Usage of Datums in Profile of a Line. 1

[Madhu454]

Hi All,

Few months back I saw a tip on profile of a line with datum referred. The drawing uses profile of a line with datum A to control the line elements of the surface , also the line elements are located to the datum A with basic dimension. Since each line elements are located by basic dimension to datum A, the entire surface of the part is controlled. It is as good as using profile of a surface instead of profile of a line. My opinion is to use profile of a surface itself instead of using profile of a line and confusing the people.

From the above explanation we can come to a conclusion that, using the datum’s in Profile of a line to locate the line elements can be avoided. Does anyone have different opinion?
Now using the datum’s in Profile of a line to control the orientation of the line elements wrt datum’s, In such a case instead of using the profile of a line, we can use any of the required orientation controls say parallelism, and use the text EACH ELEMENT beneath the FCF. This method will be straight forward instead of using profile of a line with datums to control the orientation.

I heard that using the datum’s with profile of a line is very rare, is that true? Also I would like to know any such practical example where we must go for profile of a line with datums?

I would like to know more about the usage of datum’s with profile of a line ? Can anyone help me on this.

Madhu

 

[DeanD3W]

John-Paul,
It sounds like we will agree that profile of a line is not a purely 2D tolerance if applied to features other than linear extrusions.

When profile of a line comes up in a discussion I just like to point out that the tolerance zone is required to be normal to the surface, so for any "tapered" features the tolerance zone must tilt out of the plane of a given cross-section. If we restricted its application to linear extrusion type of features then it would be a 2D spec and I would agree that it could be viewed as a simple way to specify the size or shape of cross-sections of a feature.

We need to extend some of the coordinate system methods introduced in 2009 to allow explicit specification of the orientation of the cross-sectional planes associated with a profile of a line spec.

I would be fine with deleting Concentricity and Symmetry. Profile of a line may be OK though, if we can just improve its definition a bit.

Dean

http://www.d3w-engineering.com

 

[Belanger]

My concern was that if we say profile of a line is 3-D, then that implies that it controls a feature in the depth direction; it does not.

I guess it makes sense to be concerned with the orientation of the cross-sectional planes associated with a profile of a line spec. But we already have a simple solution for that -- datum references.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[Belanger]

Wait ... I'm still thinking. Are you trying to orient the profile of a line zone to tilt as we move back along the surface? (i.e., trying to stay normal to a horizon that's curving away.)

I guess I'm having trouble because of the statement "the tolerance zone is required to be normal to the surface." I don't quite agree with that. A couple of weeks ago there was a thread where we debated how troublesome that concept was for runout!

http://eng-tips.com/viewthread.cfm?qid=310831

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[pmarc]

J-P, Dean,

Just for clarity, per 8.2.1.2 of Y14.5-2009 tolerance zone for profile of a line "is normal to the true profile of the feature at each line element", which in my opinion does not equal to the statement: "the tolerance zone is required to be normal to the surface."

Unfortunately, Y14.5-2009 does not give any example of profile of a line application for features other than linear extrusions, but there is an appropriate example in '94 edition (fig. 6-18). Profile of line control is applied to a view showing true curvature of a surface and the figure shows how the profile tolerance zone looks like. The zone does not tilt at each cross-section, it is normal to the line element and is oriented and located relative to datum reference frame.

 

[CheckerHater]

Belanger,
I think you are reading what is not written there.
The tolerance zone still "flat" and 2-dimensional, the tolerance is MEASURED normal to the surface, while staying inside of the "flat" plane.

 

[fsincox]

"I would be fine with deleting Concentricity and Symmetry."

How did that get in there?
I would not.
Frank

 

[Belanger]

CH -- right, it's a 2-D zone, not 3-D. Thanks for spelling it out better than I was doing. I think Pmarc also reinforced my issue with the "normal to the surface" statement. In my mind the actual surface can do all sorts of crazy things that weren't intended by the designer, so the phrase "normal to the true profile" is the key.

