Real-world application of ALL OVER profile tolerance

[Nescius]

The concept of an ALL OVER profile tolerance is easy to understand in cookie cutter examples, but I'm very interested to learn how the forum membership has used and seen this tool in the real world.

From the 2009 standard, 8.3.1.6:

"A profile tolerance may be applied all over the 3 dimensional profile of a part unless otherwise specified."

As always, the phrase "unless otherwise specified" is problematic. Specified how?

Let's consider a cube, fully defined with basic dimensions, with an ALL OVER profile tolerance. If I apply a refining, tighter profile tolerance or flatness tolerance to a single face of the cube, is it no longer located/oriented relative to all the other faces?

It would be easy to say that the ALL OVER tolerance still applies to all surfaces and the single face in question must simply meet both tolerances...however, what if I desire to allow one face of the cube to vary by more than the ALL OVER tolerance? Meeting both tolerances becomes meaningless, then. Furthermore, if the more generous tolerance qualifies as "otherwise specified", and the ALL OVER tolerance is trumped, is the face in question no longer located/oriented relative to the other faces at all?

Must the part be fully defined with basic dimensions, no +/- allowed? This is a much-debated topic itself, +/- dimensions and a profile tolerance applied to a closed outline...profile not controlling size.

In theory, the ALL OVER profile tolerance is a powerful tool...perfect for a part with some very complex aesthetic or ergonomic surfaces, but also a handful of more conventional features that must be controlled differently.

Is this a gray area with open interpretation or am I missing something?

 

[mkcski]

The lack of a Standard (definition) is exactly why you two guys are having this extend discussion. And, seriously, is really thought provoking and valuable. I will be paying more attention to these "global" conditions and how exceptions to them are interpreted.

Certified Sr. GD&T Professional

 

[pmarc]

The most solid thing that could ever be used - logic (not like I am saying you are not thinking logically) ;-)

But seriously, do you really believe that in fig. 4-43 the UOS profile requirement is not nullified in case of all those features controlled by additional geometric callouts?

 

[pmarc]

mkcski,

In the thread CH referred to in this discussion (see one of his previous posts) I said: "in my opinion one of the most important takeaways from it [the entire discussion] is that the UOS notes must be used wisely. It is not that they should not be used at all. In some cases they can do really great job, but sometimes they may get us into serious unexpected troubles."

 

[mkcski]

Pmarc:

Thanks for refreshing me and I totally agree.

Certified Sr. GD&T Professional

 

[CheckerHater]

I am just trying to find the best place to draw the fine line.

To me UOS means "if local requirement contradicts default requirement, the local requirement takes precedence".
If UOS is Ra128 and local is Ra63, then yes, indeed, local spot shall have better surface.

The gray area starts when there is no contradiction. It is possible to have Profile .05 all-over and Flatness .02 someplace (that still satisfies Profile .05)

Same with indirect requirement. Size may control many things, but it doesn't mean they are all explicitly applied.

This is why logic is not enough.

To me, when there is a possibility of misinterpretation, it's designer responsibility to specify, where UOS will be nullified, like "All-over this side of parting line" or "All-over as-cast surface"

At least, this could be less ambiguous.


"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[mkcski]

My thinking is: I prefer to use the word "refinement" (of the default) as opposed to nullifying it, because the (global) default requirement is always in effect.

Certified Sr. GD&T Professional

 

[CheckerHater]

That's basically in-line with my thinking: If requirements collide heads-on, you pick one of them. Otherwise (no pun intended) it's refinement.

Unfortunately "refinement" is defined in the standard as well as UOS

"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[mkcski]

Checkerhater:

The phrases UOS and USO, and the word refinement are used frequently in the 2009 Standard. But I have not seen a specific definitions. Please clarify.

Certified Sr. GD&T Professional

 

[CheckerHater]

That's exactly what I mean - no definition for both. (also for “default”, “direct”, “indirect”, etc.)

"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[mkcski]

My bad. I missed your Smilee Face

I checked and they are missing in the DRAFT of the next release. Oh well

Certified Sr. GD&T Professional

 

[pmarc]

For what it's worth, here is another quote from another related discussion

http://www.eng-tips.com/viewthread.cfm?qid=409569:

This is taken from the new draft of Y14.5, but I am more than sure that I saw this also in one of the other, already published, ASME Y14 series standards (I think it was Y14.41, but it would be great if someone could confirm):
"1.4.9 Unless Otherwise Specified (UOS)
The phrase "unless otherwise specified" or UOS is used to indicate a default requirement. The phrase is used when the default is a generally applied requirement and an exception may be provided by another document or requirement."

Are we now going to discuss what exactly "exception" means for each particular dimensioning and tolerancing scenario? ;-)

 

[CheckerHater]

the default is a generally applied requirement and an exception may be provided by another document or requirement

If default is Profile, is Flatness exception or refinement?

