【转帖】measuring gd

yang686526

measuring gd&t
is there any site that explains the actual measuring methods and procedures for various geomteric tolerancing methods?
its all well and good to spec concentricity, cylindricty, tir, circular runout, etc, but if the shop guys dont know how the measuring procedures that differentiate between them, i may as well just use tir only.
-------------
randy
there are a lot of references given here :
randy:
there isn't that much out there on measuring and don't expect a asme gd & t prof. to know how to measure either. it is not part of their exam.
just some thoughts though.
there are other areas that blow the minds of people on the shop floor. when measuring flatness, should the burr be included or is it a separate entity?? mmmmmmmmm?
tir (total indicator reading) means range of an indicator and most people in the usa use fim (full indicator movement). never put tir on a drawing. if one places circular runout, as an example, tir is understood.
the standard and most text books only show dotted lines reflecting the tolerance zone. i always re  
i answered a similar question on another forum a while ago.
let me say that designating gd&t is a mixture of art and science. it needs to define the boundaries of variation and reflect variation that is consistant with function.
you as the designer should contemplate with the design engineer what the constraint on geometry variation is for. is it dynamic or static, fit or alignment, axis or surface, boundary or virtual condition...?  do you want to refine controls for form over orientation over location, combine controls, or do them in another order? that is what drives your feature control selections.
remember all inspection, no matter how rigorus, is abbreviated. it is accomplished with readily available imperfect instruments on a finite set of an infinite number of points. it is the inspectors job to process the observed variation consistent with the defined degrees-of-freedom and draw a conclusion about the data in comparison to the boundary defined on the drawing.
  
here is the previous post:
      
walt, this is a very long response to your question!
i remember when our product specifications were first converted the from notes that engineers crafted to describe tolerance with words like contour, profile, eccentricity, concentric, symmetrical, runout, true position, fit, interference, slip, out-of-round, flat, parallel, square, perpendicular, etc鈥?datum references were explained in the note or perhaps not even mentioned because the assumption was considered obvious.  the concept of a variable tolerance was explained with the terms "allowance for size" and "combined effects of size and eccentricity". if anyone needed the ultimate dispute settlement interpretation they had to find the author of the note and resolve the difference between design intent and measurement practice.
along came ansi symbology and draftsmen were tasked with converting legacy design notes for application to new or revised designs. the words set-up and gage disappeared from the drawings, position callouts were applied liberally and mmc symbols were declared by default in general notes. many in engineering lost the art of conveying their design intent, draftsmen were re-titled designers and the art of depicting the tolerance strategy fell to the designers. tolerance stacks reinforced the strategies and feasibility disputes helped to arbitrate problem strategies.
why am i telling you this? because i noticed that there was a shift in responsibility for establishing dimensional strategy from the engineer to the designer over the last few decades. while that shift might be good for standardizing what comes out of design it is also bad for coordinating design intent with subsequent customers of the product design.
the design specification is vulnerable to a narrow myopic perspective of how the strategy manifests itself in other areas unless it is arrested collaboratively on occasion by each affected discipline.
there is no way for me to sugar coat this so i will just be blunt.
    cad designers live in a world of perfect geometry where lines are straight, gravity is zero, boundaries are absolute, and variation is linear. it all stacks up to prove that this strategy works.
    gage makers and inspectors try to comply with the literal design definition but struggle with design intent when the design does not reflect a physically stable reference or cannot be repeat-ably measured. they make datum reference substitutions at their convenience when limited choices of measurement tools and fixtures coincide with inaccessible component geometry. they are influenced by gravity, imperfect geometry, variable boundaries, and an abbreviated collection of an infinite set of measurements.
    inexperienced machinists "process engineers" are torn between the suggested process steps and their gut feelings of the process steps necessary to achieve an acceptable end product. experienced machinists ignore the suggestions and stick with previously trusted and proven processing methods.  if only the drawing could match the process life would be good for them.
    sta and quality auditors summarily reject the attribute gauging that the designer expects will verify the variable tolerances he declared. they demand by quality policies and iso procedures that capability ratios be met using variables data. manufacturing consequently must conform to all tolerances as if they were rfs in spite of the liberty the designer has included in the design stack.
    engineering is so busy doing internal and external regulatory procedures and documenting each study that there isn't enough time to understand how the specification, that they are ultimately responsible for, affects each of the disciplines that apply it.
i have worked in every one of these disciplines in a very large organization and witnessed the autonomy that each discipline exercises in applying the specification to their outcomes. seldom would i consider it well coordinated but it amazes me that it all comes together and eventually works. no successful company stops chasing problems until it does.
here are little bits of advice for each discipline.
cad
your best guess of how someone has to inspect or process a design based on the way you declared the gd&t may not even be considered by the subsequent operations. if the datum structure that you declare is unstable or "out-of-order" i.e. |secondary|primary|tertiary| it will be substituted without hesitation.  note!!! functional gages are never functional if the modifiers are not declared according to function. they are simply attribute gages. how can you decipher if they are applied functionally????  if the feature's functional performance worsens as its location deviation increases then it should not be allowed additional location tolerance relative to size!!! static fastener clearance holes are usually the best example of proper variable tolerance application but if the hole is sensitive to gasket bearing area even this application might be functionally designated rfs. "just because the tolerance applies to a feature of size" is never the only criteria for mmc designation. the spread between a pair of dowels and dowel holes is typically a fit scenario that would beg the mmc modifier but the location of the dowel pattern on the structure is not. why should the tolerance of the patterns location relative to the structure be greater just because the hole sizes are larger? subsequent feature controls on either side of the alignment dowels should reference the pair rfs, after all you already have to stack the worst case sizes of the dowels and dowel holes and their position allowances why would you add to the additional position tolerance for mis-alignment, just so that an attribute gage can be built? it's non-functional!!!  
engineering
invite representatives from each of the disciplines above to sit in the same room and disclose how the specification will be addressed in their process.
study gd&t and walk the design, manufacture, inspection, and quality review process to see for yourself how they are actually implemented with your design.  
the end-process print of an inseparable assembly should declare every feature tolerance for material property, form, orientation, and locations including clearances that make that are critical to function. all sub-assembly geometric tolerances are subject to change beyond the level that they are declared. the formidable obstacles in terms of quality responsibility, financial liability, and contractual obligations need to be addressed if the product design and ultimately the customer are to be protected. a final inspection should be performed to verify that all features assumed not to have changed by the assembly process indeed have not in addition to those that are directly related to the assembly.
inspection
even the slightest difference in inspection procedures will often produce remarkable differences in the data. if differences need to be resolved the inspectors on both sides of the dispute need to talk with each other if not see each others procedure.
the producers inspection needs to check the vulnerabilities of the process i.e. roundness when odd-lobbing makes size variation invisible to certain inspection methods.
when datum substitutions are chosen they must be conveyed prominently in the inspection document so that an engineer reviewing the data will understand that either his design, or your inspection is lacking. also so that others comparing their inspection to yours will understand if they made identical or alternate substitutions.
recognize that cmm data can be used as attribute data when variable bonus tolerances are applied. and always disable the "datum shift" bonus tolerance allocation because it typically applies it independently rather than unanimously to every related pattern or simultaneous requirement related feature.
if quality or sta people tell you to do statistical process capability evaluations of variable tolerances "those declared @mmc or @lmc" tell them that the capability formulas only work with constant value tolerances @rfs. furthermore tell them that coordinate tolerance distributions that are well centered on their basic coordinates are always produce skewed "non-normal" distributions and must be transformed or analyzed with the appropriate skewed distribution function typical formulas won't work.
gage-makers: no advice necessary
process gages do not need to check product specification characteristics! in fact gages at any given step in the process should be used to monitor and control the process. process limits are generated by the process variation itself and are not related at all to the product specification the gage datum structure should compliment the process orientation and aid adjusting it when necessary. as far as final gages go: gage-makers are generally acutely aware of our gd&t problems and the differences between rfs and mmc gages even if we are not but it is not in their best interest to point them out. that is why you must sign for the gage design that you want.
machinists "process engineers"
the product drawing should never suggest, codify, or depict all or any portion of the process. if it does it stifles ingenuity, creativity, capitalism. constraints are never-the-less present on most product designs in the form of material properties, thicknesses, and subsequent alteration treatments. sintered iron suggests, etc鈥?br />as long as the finished product meets the specification requirements (and no unintended or adverse problems arise to change them) you are free to streamline and simplify the process to achieve its goal.