But Dean might still be on to something regarding a continuously curving surface. Think of a bubble or saddle shape to which we want to apply profile of a line. Datum references will tell us how to "slice" the part to check cross-sections, but I wonder if we're again crossing into the territory that was previously discussed with runout and the angle of its dial indicator.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[CheckerHater]

Since we agreed about 2D, I would think about something like CMM probe being oriented to X, Y, Z, rather than "wiggling" the indicator around.
Also, maybe Profile of a surface could be better for heavily curved part? This is what I started with - it's all about the part and what you want to do with it, rather than some formal rule.

 

[axym]

Dean,

I agree with pretty much everything you stated. Here's a question though. If Profile of a Line was applied to a nominally straight continuous extrusion, with no datum features, would the tolerance zones follow the as-produced bend and twist of the part? I agree that the zones would be 2D slices that would "start out" being oriented to each other, but I'm wondering how they can adapt to the as-produced geometry. Adapting to twist would requre the zones to "roll" relative to each other but remain parallel, like a deck of cards. Adapting to bend would require the zones to rotate out of parallel with each other. I realize that the standard doesn't address this but I wondered what your thoughts were.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[DeanD3W]

John-Paul - Please tell me what you think of the three alternative profile of a line call-outs applied in the attached sketch (incomplete-by-intent drawing)... What is the orientation of the cross-sections for profile of a line for each?

pmarc - You are quite correct that "normal to the true profile" is the better wording... "Normal to the surface" was sloppy wording on my part... Apologies to all for that mistake.

CheckerHater - Unless the profile of a line is applied to a linear extrusion, the tolerance zone does not lie in a plane... It tilts out of the plane as needed in order to be normal to the surface. There is no other way for the tolerance zone to be normal to the true profile for a tapered or compound curved surface. I think we will agree that for a linear extrusion profile of a line is OK, but for other features profile of a surface is a better choice (I don't want to deal with a tolerance zone that tilts out of the cross-sectional plane either).

fsincox - Admittedly, the deletion of Concentricity and Symmetry would be a discussion for another thread... I do have what I think are good reasons for saying this.

axym - The issue you're pointing out is very real and at least for me, makes by brain hurt a bit . If a linear extrusion is a bit wiggly (a "wiggly extrusion", is always the case to some degree I suppose) then I can't say explicitly what the orientation of the cross-sections should be be. To orient them to the as-produced surface might be a nice ideal, but it's not often that the work needed to do that will be considered justified. Some sort of median surface is all that I am thinking of that the cross-sections could be oriented to, for more complex feature shapes... I think I would prefer to limit the application of profile of a line to linear extrusions only, then ignore any wiggliness of the extrusion... Clamp them is a vice as best you can, cut them, then measure with a vision system with a dxf overlay... If if the degree of part imperfection is great enough relative to the tolerance value then maybe an uncertainty calculation and guard banding should be implemented..? If a cmm instead of a vision system is used for measurement then I would still want to orient the cross-sections for the data to a simple "vice" method. Too crude for Evan's sensibilities/sensitivities? :^).

Dean

http://www.d3w-engineering.com

 

[CheckerHater]

Dean,

Allow me to disagree.
Per Y14.5-2009 Para 8.2.1.2 "Each line element tolerance zone established by the profile of a line tolerance requirement is two-dimensional (an area) and the tolerance zone is normal to the true profile of the feature at each line element."

It clearly sais "two-dimensional", which is "flat" - there is no tilting.

And "normal to profile" does not mean "normal to surface".

 

[Belanger]

Dean,
The 3rd option is the only adequate one.
In the first case, there is no datum reference. The standard won't say that this is illegal, but the obvious problem is that the sections to be sampled for the lines we trace are assumed to be in the plane of the view. OK, but if that front face that we're looking at is tilted slightly away, then what do we do? Keep the section cuts perpendicular to the floor instead?

The second case has a datum reference which creates a single point. This also doesn't help us nail down the orientation of the planes that we wish to sample.

