Controlling "sharp" edges through geometrical controls 1

[semiond]

Searching the forum for information regarding application of GD&T on edges, i found thread1103-310372, where it seems that most participants agreed that edges (including those that connect surfaces at an angle smaller than 90 deg.) can legally be used as datums (recommended or not - is a separate question).

In relation to ASME Y14.5 2009, my question is: could the two edges which the OP of that thread intended to assign as datums, be controlled for orientation between each other by the use of perpendicularity control? If i was asked this question my answer would be "probably not" because the definitions in chapter 6 seem to only mention a surface, center plane, or an axis as the subjects for orientation controls. On the other hand, maybe the fact that it's not mentioned in the definitions, doesn't mean it is not allowed? As for inspection - i don't think it would be too difficult with an optical comperator or a microscope (simulating a tangent line to the datum feature edge and then finding the size of the zone within which the other edge lies).

For approaching this issue, please assume that the uncertainty related to deburring/ rounding size removed from the equation, since the rounding/deburring are small enough and tightly controlled.

 

[DeanD3W]

semiond,
Sorry to be late for the party here. I'm just getting back to "normal life", but my time to login here is still a bit spotty. You point out a gap in the standard. When you need to apply straightness or an orientation tolerance or profile (I will say profile of a surface, since I think profile of a line is problematic) to an edge, the way I have seen and agree with is to put a modifier below the feature control frame "PROJECTED EDGE".

This is not in the standard yet, but I think it should be. There is a new standard in the works that will be about "Force and Direction Indicators" (that's sorta close to the correct title). For now, since we don't yet have the tools from that standard to use for the projection direction spec, for cases where the projection direction is not clear enough, or if you just like to be more explicit, then I think a note may do. With a PROJECTED EDGE modifier and maybe a flag note that says something like "PROJECTION DIRECTION IS NORMAL TO DATUM A", or maybe "PROJECTION DIRECTION IS NORMAL TO YZ[A,B,C]" (the YZ plane of the datum reference frame A,B,C) then all should be well.

The tolerance zone would be just as you said, two parallel planes that are parallel to the projection direction. This is not yet supported in Y14.5, but I am intending to do what I can to get it added asap (we can talk again in 5 to 10 years )

To bring up the use of an edge as a datum feature (with notes added to clarify, since this is not supported by the standard), I think it would be good to keep in mind that this may be completely fine if the part will be measured with a vision system or other non-contact method.

The issue you have brought up is a very good one. I hope this helps.

Dean

 

[semiond]

DeanD3W,
This is extremely helpful, thank you.
I certainly want to adopt the "PROJECTED EDGE" modifier.

I'd like to ask why you think that Profile of a Line is problematic in this case?
I can assume that since there is no actual surface with significant area to deal with in case of an edge, it will feel more natural to specify Profile of a Line than Profile of a Surface. However, i can see that it might be problematic to redefine the profile of a line control to give origin to a volumetric tolerance zone rather than a 2D area. Is this also your concern?

About the new standard in the works that you mentioned - will contents from it be integrated in the future revision of Y14.5?

Thank you again for the input. This is both helpful and encouraging.

 

[drawoh]

semiond,

I am still not sure there is a problem here. Industry practice is to make sharp edges not sharp, obviously for safety reasons. There must be some reason you want a sharp edge. If the edge is sharp, you do not want to touch it for fear of damage to you, to your equipment or to your edge. The sharp edge has to be due to some functional requirement you probably can test for.

Above, I note a test for a blade edge. You specify a material, a cutting force and resultant depth of cut. This test probably is destructive to the blade.

If your sharp edge is part of an optical beam splitter, you test for clear aperture, and for any weird optical effects that result from anything other than light passing through a perfect sharp edge.

--
JHG

 

[semiond]

drawoh,
Testing functional edges for sharpness is a separate topic, and there are various types of specifications and testing/inspection methods for this - the ones you mentioned are indeed good examples.

However, sharpness validation doesn't cancel the need to define and validate the location, orientation, and form of the functional sharp edge relative to the part's datum features. This can not be carried out through a cutting test. The inspection should of course be performed by a non-contact method.