sta and quality auditors
you have to help everybody understand that final inspections are important. it is not sufficient to accept process checks sequentially assuming that subsequent steps had no effect on the end product. manufacturing will refuse to inspect inseparable assembly details that were manufactured at another source and considered only as input to their process. these items should be captured and checked as "pass-thru" characteristics. some day designers and engineers wise up and fight manufacturing, purchasing and quality and make "inseparable final assembly prints."
    earlier i gave inspectors ammo to refuse to give you capability ratios on variable tolerances since the equations only work for constant tolerances. they can give you ppm defective predictions but in order to achieve an acceptable level with any confidence the sample size would have to be astronomical (without defects). the process on the other hand could easily be condemned with the same strategy.
    there is another way. the distribution of variable tolerance (consistent with the distribution of feature bonus tolerance) can be compared to the distribution of the geometric deviation with the following equation. (usl+mean[bonus]-mean[pos-dev])/(3*(sigma[size]^2+sigma[pos-dev]^2)^0.5). the equation figures the ppu of a geometric deviation relative to its variable tolerance when both distributions are normal. it works identical to the stress vs. strength reliability distribution model.
walt to answer your original question:
"shawn's question about surface roughness on a screw thread crest
reminded me of an ongoing debate in our industry. some y14.5 committee
randy1111,
   mechanical design, inspection and gd&t require you to have mechanical aptitude, and geometry and problem solving skills.  if you want to follow a few simple rules, you are in the wrong business.
problem:   concentricity (position?)
   i have a housing with two bores which must be located accurately with respect to each other.  
1.] there are machines that do this inspection.  these are feasible if you do a lot of this.  at some point, it is worth designing an inspection fixture.
2.] if you do not have a specialized machine, use your standard inspection fixtures to provide a pin located horizontally parallel to your granite reference surface.  mount your part on the pin, and measure the top and bottom of both holes using a height vernier.  rotate the part around and keep doing the measurements.  this gives you the dimensions across of both pieces, and the positional shift.  there are some weird shapes that will defeat this inspection.  some inspection of roundness would make this process more reliable.
   this is just an example.  the quality of your inspection depends on the tools you have available, on the ability of your people to interpret the drawings and use the tools, and the ability of your designers to produce the drawings properly in the first place.
   check your local college for metrology courses.  these probably will not teach gd&t, but they will teach you how to use the inspection tools.  
                            jhg
pauljackson,
   i believe that designers must consider inspection methods, at least to the extent of being confident the part can be inspected.
   i would like to push the concept of inspectability.  a part is inspectable if there is a realistic way to inspect it, and a realistic expectation that it will pass inspection.   a part is not inspectable if...
...the drawing is missing dimensions and/or tolerances,
...tolerances are obviously unachievable,
...drafting standards are not followed,
...there is no apparent way to inspect the part.
   the inspector can do his job any way he wants, as long as he can do it.  if he cannot, i should be able to show him a way.
   in my gd&t course, the instructor insisted that the part must be fixtured to the datums while it is being fabricated and inspected.  all other fixturing procedures are unacceptable.  if you are preparing drawings, you must at least understand fixturing.
                       jhg
i don't know if it's as simple as a site but just before i started at my company they got training from gary whitmire.
looking through his course notes he emphasized inspection as well as the basic gd&t.
he is heavily involved (think he may even own or at least manage) genium group and this is the section on their drafting zone website that talks about inspection.
ken
assuming you understand y14.5m-1994, dimension & tolerancing, then you may want to check out asme y14.43-2003, dimensioning & tolerancing principles for gages and fixtures.
folks-
the part manufacturer is not obligated to constrain the part as shown on the drawing.  he can be as creative as he wishes to constrain the part any way he desires but "the rubber hits the road" when the part gets into inspection where the part must be constrained according to the drawing.  in short, you can cut corners fixturing the part for fabrication but you can't do the same in inspection.  of course inspection often has more flexibility to produce the datum simulators required by the drawing;  the machinist doesn't have the same flexibility.
tunalover