"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[JNieman]

I believe that if you /replace/ the profile callout with flatness, you've lessened the definition of the part. That feature may no longer be otherwise located. Thus I think REFINEMENT would the more logical treatment. Though I think this conversation has exceeded practical application, it's kind of interesting.

I hate "unless otherwise specified" PROFILE FCFs as notes (or worse... just writing "profile .020 applies u.o.s.") more than I appreciate their efficiency. It seems to be overwhelmingly used as a crutch or C.Y.A. rather than to convey the design requirements.

 

[Nescius]

For parts where I work, I would prefer that an all over profile tolerance simply apply. Period. Any additional tolerances must ALSO be met. Overlapping, but not conflicting, not UOS. My parts are usually very small, an overall aesthetic shape with a handful of functional features. The tolerancing strategy is usually simple enough that I can "see" the downstream interaction between the all over tolerance and the others. If there are features that I don't want the all over tolerance to apply to, I simply use a note under the all over FCF.

Since the standard says:

"A profile tolerance may be applied all over the 3 dimensional profile of a part unless otherwise specified."

One could argue that UOS applies by default. How would you all feel about a note, "All over tolerance shall apply regardless of other tolerances.", under the FCF to negate the (possibly) default UOS nature of the all over control?

 

[drawoh]

Nescius,

The company I work for has switched from detailed fabrication drawings, to Model Based Definition (MBD). We prepare a drawing, but it does not have dimensions. We have a note on the drawing that unless otherwise specified, everything is to be fabricated to a 0.4[ ]profile WRT the specified datums. I hate MBD, however, this is an application of an all-over profile tolerance.

If the standard all-over tolerance is not suitable, we apply tighter tolerances to critical features either with new profiles, other GD&T FCFs, or by applying feature-of-size dimensions and tolerances.

--
JHG

 

[Nescius]

Thanks, drawoh.

I'm not experienced in MBD at all, but I can see why you hate it...drawings with no dimensions. I'm trying to make use of the all over profile tolerance, but the drawings I'm using it on have all the dimensions needed to check all the tolerances I apply. The only exception is the all over tolerance; you'd need the solid model to check that.

 

[drawoh]

Nescius,

MBD is an issue separate from all-over profile tolerance. What is wrong with applying a set of basic dimensions completely describing the part, applying datums, and providing a note that states that, unless otherwise specified, everything must be made to a profile of 0.8mm?

--
JHG

 

[pmarc]

If default is Profile, is Flatness exception or refinement?

Regardless of both tolerance values, flatness is exception to default profile. I am not sure what the word "refinement" has to do with it.

How would you all feel about a note, "All over tolerance shall apply regardless of other tolerances.", under the FCF to negate the (possibly) default UOS nature of the all over control?

I will say this for the third time in this post ;-). Simply attach the all over profile callout to any feature of your part and do not use notation "UOS" or "UNLESS OTHERWISE SPECIFIED" at all. Then apply as many additional geometric callouts as you need and you are done.

If you are still afraid that it might be misinterpreted, your idea ("All over tolerance shall apply regardless of other tolerances.") should work fine. The other way of saying pretty much the same thing could be: "This tolerance applies in addition to all other tolerances".

 

[Nescius]

Thanks, pmarc.

What is wrong with applying a set of basic dimensions completely describing the part, applying datums, and providing a note that states that, unless otherwise specified, everything must be made to a profile of 0.8mm?

Of course, nothing would be wrong with this, assuming UOS doesn't cause confusion. The only caveat is that some geometry is deceptively difficult to describe on a traditional drawing.

In my case, some of this difficult-to-describe geometry actually needs to be actively and thoughtfully controlled. The features I'm controlling are aesthetic details on a type of consumer good where "fit and finish" is highly scrutinized. It's a subjective distinction, but I'm using the all over tolerance in a more active sense, as opposed to a failsafe.

Often, the most difficult choice I face when tolerancing a part is trying to determine how round, straight, flat, or symmetrical something needs to be in order to look good.

 

[3DDave]

You probably would be better off by pointing at the surfaces to be controlled with a note that says manufactured part shall deviate from the mathematically defined surface as supplied in a STEP file relative to certain datums by no more than a certain amount.

One could also discard the datums if the part is a completely free item, such as a hand-held device that mates with no other parts and has no particular orientation or location needs.

The creeping definition of Profile to include things that aren't 2D projections is apparently an end run that muddies the conventional use of Profile.

Aesthetics isn't one of the variables directly controlled by Y14.5. For that you need to specify things like - surface variations in terms of deviations from ideal curvature, which are more noticeable than displacements from ideal form.

Tougher are conditions where two parts meet and it doesn't matter much what the variation is, but the relation between the two parts does matter, basically requiring the tooling of the multiple parts to be coordinated in some way.