 

[DeanD3W]

semiond,
Yes, I think profile of a line would be problematic for controlling an edge because its tolerance zone cannot currently be the needed boundaries that extend infinitely in the projection direction. A given cross-section that profile of a line would apply to can be assumed to be perfectly flat, so the tolerance zone lies in, or at least originates from, that cross-sectional plane. To control an edge with a "PROJECTED EDGE" modified tolerance, a tolerance zone that accommodates the lack of straightness of the edge in the direction that is parallel to the projection direction, we need the two parallel planes you described, or for more complex edge shapes, at least two boundaries that extend some sufficient distance in the projection direction.

I think the PROJECTED EDGE modifier (not yet in Y14.5, as you know) will need to be limited to edges that, by design, lie in a plane, with the projection direction being normal to that plane. More complex edges, such as with a "non-flat" razor blade would need a more complex control. As step in the right direction that will provide a useful tool in many cases will be to get PROJECTED EDGE into the standard.

drawoh,
I agree with semiond about the need for a dimensional tolerance that can be applied to an edge. An example I always bring up is for a producer of razor blades. The location, orientation, and form of an edge wrt a datum reference frame that is established from the features that mount the blade in the handle or holder would be very important, and also likely to be measured with a vision system, as opposed to any type of contact measurement method.

Tolerances can be applied to the two surfaces that create the edge, but for as the angle between the surfaces gets tighter, the resolution of those tolerances, so their ability to control the edge location, orientation, and form, is reduced. It will be more efficient to apply a direct tolerance to the edge. If PROJECTED EDGE, supported by a note to explain what it means provide that tool for now, then all we need to have it included in Y14.5.

Best Regards,
Dean

http://www.validate-3d.com

 

[3DDave]

Profile of a surface is sufficient for the flat-view of a sharp edge. There are other controls that apply to the intersecting faces, depending on exact configuration, that are sufficient to the task. There's no need for 'projected edge' as the only requirement is the profile of the part lie in the profile zone. Under sufficient magnification any real edge is a surface made up of lumpy atoms, often arranged in lumpy crystals.

It doesn't matter that the current descriptions have reduced control - it's a precision operation, so every description has reduced control. I never got the concept that having a magic wand description would somehow transform a difficult manufacturing process into an easy manufacturing process. Before calling to add a new control, see if the existing ones already work by understanding the exact problem that is to be solved. Since the current ones already work, it seems like it must be a lack of understanding of the current problem.

What you want to ensure an edge is sharp is a surface roughness callout and a small radius limit, in addition to the subtended angle.

"The measured cutting edge radius values for carefully manufactured diamond tools were in the 20 to 45 nm range. This method was also applicable for other sharp edge samples, having even under 45 nm sharpness."

https://www.sciencedirect.com/science/article/pii/S0007850607610087

Break out your SEM to check your work.

Also of interest:

https://northarmknives.com/razor-sharp/

which links to

https://northarmknives.com/wp-content/uploads/2016/01/knifeshexps.pdf

 

[semiond]

DeanD3W,
I agree about Profile of a Surface. It is apparently better to rely on a control that already establishes a 3-dimensional tolerance zone. Do you also think that flatness should be used rather than straightness? A bit non-intuitive too, but maybe unavoidable?

I think the PROJECTED EDGE modifier (not yet in Y14.5, as you know) will need to be limited to edges that, by design, lie in a plane, with the projection direction being normal to that plane. More complex edges, such as with a "non-flat" razor blade would need a more complex control.

Often there are edges that should appear to be curved in one view and and straight in the view perpendicular to it. That is due to the shape of the intersecting surfaces. The "straightness" of the edge on the view where it should appear straight is actually very important. The projection by the vision system or the optical comparator is done in the direction at which it appears straight. My initial assumption is that a PROJECTED EDGE modifier should be sufficient for these cases too, even though those edges don't lie on a plane nominally. In fact this is the exact way how inspection of geometric tolerances on edges is done already - without much regard to the fact that it's not supported by the standard and that those controls are currently defined only on surfaces. I'm interested to know - why do you think that a more complex control is needed for these cases?