The third case controls all degrees of freedom, and now we know exactly how to sample the sections for profile of a line. I'm not saying it's required to have datums for all DOF. A single datum reference of the front or back face (the surfaces that intersect the hole) would have been fine.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[powerhound]

Regarding the cross sections of an extrusion; I have always interpreted the tolerance zone of profile of a line, without the use of datums, applied to an extrusion, as being oriented perpendicular to the derived centerline of the extrusion. I understand that a 2D cross section would really constitute a point, to which there really is no perpendicular orientation, but it shouldn't be a stretch to imagine what the orientation of the zone should be.

Powerhound, GDTP T-0419
Engineering Technician
Inventor 2010
Mastercam X5
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[DeanD3W]

CheckerHater - Of course we can disagree since the standard does not make this clear enough. The issue I am pointing out is one the standard should eventually address. As written now, Y14.5 specifically allows the application of profile of a line to tapered features. This is a problem since at some point a feature's taper can be great enough to cause a significant practical issue. I don't think you will want to apply profile of a line to a feature which tilts at 89.99 degrees from a cross sectional plane... Even tilting at 80 degrees is likely unworkable, if the tolerance zone is treated as being in the plane of the cross section. For features other than a linear extrusion, as the angle of tilt relative to the cross sections grows, at some point there would be a significant problem since small form errors will cause very large variation at a given cross section. Just what is the limit to the angle of tilt that would make profile of a line a poor choice? If I have a better alternative then I won't use a tool that has issues. This "tilting issue" with profile of a line is the reason that I believe it should be restricted to linear extrusions. Even if this issue were resolved I still probably wouldn't ever choose to use it since profile of a surface can be applied on a unit basis (incrementally), the same effect can be achieved without the tilting issue, or profile of a line's other issue of the inability to detect a step in the feature from one cross section to the next.

It's not often that profile of a line is a function driven spec anyway. I think the functional need is usually for profile of a surface.

John-Paul - If option 3 is the only one adequate one then we shouldn't use profile of a line on this part... With both A and B referenced all 6 degrees of freedom are constrained, so the profile of a line tolerance zones would stack into an exact equivalent of profile of a surface. We should just use profile of a surface here. If we're dealing with simple rectangular blocks then relying on datum features referenced to provide the orientation of cross-sections might seem to work well enough, but for more complex parts, or if we want cross sections that are at an odd angle to any plane of the datum reference frame, then we're left without a good way to say this. A method needs to be added which clearly and explicitly specifies the orientation of cross-sections, just as we need for line element straightness applied to any ruled surface other than a cylinder.

If profile of a surface tolerance zones are not "location constrainable" then profile of a line would not ever be equivalent to profile of a surface, but I don't think there is anything in the standard that would clearly make this a trait of profile of a line tolerance zones... They're not described as being any less constrainable than profile of a surface zones as far as I know. There may be some implication of this, since the examples shown have combined profile of a line with directly toleranced dimensions, but that is only further illustration of my point that profile of a line is not well enough defined in the standard.

Dean

http://www.d3w-engineering.com

 

[Belanger]

Dean, I don't think we're that far apart. You bring up a good point about the difficulty in being normal to a true profile if the surface is a compound curve. (I've waged that battle in regards to runout.) And I hope you see why I like the datum references for profile of a line, in order to constrain the orientation of the cross-sections. However, I didn't intend for the distance to the datums to be basic. (Sorry... I shouldn't have said all DOF; I meant just the rotational/orientation degrees.) If the location of the curve from the datums is kept as ± tolerances, then profile of a line is not equivalent to profile of a surface.

However, it seems that many people seem to think that profile tolerances must always be located to the datums with basic dimensions. That's reading something into the standard that is clearly not there.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[axym]

CheckerHater,

The 2D vs. 3D issue is a tricky one. "Two-dimensional" is not synonymous with flat. In mathematics and geometry, a curved surface is considered to be two-dimensional even though it does not exist in a two-dimensional plane. By the same token, a curved line (e.g. a Derived Median Line) is considered to be one-dimensional.

