Tolerance strategy for precision molded parts

[PaulJackson]

Attached is a specification I propose for limiting the profile error permitted within a specified area... while permitting the over all tolerance to constrain the contour within the larger boundary. Naturally the lower portion of the composite is only oriented to the datum features.

Paul

 

[PaulJackson]

With the pace of product development these days where prototype cast or molded structures are printed on a 3D plotter, patterns are grown and generated from the solid model, casting or molding prototypes are delivered before the layout stacks are completed and the profile inspection comes in the form of whisker plot comparisons to the solid model… we could really use a simple yet robust method to constrain the cast or molded geometry, that recognizes the inherent variability of cast and molding processes in the contour extremities, but also recognizes limitations on local contour detail for function and material weight savings.

Simply using overall profile tolerance means that if one has those additional concerns… then additional profile constraints need to be detailed on the drawing. If those details conflict with the predicted variability in curing profile change… then the contour may need post processing.Historical cast tolerance methods “shrink equation allowances” have their unique problems in that they are origin dependent, i.e. if two features have functionally similar requirements yet one only one is found non-conforming due to its proximity to the measurement origin… the strategy often fails to support function. The ISO categorical tolerance strategies are sensitive to dimension magnitudes regardless of their proximity to the origin… but like “shrink equation allowances” they fail to establish the fundamental coordinate system necessary to constrain measurement orientations and translations to a common origin.

The potential advantage of this composite profile strategy is the establishment of that datum reference frame, a uniform tolerance relative to the component extremity, and window refinements “as necessary” to limit subordinate sizes and contour detail… one caveat is that those refinements need to be complementary to the predictable variation for the size of that window. Another is that readers of the drawing need to understand composite feature control frames to understand the translation liberties in the refinements.

Many designs position datum feature targets on a cast or molded structure so that the coordinate system X0, Y0, Z0 reference is on the extremity of the structure and… if the requirement for lowest functional variability is nearest that origin and increasing as it departs from that origin I would say that is good… but from what I have seen in castings over the years that is seldom the case. I typically try to define targets positioned on the extremities to stabilize the structure but then position and equalize translational targets on the structure so that the origin nearest the structure center. By doing so the effects shrink variation relative to the origin are minimized. Sometimes I put all of the targets in the one cast section corresponding with the most critical cast attributes to minimize mismatch and closure variability and sometimes I put them spanning sections to equalize that variability as function requires.

I envision that this profile strategy will be inspected just as it is written where there is a global or over all tolerance wrapped bilaterally about the basic contour that all points of the surface must reside within, and then a narrower bilateral zone of given window boundaries… that is oriented to the composite DRF as specified but able to translate within the global zone… that all points of the surface within the window must reside. In effect it can be like using a cross-sectional overlay with overall boundaries, oriented and located to the DRF… and another cross-sectional overlay with narrower boundaries oriented to the DRF but able to translate within the larger zone.
Paul

 

[fcsuper]

You may wish to invoke the ALL OVER requirement instead of using "ALL SURFACES". It is covered in ASME Y14.5-2009

Matt Lorono, CSWP
Product Definition Specialist, DS SolidWorks Corp
Personal sites:
Lorono's SolidWorks Resources & SolidWorks Legion

 

[ptruitt]

I am still thinking about cubical vs. spherical tolerance zones. One of the attached illustrations shows a cube, with a basic surface and its boundaries, similar to Paul’s. The other illustration (same PDF) shows the same basic surface and boundaries, but clipped by a circumscribed sphere. My contention is that the boundaries of each are likely to work equally well, functionally, in the design. (The same argument as square vs. circular tolerance zones.) The table below the illustrations show how the surface area of the two boundaries, as clipped by the sphere, have a more linear variation from the basic surface than the cube. Don’t ask me to explain how one might measure either of these in practical applications, because I don’t know. Does this make sense to anyone else?

Peter Truitt
Minnesota

 

[MechNorth]

J-P, my automatic assumption (perhaps erroneous, but there it is) is that the areas (40x40) would be based on xy, yz, xz planar projections as established by the DRF (a more evolved concept in '09). Topographically, as viewed from one of these planes, the area would be equal, however the actual surface area would vary with the topography.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com