I really am looking forward to the PROJECTED EDGE modifier to be introduced in the Y14.5 standard

 

[semiond]

3DDave,
See my last reply to drawoh.
Sharpness and wholeness of an edge have very little to do with the subject matter. Just like surface roughness doesn't replace geometrical tolerance controls on a surface, similary specifications of sharpness (see: "edge preperation") are not a substitute to a geometrical definition of an edge. But thanks for the interesting links.

 

[3DDave]

semiond, You don't have to thank me for links you ignore. You don't even have to remind me to read your previous replies, for whatever they are worth. The links I posted included the 'non-contact' measurement methods you demanded from drawoh.

 

[semiond]

3DDave, I didn't demand anything from drawoh. There are enough non-contact measurement methods to choose from and I was not asking for metrology advices.
I thanked you for the links because they contain interesting information, although not directly related to the discussed topic - which is geometrical controls (feature control frames) applied directly to sharp edges, and how this can be done in a drawing without abusing the definitions of the Y14.5 standard.
One more thing - I never implied that it is impossible to control edges through the intersecting surfaces they connect between, using the regular surface controlls well defined in the standard. It's just that it is often the least practical and efficient scheme.

 

[3DDave]

semiond, care to clearly demonstrate the contention "it is often the least practical and efficient scheme." That's a bold claim to make and you have offered no proof that you understand how to approach a solution of any kind beyond waving a magic wand of undefined non-existent specification. Trust me - I was not giving you any "advices" (sic). It was a comment to Dean.

 

[chez311]

semiond - I think I echo some of 3DDave's concerns.

I think you'll find that for many applications, especially cutting implements, the geometry of the surfaces beyond the actual edge, which make up the edge itself, are also very critical. I guess I also don't understand why profile of a surface or profile of a line of each cross-sectional element (or perhaps standard controls like flatness paired with a traditional angle/radius tolerance) won't suffice instead of trying to control the exact "edge" itself - which really won't ever truly be a line anyway and trying to characterize it as such may lead to ambiguity and issues in measurement, especially as you deviate from a theoretically perfect "razor sharp" edge. Considering the edge is directly related to the surfaces which intersect to create it - any deviations in the those surfaces will show up on the edge, which I expect would make controlling those surfaces just as effective as controlling the edge itself - if not more, due to be able to directly apply the concepts from Y14.5

I figure I can see value to utilizing the projected edge concept that DeanD3W mentioned but as a supplemental verification/measurement, it doesn't seem to me that it would be as complete a control as profile of a surface or surface elements as well as not fully capturing the functional requirements in most cases, except for perhaps optical applications (I can't say for sure but even then my gut tells me the surfaces themselves might also still be important, not just the projected edge). For example what if the tip was comprised of two concave surfaces which would result in a very thin edge prone to deformation, chipping, and easily dulling (as opposed to the intersection of two nearly planar features) - this might appear near perfect under a projected edge measurement but a profile measurement of the surfaces would show this deviation.

 

[DeanD3W]

chez311,
I agree that the surfaces that form an edge should also be controlled. I think a profile modified with PROJECTED EDGE (with notes as needed to clarify what it means) is still a useful refining tolerance.

3DDave,
I disagree with your opinion. Profile applied to an edge is not supported by any GD&T standard. A modifier and additional clarification is necessary in my opinion.

Best Regards,
Dean

 

[semiond]

3DDave, If you were following the thread, you could notice that I already gave a good example for a product which requires application of geometrical controls directly on an edge, and where trying to control the edge's geometry indirectly through the intersecting surfaces is not a good idea.
Also, I don't need to supply you any proofs for anything. If you don't think my query is justfied, that means you haven't dealt with an application that requires what I'm looking for. And since you didn't, it is an even bolder statement than mine, to say that there are already sufficient solutions that would make Dean's suggestion redundant.

