Plus Minus Width Callout

[[Ia...]]

I'm trying to prove that a linear callout with a plus minus tolerance means that ALL points on one surface must be in spec with ALL points on the other surface. I'm arguing against using calipers for a large width. But I can't find anything in the ASME standard to prove my point (or if my thought is wrong, please correct me). Does anyone have any documentation on this?

[[Da...]]

I don't know of any documentation and I don't think you should need any. It's about understanding the limitations of the gage. If the plane you are checking has a concave spot in it, calipers will not show it but instead only show the high points of each face, thus the part could be out of tolerance and the caliper check it as good. The CMM gives you the best chance of seeing the face and its' flaws.

[[Ro...]]

ASME 14.5-2009

1.4 Fundamental Rules, Section (n)
Unless otherwise specified, all tolerances apply for full depth, length, and width of a feature.

[[Ja...]]

ASME Y14.5 2009

2.7 Limits of size

Unless otherwise specified, the limits of size of a feature prescribe the extent within which variations of geometric form, as well as size, are allowed. This control applies solely to individual regular features of size as defined in para. 1.3.32.1. The actual local size of an individual feature at each cross section shall be within the specified tolerance of size.

[[Da...]]

In ASME, it's pretty straightforward. There's "Rule #1" and "Rule #2". Let's start with Rule #2. It states, that the default for geometric tolerances is RFS (Regardless of Feature Size) and RMB (Regardless of Material Boundary). But in this instance, a simple plus/minus tolerance is NOT a geometric tolerance. Those are the ones with a little symbol and an optional datum.

For simple tolerances of distances of regular features of size (those are basically the ones with opposing points), Rule #1 applies, the Envelope Principle. It states, that the surface or surfaces of a regular feature of size may not extend the envelope that is made of the boundary of the perfect form at Maximum Material Condition.

For example, if you have a simple outside width of a part (two opposing planes where ideally every point on one plane has an opposing point on the other, making it a regular feature of size), and the nominal size on the print is 2 inches plus/minus .01, then the Maximum Material Boundary would be 2.01 inches (the maximum envelope). Rule #1 dictates, that this maximum envelope must not be exceeded anywhere to be in tolerance (so, every point of one plane to the tangential datum plane made of every other point of the other plane, or every point to the collection of every other point, if you will). Also, the smallest distance must not be smaller anywhere locally (that means ONLY opposing points) than 1.99 inches. Basically, that would mean: one check for the maximum dimension and virtually infinite checks for the minimum dimension.

The MMB (the maximum) cannot be checked by use of a caliper, the smallest local distance (the minimum) on the other hand can very well be checked that way. You could check the MMB on a height gauge, though, when the granite table acts as a datum simulator.

When it comes to irregular features of size, Rule #1 does not specifically apply, which means those better be defined by geometric tolerances, for which Rule #1 then explicitly does not apply.

[[Ia...]]

I am now arguing that an external width is indeed a feature of size. What a world.

Thanks for all the replies.

[[An...]]

_

[[Da...]]

Please sign in to view this quote.

Not sure, whether there's a question or a statement in there, but ASME defines local sizes as being perpendicular to the derived median line of a feature, the derived median line being an imperfect line made of all points of any cross section of the Unrelated Actual Mating Envelope. So, there's no guessing needed.

Please sign in to view this quote.

[[An...]]

_

[[Da...]]

The problem you're talking about is taken care of in ASME Y14.5.1-2019, Chapter 2.3.3.1, "Establishing the local size spine". It states, that a flawed derived median line will be substituted by a smoothed-out derived median line, that will be tangent-continuous. You can use the spine to create sections, you take the one (1) centroid of each section and use those to create an imperfect line, imperfect meaning, that it is a line that can be curved. A little reminder of differential calculus, a continuously curved line is a line that has got a slope in every intrinsic point, thus allowing to define a gradual normal direction of that line in every point of that line. The direction perpendicular to that normal direction is the direction of the local size, so yes, this imperfect line CAN be used for other things than straightness.

Correction: I will have to rephrase that. Of course, to create the imperfect derived median line in the first place, you will have (in ASME, because it's different in ISO) to use the center line of the UAME to create the cross-sections, then take the center points (the centroids) of the real profile within each cross section.

I suggest you look it up yourself, because I can't post neither the whole standard nor a mathematics book here. I'm sure, you understand the concept, though.

If you posted more words than pictures, I'd probably understand your problem better. I don't have all the answers, I'm not always right, but I'm willing to learn.

Thank you!

[[An...]]

_

[[Da...]]

That is an excellent question, and it reveals one major problem with the ASME standards. The standards we usually discuss are not only the main standard, currently ASME Y14.5-2018, but also the additional standard explaining the mathematical principles behind the main standard, currently ASME Y14.5.1-2019.