The standard states that "the tolerance zone is normal to the true profile of the feature at each line element". I would take "true profile" to mean the as-designed surface, and therefore get zones that may tilt out of the cutting plane that defines the line element. But I can see how you (and many others) would take "true profile" to mean the intersection of the cutting plane and the as-designed surface, and therefore get zones that are all flat and parallel to each other.

So I would agree with Dean that the tolerance zones would in fact tilt relative to the cutting planes, for geometry other than continuous extrusions.

powerhound,

Your interpretation of Profile of a Line for a continuous extrusion makes it behave in a similar way to Circularity, in which the zones must be perpendicular to an arbitrary "spine" (which is a line that isn't necessarily straight). This is perfectly reasonable and practical, as it lets the cross-sectional tolerance accommodate the as-produced bend of the feature. We can go the simple and easy route and choose a straight spine, or we can do the extra work and extract one that follows the feature's as-produced centerline as you mentioned. Our problem is that orienting the zones to a curved spine is one of many possible interpretations, and the standard does not provide any guidance.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[CheckerHater]

Dean,
I still don't understand what the big deal about taper is.
Unless we are talking about different things; then please specify the scope of the problem: is it impossible for the mankind in general, or is it impossible for the machines you are selling?

Evan,
2-dimensional means "the one that can be described with 2 coordinates" and yes, indeed it is flat. I would greatly appreciate if you direct me to college-level math book that states otherwise.

 

[Belanger]

Evan, I'm with CH that 2D means flat, but I might be missing something. I don't understand how a curved line (think of a drawing on an Etch-a-Sketch) can be considered as only 1D, and a wavy surface with depth can be considered as only 2D. Then what the heck is 3D?

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[CheckerHater]

Forget the college book, let's get closer to home.

From ASME Y14.5.1M-1994:

"A profile is the outline of an object in a given plane (two-dimensional figure). Profiles are formed by projecting a three-dimensional figure onto a plane or taking cross-sections through the figure."

For anyone resonable, including members of ASME committees, 2-dimensional = PLANE; 3-dimensional figures are projected onto PLANE, to become 2-dimensional.

2-dimensional IS "flat", profile IS NOT surface, one cannot live in denial forever.

 

[axym]

I'm sticking to my guns that 2D is not equivalent to flat. Part of the problem is that Y14.5 uses certain terms in an inconsistent way, and does not always follow the way these terms are defined in mathematics and geometry.

John-Paul,

A curved line is 1D because it has no thickness or width, only length. A wavy surface is 2D because is has no thickness, only length and width (and therefore area). 3D entities are solids that have length, width, and thickness (and therefore volume). I didn't invent this - I'm just the messenger ;^).

CH,

The Wikipedia entry for "surface" has some good statements in the first couple of paragraphs (I realize that this isn't an actual textbook). Here's a quote that mentions the 2D nature of surfaces:

"In mathematics, specifically in topology, a surface is a two-dimensional topological manifold. The most familiar examples are those that arise as the boundaries of solid objects in ordinary three-dimensional Euclidean space R3 — for example, the surface of a ball."

Armed with this understanding of the proper math concepts and terminology, we can nit-pick the errors in the passage from Y14.5M-1994 that you quoted:

"A profile is the outline of an object in a given plane (two-dimensional figure). Profiles are formed by projecting a three-dimensional figure onto a plane or taking cross-sections through the figure."

An outline of an object in a given plane would be a line, which is a one-dimensional figure. The figure that is projected onto a plane or has cross sections taken through it would actually be a surface, which is two-dimensional. So here's a corrected version:

"A profile is the outline of an object in a given plane (one-dimensional figure). Profiles are formed by projecting a two-dimensional figure onto a plane or taking cross-sections through the figure."

So I agree with you that the statements in Y14.5M-1994, if taken as gospel, would lead a reasonable person to conclude that 2-dimensional is flat. But I continue to live in denial as you say. The members of the Y14.5 committee are also reasonable, at least most of the time - right Dean ;^) ? But they are not mathematicians, and so their application of terms and concepts from math and geometry is not beyond question. These things are much better dealt with in the Y14.5.1 Mathematical Definitions standard.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca