Thin Datum Surfaces 5

[AMontembeault]

I have a sheet metal part, which is more or less square, with a bilaterally symmetric, sinusoidal contour cut on one side. I set my primary datum "A" at the large planar surface of the part, secondary datum "B" as the midplane of the part, and tertiary datum "C" as the edge opposite my contour. The contour is controlled by line profile with respect to A|B|C.

The problem we're running into pertains to ASME Y14.5-2018 section 7.9, specifically that all datum features must be controlled by appropriate geometric tolerances and/or size dimensions. Putting flatness on A, and perpendicularity on B with respect to A, and perpendicularity on C with respect to A and B makes sense in theory, but in practicality its proving to be very difficult to measure, given how thin the part is and how little real estate is available to probe with a CMM.

The only way I can think of to inspect this part is through the use of fixtures/functional gauges, but was wondering if perhaps anyone had run into something similar and might have better ideas. A change in the GD&T, or a different method of inspection?

 

[TheTick]

datum "B" as the midplane of the part

No such animal exists. Features or faces define datums. Midplanes are derived from those specific features.

Honesty may be the best policy, but insanity is a better defense.

http://www.EsoxRepublic.com

-SolidWorks API VB programming help

 

[AMontembeault]

No such animal exists. Features or faces define datums. Midplanes are derived from those specific features.

I should perhaps be more specific. The B datum is the median plane derived from the width of the part.

 

[TheTick]

Sorry if I was too pedantic.

Sounds like datum B is best controlled by profile or position, maybe with straightness for good measure. Trying to nail down perpendicularity on a sheet edge is folly.

Since B is secondary, you could bump the edges with posts. Probable best to define B with targets.

 

[3DDave]

I would ignore that directive from the 2018 committee. They mean well but this is a problematic requirement.

 

[pmarc]

Assuming this would be OK from functional standpoint, a trick that worked for me a few times was getting rid of datum features B and C, defining the entire contour with basic dimensions and applying surface profile tolerance wrt A all around the part. This usually closed the conversation about inability of inspecting the perpendicularity error on datum features B and C - for some unknown reason, inspecting in a single plane only was then sufficient to everyone.

 

[3DDave]

The downside to the all-around approach is that it leaves a lot of acceptable variation on the table in favor of overly tight requirements - it also may not correctly represent the case of when the width of the part is clamped and that clamping defines the in-plane orientation of the part - a constraint that, if accepted under the all-around approach, may force the sinusoidal (in this case) form out of position.

 

[pmarc]

That is why I started with: "Assuming this would be OK from functional standpoint".

 

[Burunduk]

pmarc, I'm intrigued by the "unknown reason". If the CMM can't probe the thin edges for perpendicularity inspection, how do they inspect profile on those surfaces? The tolerance zone is of the same type (other than spanning all-around) and the relationship with the datum is the same as with perpendicularity.

AMontembeault,  suppose you want to make the CMM inspector's life easier, and you don't require perpendicularity. What's the alternative? Plus minus tolerances from the general tolerance block on the 90° angles? Why should the CMM probe that failed to verify perpendicularity be able to verify these? Or do you consider leaving your datum features totally uncontrolled with reference to each other? 

Geometric tolerances provide unambiguous definition of geometric limits in the 3D space of the part. They don't force you to inspect the part with specific equipment or methods. 

 

[3DDave]

I assumed that the assumption wasn't as good as making the trade-off explicit.

 

[pmarc]

pmarc, I'm intrigued by the "unknown reason". If the CMM can't probe the thin edges for perpendicularity inspection, how do they inspect profile on those surfaces? The tolerance zone is of the same type (other than spanning all-around) and the relationship with the datum is the same as with perpendicularity.

That's exactly the intriguing part. I would have thought the same thing, but apparently being able to measure in 2D other characteristics that the surface profile is able to control, like overall size and form of the part, is enough.

It is sort of similar to a situation where a position tolerance at RFS is applied to a hole in the sheet metal part. People usually don't have issues with this kind of callout even though technically it requires finding the UAME axis of the hole, not just its 2D center point, to be able to properly verify it.

Don't get me wrong, just like you I agree that the drawing should have all necessary perpendicularity tolerances applied to the datum features B and C regardless of the thickness of the part. I merely wanted to provide an alternative solution that does not violate the requirements of para. 7.9 in the standard, yet may be easier accepted by those that have issues with the requirements.

 

[chez311]

I have to assume it has to do with the fact that you're not asking someone to directly quantify the orientation of a surface and the associations we have between "orientation" and "angle", even if that same requirement exists for profile - for example even if theres no practical difference between a perpendicularity tolerance of a planar surface and surface profile of a planar surface where no datum features constraint rotation/orientation (both control form/orientation) intuitively people likely associate them differently. People see orientation (parallelism/perpendicularity/angularity tolerances) and associate that with measurement of an angle even if though the definition in Y14.5 only utilizes the basic angle of the feature and does not require the output of a discrete angle measurement of the feature.

As a result its enough for most inspectors to take only a single 2D section for profile and be confident that the process used (stamping, laser/waterjet cutting, machining) for a thin sheet metal part created a surface which is reasonably perpendicular/flat and in most cases thats a reasonable assumption - and even so, many applications could probably tolerate a notable amount of angular deviation as long as the projected profile falls within tolerance. But slap a perpendicularity tolerance on there and suddenly they feel they need enough points to quantify/measure that angle.

 

[pmarc]

Well said, chez311.

That is why I like to call the approach I proposed "psychological approach to dimensioning and tolerancing" ;-)

By the way, once switched to basic dimensions and the surface profile approach, all sorts of combinations of profile callouts wrt A for particular portions of the entire contour can be applied to better diversify the amount of allowable tolerance yet keep everything as simultaneous requirements.

 

[3DDave]

It's not the psychological problem for the inspector here - it's CMM software that cannot manage a thin feature in evaluating perpendicularity.

In 40 years I never had any inspector check the default angle tolerance and it didn't appear on any inspection reports. Ever. I saw a full DoD program that didn't even have that tolerance on the drawings and not one case directly applied to any right angles.

One combination that cannot be allowed by profile is that one side of the width can be particularly straight and all the variation in width is allowed for the opposite face per the limitations of a feature of size. Adding "perpendicularity" doesn't help - it narrows the range of potentially usable variations.

The clear side-step is to put the item in the CMM with 3-2-1 blocks, gauge blocks, or angles on both faces and push them into contact and probe those surfaces - the datum feature simulators - instead.

 

[pmarc]

It is a psychological problem if it works with surface profile but doesn't with perpendicularity.

If I am understanding your last paragraph correctly, you are describing the establishment of datum reference frame, not the verification of the actual geometry of the datum features.

 

[3DDave]

It's a CMM - it doesn't have a psychology. What you refer to is acceptance by people willing to change the function of the description to match their own limited understanding - a psychological problem, but not the one the OP has.

Like all inspection, the inspector can look to see if there are gaps that would need a feeler gauge to confirm. Since the vertical projection of the zone can be checked that way it's fine to confirm the application.

 

[pmarc]

It's a CMM - it doesn't have a psychology. What you refer to is acceptance by people willing to change the function of the description to match their own limited understanding - a psychological problem, but not the one the OP has.

One of OP's questions was if the problem could be avoided by "a change in the GD&T". I gave the change proposal that worked for me a few times. Just by changing the geometric control type from a pure orientation tolerance to a surface profile tolerance (where orientation is just one of the components of the characteristic), CMM inspection and the assessment of the feature conformance have suddenly become possible without having to worry about the thickness of the part. So something must have changed even though the drawing hasn't stopped to require all elements of the thin side faces of the part to lie within the tolerance zone and that is why I called it psychological as it was used by me exactly to affect people's "will to change the function of the description to match their own limited understanding".

 

[3DDave]

If the CMM software wasn't defective this would not be a discussion. Coping with bad software is a problem as it reinforces acceptance of bad software.

 

[pmarc]

I don't think it is a problem of the CMM software. OP described the problem in the initial post: "[...] but in practicality its proving to be very difficult to measure, given how thin the part is and how little real estate is available to probe with a CMM."

 

[Burunduk]

pmarc, chez311,
Regarding the solution of applying surface profile  on the very thin peripheral faces of the sheet metal part:
If everyone is aware and accepts that the profile tolerance will be verified only at a single 2D section (at some probe-contactable height above datum A), and that the normal orientation of the surfaces relative to datum A will not be controlled, wouldn't it be more appropriate to change the drawing tolerance to profile of a line, to reflect that? Although profile of a line is usually checked in more than one cross-section, I don't think it is a mandatory requirement for all cases.