for meeting the competition and opening up new potential markets for their products. Their
products are very uniform from piece to piece, so it is just as easy to make all pails good as to
make all parts bad.
Process Control. More use should be made
B 1345Fig. 7 2Measuring Slide Rolling Sliae Vertical Slide Control Slide Helix Guide generative 7
rather than double-
flank composite methods. This also applies to
6 - Compensation 7, 9, 12 - Compensation 11 & 13 involute 10 -
gears that are finer than 20 DP. This kind of information fosters better diagnosis Our AGMA of probneed lems as wen as better process control.
Work Adaptor Optional Base Disk and Straight Edge
and lead tester.
changing. Elemental tolerances and inspection need to be extended to finer pitches (even as far as 80-100 DP). We should be using accumulated pitch variation rather than run out. The standard should also allow the optional use of single-flank inspection as well as double-flank and elemental methods. One way of doing this is to make use of the
new ISO standards currently
Tesler Witll 7 CNC-cornroll'ed Axes
Vertical Slide Movement. Linear Scale Aesolution: 0.5 J.Im (20 rnicroinchesl, TAxis - Radial Slide Movement. T/ linear Sea la Resolution: O.5[.lm
Z Axis -
~. Axis - For Angular Positioning of ---I-"+--tlll-fl-+ Probe, ± 90"
Y Axis - Carriage Positioning.
Linear Scale Resol ution: 0.5
and published. and perhaps
Refereneese I. Moderow.
In fact, AGMA is now very of these standards, someday that is what our gear
active in the development indu try will be using .?
II' Axis - Work Rotation. Angular Resolution: 0.0001"
X Axis - Tangential Linear Sea Ie Resol ution: 0.5J.1m
Robert H. "Gear Inspection and Chart Interpretation," Gear Technology. Vol. 2, 0.3, May/June 1985, p. 30. Acknowledgement: First presented or lite AGMA 21st Annual Gear Mamifacmring Symposium. October /0-12, 1993. Reprint. ed with permission.
Tell Us Wha~You
fig. 8 Generative
and lead tester.
and/or useful, please circle