chez311, you mentioned cutting implements. You are right that the surfaces that meet at the cutting edge are very important by themselfs. They are called the rake face and the flank faces, and they influence the performance greatly. But they are never used to fully define the geometry of the tool. They are traditionally controlled for angular relationionships to theoretical planes, (rake angle for the rake face, relief angles for the flanks). Those theoretical planes (the references) usually can only be defined geometrically through the cutting edges, and not vice-versa: a plane through the theoretical edge, normal or parallel to the tool's shank - is a reference for an angle measurement (By the way, maybe I'm going too much into detail here, but those angles are not defined directly to the tool shank from practical reasons, and the theoretical planes are easy to simulate during inspection).
Rake and flank surfaces are almost never directly controlled for location. The cutting edges are controlled separately for form, location, and orientation. Much of the inspection equipment that is designed with careful consideration of the needs of this industry, such as various vision systems, include built in programs for GD&T controls of detected ("projected" as Dean rightfully defined) edges. Those are also specified in drawings, mostly without any care to "legality" and correspondance with the definitions of any standard. What is missing is the standard giving some kind of treatment for this issue. For example - the "PROJECTED EDGE" modifier that Dean suggested. But even if it won't get into the standard, I'm still glad that Dean pointed on that solution for me - a custom note, accompanied by a proper definition (of the projection direction) is good enough. This is the kind of suggestion that I find useful.

 

[3DDave]

Dean,

Profile applied to an edge is not given as an example figure. That doesn't mean it is not supported. The width of a really sharp edge is on the order of a half-micron. So is it not allowed to use profile of surface on a part that is a half-micron thick? Does the profile of a blade disappear from the time it is thick stock until it is sharpened? This would mean that I can't create a decorative cutout to lay a knife blade in. Since I can, it is obvious the profile does not disappear, no matter how tiny the extent in one direction it may be.

 

[3DDave]

semiond,

You are making the claim that a change is required, so it is entirely up to you to make the complete case in a concise way. Asking people to read post after post to try to pick a thread of continuity out of it is unreasonable.

It does raise an interesting question - if no one in your industry uses the Y14.5 Geometric Characteristic symbols, Datum Feature Symbols, and Feature Control Frames to describe cutting tools, then are you planning to personally convert the millions of users to do so? Nothing right now prevents using a point-normal mathematical description for each face and a profile tolerance to get what you want right now, but it isn't done. Who will bear the expense of this conversion of convention and what is the cost savings you have estimated will pay for it?

 

[DeanD3W]

3DDave,
What definition of a feature do we have that includes edges between two intersecting faces? The answer is none, I believe, so applying profile to something that is not defined as a feature is an extension of principles out into "space" (outer space) to the degree that I think we need a modifier and further explanation to make it clear and understandable. Otherwise it may not be clear in some cases how the tolerance zone is to be oriented. Part of the picture here will be the tools to specify the projection direction. Those tools will be available in the future standard I mentioned in the earlier post.

What if there is a drafted feature that mates with another feature that does not have complimentary draft? The as-designed orientation of that mating feature would likely be the desired orientation of a "PROJECTED EDGE" modified profile tolerance zone. Since draft angles can be very small, that may be a subtle difference from profile applied to the drafted surface itself. This is why the PROJECTED EDGE modifier is helpful to clarify what the tolerance applies to and how it works (what its tolerance zone is comprised of). I think it is helpful to the point that it should be required.

I agree that profile of a surface can be applied to a 2D feature, such as something printed on a surface, with negligible thickness of the print. We don't have the dilemma of how the tolerance zone must be oriented in that case. In that case the tolerance zone simply lies on the surface.

Best Regards,
Dean

 

[3DDave]

Dean,

It seems that you are talking about controlling the theoretical intersection as a feature - not an edge, which is where material stops. A theoretical intersection can already be given as a basis for dimensioning. From what you imply it would also allow a position tolerance on the vertex of a cone.

If there is a tapered feature that mates with a nominally non-tapered feature it is unlikely the theoretical intersections of realized parts will coincide with surfaces on either part or with each other. One can limit the extent of the profile of surface tolerance to a reasonable depth, beyond which contact is not expected. Outside of the expected contact zone one could have a much larger tolerance.

 

[DeanD3W]

3DDave,
No, I am not referring to a theoretical intersection.

Best Regards,
Dean

 

[3DDave]

Dean - OK, if it's not a theoretical intersection then you are referring to a surface, so it's back to profile of surface, even if the surface is really small.

Have you generated a precise diagram of the situation that is distinguished from profile? If I can trace the perimeter of a part then I can compare that to the ideal profile, regardless of how small the radius between the upper and lower surfaces is.