Both standards often contradict themselves, because newer concepts are introduced, while the standard explaining the mathematical principles drags behind. As of now, it is the other way around for the first time since 1994. The Y14.5.1 is newer now than the Y14.5, but the successor to Y14.5-2018 is in the making, introducing yet again new or changed concepts. One that I know of is that the procedural algorithm for establishing a tangential plane will no longer be the standard, but rather the L2 calculation (basically a plane parallel to a filtered Gaussian plane).

Back to your question:

The answer is no, because in the Y14.5-2009 there is no mention of the way the local size shall be established. The Y14.5-2009 still refers to the additional standard Y14.5M-1994, but in a revision of 1999. The Y14.5M-1994 was also reaffirmed in 2004 for the last time.

And while the Y14.5-2009 introduces the concept of a local size, there's only a picture showing it, but no further explanation other than "distance at any cross-section of a feature of size". No word about how to establish those cross-sections. And the Y14.5M-1994 [1999] naturally knows nothing about the new concept. In the mathematical principles from 1994 up to the successor in 2019, the diameter of a feature of size must fall within two concentric envelopes, one being the maximum material boundary, one being the least material boundary, with the envelopes being at perfect form (when Rule #1 applies).

So, effectively, the Y14.5M-1994 (and every revision up to 2004) knows nothing about local sizes or how to establish them, and applies the Rule #1 differently than the Y14.5-2009.

If you take three steps back and look at the big picture, you'll see the evolution process, but it is a total mess, which only now slowly seems to get better.

And don't get me started on ISO. I mean I like it, because many of the new concepts of ASME seem to be a gentle approach towards ISO concepts and vice-versa (like killing off the concentricity and symmetry, something than might be expected to happen in ISO, too, and the introduction of the L2 calculation, that might be expected to be introduced into ISO eventually). But ISO already carries the potential to throw a shitload of new symbols onto the world, that no one seems to know or care about except for the designers that use them more and more. Those are quite useful symbols, but ask any manufacturer about them, and they will just shake their heads in disbelief.

My suggestion would be to anyone: Forget about Y14.5-2009 and Y14.5M-1994/1999/2004, and move on to the future. The new standards are money well-spent.

[[An...]]

_

[[Da...]]

Your examples are often exaggerated to an extreme level. Yes, it does make sense to evaluate it that way. Are there other ways to define the normal directions of the local sizes? Yes, of course, but that is the way ASME defines it, and in practice, it is as good as any other way.

By the way, the standard is not talking about a "SPLINE", but rather a "SPINE" (im Deutschen eher ein Rückgrat, eine Seele im mechanischen Sinne) in the sense of a curved line defined of single points, NOT the spline as a geometrical object defined by knot points and then interpolated. There is no interpolation in the spine, it consists only of real points.

In your example, you cut it down to four centroids, but in the standard, you must consider all centroids making then up the spine and then smooth it. Does reality work that way? Do CMMs work that way? No, certainly not. In what way the programmers of metrology software think about this problem, that's a question you're gonna have to ask them. They're certainly going with a compromise of some sort, maybe they really construct a SPLINE to mimic the SPINE, who knows...

[[An...]]

_

[[An...]]

_

[[An...]]

_

[[An...]]

_

[[Da...]]

Please sign in to view this quote.

Again, how the programmers of Calypso handle that, I don't really know. But you're right, they have to come up with a solution if there's only little information.

Please sign in to view this quote.

It defines the direction of distance between opposing points. If it was not smoothed, any error in the derived median line (the unsmoothed spine), would make it impossible to define a normal direction.

Please sign in to view this quote.

Absolutely! Why not?

Please sign in to view this quote.

Local sizes always perpendicular to UAME?[/quote] Another possibility, but the downside would be that you would lose all information about the form, and that's exactly what you don't want. The slightest error in the calculation of the element (due to dust, dirt, too little points, wrong filter, no outlier removal etc.) would result in a massively different UAME. The UAME itself is unsuitable to determine the normals of the cross-sections. And if your element had a large form deviation, and all your cross-sections had the same direction, they would not nearly represent the shortest distances between opposing points you would want.

In ISO 14405-1, no recommendation is made how to measure a two-point distance, only different tools are provided, like the rank-order maximum or minimum as an extension to the simple, not further defined two-point distance. ISO relies heavily on explicit specifications made by the designer. Although, on second thought, the way to measure the two-point-distance might be further defined in ISO 17450-3, which I unfortunately don't have, but you might want to look https://www.toleranzen-beratung.de/unternehmen/aktuelles/ansicht/iso-17450-32016/.

[[An...]]

_

[[Da...]]

If you want to hear that you're right all the time, then I can't help you there. I don't see this as a comedy, so no need to get rude.