Does anyone understand "spines" of a regular FoS per ASME Y14.5.1M-1994 (R2012)?

[Tarator]

Hi all,

I am trying to understand "spines" as defined in ASME Y14.5.1M-1994 (R2012).

My question is simple: How do you construct your 2 spines (1 for MMC, and 1 for LMC)?

Without the spines, you can't measure your actual local sizes and find the actual MMC and LMC size, in addition to the actual mating size.

I would appreciate it if anyone could shed some light. Thank you.

 

[Burunduk]

Without the spines, you can't measure your actual local sizes

"Actual local size" is not related to spines and not defined in Y14.5.1M-1994.

Limits of Size conformance and actual value of size are defined in Y14.5.1M-1994 by the swept spheres concept, but it is different and not overlapping with "actual local size" per Y14.5.

The relevant spine is merely a simple curve without self-intersections. It doesn't need to be established in any specific way (unlike for example a median line). As long as you can find spines that will allow conformance, these spines are valid.

 

[Tarator]

Burunduk,

My understanding is that to be able to sweep spheres you have to have a spine, whether it is a 3D curve or a 3D surface. And once you have your spines (the 14.5.1M standard mentions 2 separate spines), you can cut/slice the FoS, and measure the actual local values on each section.

14.5.1M 2.3.1 c. Actual value: These are the local values for each section... The extreme ones become the actual local sizes, don't they? For instance, if you have a cylindrical external FoS, the smallest of the measured actual values will be your actual LMC size, and the biggest one will be your actual MMC size. And those 2 values must be within the size tolerance. Also, if Rule #1 applies, you additionally need to report the actual mating size, which also must be within the size tolerance to conform.

"Actual local size" is not related to spines and not defined in Y14.5.1M-1994.

What is an actual local size? How is it defined?

Thank you.

 

[chez311]

These are the local values for each section... The extreme ones become the actual local sizes, don't they?

No. A feature has only 2 actual values for size for the entire feature defined by the swept spheres interpretation per Y14.5.1-1994 para 2.3.1, a definition for "actual local size" evaluated at each cross section is introduced in the new Y14.5.1-20xx draft which is yet unreleased. This definition does include the use of a local size spine.

The actual values are one external to the material and one internal - these would be defined by the solids G(Sd,Br) where Sd is the d dimensional spine (d=0 is a point for a sphere, d=1 is a curve for a cylinder, d=2 is a surface for a width/parallel planes) and for the external actual value of an external feature is the Br = smallest sphere to which the feature conforms (largest for an internal feature) and the internal actual value for an external feature is the Br = largest sphere to which the feature conforms. The size of the sphere is not variable along the length of the spine. For a feature which is subject to rule #1 the spine for the external actual value of size is of perfect form (ie: for d=1 the spine is a straight line, for d=2 the spine is a flat surface) and would be equal to your UAME size, and the spine for your internal actual value of size is not required to have perfect form (this is for a feature controlled at MMC/RFS, for LMC I believe this would be reversed).

As far as conformance, to put it simply there must exist two spines Sm and Sl for which the feature lies entirely within the volume between two solids G(Sm,Br) where Br = rMMC and G(Sl,Br) where Br = rLMC. Again, if the feature is subject to rule #1 (specified MMC/RFS) then Sm is required to be of perfect form (if specified LMC then Sl is required to be of perfect form). Theres some caveats to that, but thats the gist.

 

[Tarator]

chez311,

My question would be this:

Let's say you have an external FoS (made by 2 parallel planes), and let's say Rule #1 does not apply (maybe Independency modifier was applied)... According to the standard, you would have 2 spines, each of which would be a 3D surface (looks like a ribbon). And we would have 2 fixed-size balls (each swept on its own spine), one for internal (Bl, which is the smallest) and one for external (Bm, which is the largest) to the material. How do you find those 2 values (diameters of Bl and Bm balls)? You have to check each cross-section, don't you? (Each cross-section meaning multiple cross-sections, as no one can check an infinite number of sections). What I am trying to say is, you collect 2 sets of local values. From this collection, the smallest of the smallest becomes your actual Bl value; the largest of the largest becomes your actual Bm value. (Because the part has multiple cross-sections, and each section has multiple local values). I hope it makes sense.

 

[Tarator]

I have added one possible section:

 

[axym]

Tarator,

I think that chez311's description of the two sizes is pretty good. Here is what the internal swept-ball sizes would look like for the feature and spine you drew:

As you mentioned, the internal and external sizes don't have to use the same spine. We can keep trying different spines to get a larger internal size and a smaller external size.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Tarator]

Evan,

Thank you for the input. I think I do understand the two solids/volumes that are swept, and the 2 values: Bl and Bm.

My original question was "how to construct those 2 spines?". It is not explained in any of the standards at all unless I missed it.

Also, the "actual local" size is not well defined either, in my opinion:

For instance, based on the images below: Which one is the actual local size on that section, d1 or d2 (it is an external feature of size)? Let's say d1 = 9.91, d2 = 10.02.

 

[3DDave]

Algorithm wise I'd go with minimum and maximum convex hull creation for widths and then operate within those restrictions to shrink/expand to allow constant diameter spheres to pass through them without getting trapped. For enveloped constrained features the one spine is trivial and can be constructed without depending on the produced features. Of course for the minimum volume of the external feature there's no unique spine surface; the requirement is there is at least one.

The unstated supposition is that variation in the form of the feature is small enough that sin(theta) = theta, so all those jagged examples aren't really applicable. I recall the math standard was also implying the considered surfaces are curvature continuous. More than that, that changes in curvature are limited to the radius of the sphere.

 

[Burunduk]

Tarator, regarding your last post - that's why it's called "local". Both are.

 

[chez311]

Also, the "actual local" size is not well defined either, in my opinion:

You are correct, it is very poorly defined in Y14.5 and is not included anywhere in the current Y14.5.1-1994 version. The new Y14.5.1-20xx draft even specifically acknowledges this and provides two possible definitions while also specifically stating that neither supersedes or guarantees conformance to the swept spheres interpretation.

 

[chez311]

3DDave/Evan,

OP's inquiry has got me thinking how the swept spheres definition of size would be applied to a feature measured on a CMM with a discrete number of measured points. As 3DDave mentioned for features subject to rule #1 the envelope constrained spine/solid is trivial as it must have perfect form, but what of those that are not envelope constrained where the spine is not required to be of perfect form? The convex hull creation algorithms might work reasonably well for a width shaped feature, however I'm not so sure on cylindrical features - I guess there are other algorithms to better interpolate an approximately cylindrical solid from the discrete points gathered by the CMM?

 

[Tarator]

Algorithm wise I'd go with minimum and maximum convex hull creation for widths and then operate within those restrictions to shrink/expand to allow constant diameter spheres to pass through them without getting trapped.

The convex hull creation algorithms might work reasonably well for a width shaped feature, ...

I have never heard of "hull creation". Would you be able to explain what it is or guide to a resource where I could read?

 

[chez311]

Tarator,

Admittedly I was not familiar with that particular type of algorithm until 3DDave mentioned it. A google search gave me a general idea of what they are about.

https://en.wikipedia.org/wiki/Convex_hull_algorithms

https://en.wikipedia.org/wiki/Convex_hull

 

[Tarator]

Thank you, chez311.

The rubber band analogy is very good actually.

 

[Burunduk]

A note about "actual local size" and its incomplete definition which was touched on in this thread:

a definition for "actual local size" evaluated at each cross section is introduced in the new Y14.5.1-20xx draft which is yet unreleased. This definition does include the use of a local size spine.

I don't have the draft and don't know what that new concept is, but "Local size spine" sounds like a bad idea doomed to be ignored in the industry.
Although there is an issue with how actual local size is defined in Y14.5, what it attempts to describe is two-point measurements. Calipers and micrometers do not rely on any spines for the measurements they are used for.

The only thing missing from the actual local size definition is the specification that the distances shall be taken between opposed points, and the measurement direction shall be normal to the surface at both points.

 

[greenimi]

The only thing missing from the actual local size definition is the specification that the distances shall be taken between opposed points, and the measurement direction shall be normal to the surface at both points.

Then I guess: opposed points should be clearly defined and also what means "normal to the surface" should be clarified.
(runout section has its muddy definition of "...indicator fixed in a position normal to the true geometric shape", but what is the true geometric shape is not really clear). You don't want to close one "door" and open up 5 others, so to speak.

 

[Burunduk]

Then I guess: opposed points should be clearly defined and also what means "normal to the surface" should be clarified.

I agree. If "opposed" is properly defined and examples are provided, then also the definition of a regular feature of size would benefit from that.

I guess if "opposed" is defined clearly (the way most people understand it anyway and as described in some books) then the "normal to the surface" addition is not even necessary, as measurement between opposed points will always be normal to the surface (although not every measurement normal to the surface is necessarily between opposed points).

 

[axym]

Hi All,

This discussion of size definitions is similar to countless others that have occurred before. Eventually, the same massive issues are pointed out:

-The swept-sphere definition in Y14.5.1 is clear, but has immense practical difficulties. One of the two swept volumes can be approximated with a physical object, usually something like a gage pin, sleeve, or gage block when Rule #1 is in place. The other swept volume is "inside the material", can have a curved spine, and cannot be verified physically. Accurate calculation of enveloping swept-sphere surfaces from point data is not straightforward.
-The sphere-based size definition is not mentioned in Y14.5.
-The definition of actual local size in Y14.5 is problematic. The two main components, "any individual distance" and "any cross section" are not clearly defined and become ambiguous on as-produced surfaces.
-Many people would agree that what Y14.5 tries to describe for actual local size is two-point measurements, and there are many indications in the standard that are compatible with this interpretation. However, many other people would disagree as there are also references to circular elements.
-2 point measuring devices such as micrometers conflict with the concept of measurement normal to a spine. Mic measurements are normal to the surface but are generally not coordinated with each other, and spine-based measurements are coordinated with each other but are generally not normal to the surface.

I would say that these issues limit the application of simple definitions of size to "well-behaved" geometry, in which the magnitude of form error is small compared to the size tolerance and to the overall dimensions of the feature. If we try to extend these simple definitions to apply to more irregular (and thus complex) as-produced geometry, it rapidly becomes impractical.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Burunduk]

-Many people would agree that what Y14.5 tries to describe for actual local size is two-point measurements, and there are many indications in the standard that are compatible with this interpretation. However, many other people would disagree as there are also references to circular elements.

axym, could you provide examples of those references to circular elements that suggest that Actual Local Size does not describe two-point measurements?

In my opinion, if the definition of Actual Local Size changed to include "individual distance between opposed points on the surface(s)" there would be no problem remaining, other than the fact that a way to calculate a single "actual value" for a feature's size (like the swept spheres concept provides) from the measured Actual Local Size values would still need to be defined.