Justification for Best fit alignments

[Fuutgut]

I am trying to find a justification in ASME Y14.5 for doing best - fit alignments when there is a true position on a datum feature.
For example: a simple disk with a face as datum A, OD as datum B, a slot as datum C. (Ie. ASME Y14.5-1994 Figure 4-6)
A true position call out on datum C should allow to do a best fit on the rotation, correct?
I need to be able to reference where Asme y14.5 allows this.
Thanks,

 

[Belanger]

When talking about a datum, best-fit is typically not the way to go. In the figure mentioned, the M modifier on datum C (when that datum is referenced) means that there may be some slop or datum shift between the true datum plane and the derived center plane of that slot.
And the derived center plane of the real slot is based on the actual mating envelope, not best fit.

But feel free to elaborate on the question.

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

 

[CheckerHater]

Will somebody please educate me: what exactly is "best fit"?

 

[Belanger]

CH -- that's usually tossed around in the CMM world as a way to find a perfect plane, axis, or point from a non-perfect physical feature. Think of a typical datum feature A as a flat surface; in reality it may be bumpy. A best-fit algorithm on a CMM will take the points probed on the bumpy surface and find a plane that goes through the middle of the bumps (with equal distribution of bumps above and below this perfect plane).

According to Y14.5, this is exactly what NOT to do when deriving a datum. It is supposed to be taken from only the high points (or "tangent plane") of the bumpy surface. Unfortunately, the default setting on many CMMs is best-fit, I think.

Now, the datum C mentioned in the OP is created from a feature of size, not a surface, but it's the same idea of high-point contact. And it happens to be modified with MMB, which is another monkey wrench in the explanation.

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

 

[axym]

Fuutgut,

Are you able to clarify the question a bit? It sounds like you're referring to best fitting the coordinate system alignment, but I'm not sure exactly what you mean. Are you referring to best fitting the rotation when evaluating the Position tolerance that is applied to Datum Feature C, or the Position tolerance that references C as the tertiary datum feature?

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[CheckerHater]

Thanks!

So, in your example, if we specify tight flatnes requirement for datum feature A it will reduce the difference between "high point" and "best fit" datums and make drawing more CMM friendly"?

 

[CheckerHater]

Sorry, I made it unclear I was asking JP...

 

[fsincox]

Guys,
I think it may also be used to check the position of a pattern of multiple feature to "itself" as in the second line of a composite tolerance, on a CMM also?
Frank

 

[Fuutgut]

Sorry for the lack of clarification.
I am not talking about the least squared algorithm vs actual mating envelope as Belanger describes.
Axym is on the right track, but let me clarify.

Please see the attached picture.
Datum C has a TP callout to 0.5 and Datum C is the tertiary datum in the TP of the hole pattern.

The first results from the CMM show the location of datum C as perfect, since it aligns rotation to it.
However the bolt holes are failing TP, since they are off-rotation.
I depicted these first results in yellow.

According to our CMM programmer, since datum C has a TP callout, we can change the rotation alignment when evaluating the TP on the bolt hole pattern, so long as datum C's TP still shows in.
He calls this technique "best fitting" and says it is commonly used in aerospace.
So, after the best fitting, the datum C TP shows in spec (let's say at 0.3) and the bolt holes also show in spec (let's say 0.07).
I drew the results after best-fitting in orange on the attached picture.

My goal in posting this thread is to:
1) Validate that what our CMM programmer is saying conforms to ASME specs.
2) Be able to justify this technique as valid (hopefully) to management.


Thanks for your help!

 

[SeasonLee]

Fuutgut

May be you will find out the answer from this Tec-Ease tip.

https://www.youtube.com/watch?v=zABW_9bc8tg

Season

 

[axym]

Fuutgut,

Short answer - the "best fitting" rotation shown in the sketch is not allowed. There isn't anything in the ASME Y14.5 standard to support this.

The method that your CMM programmer described might be a misinterpretation of something that is allowed, however. If a datum feature is referenced with the (M) modifier (called Maximum Material Condition in the '94 standard and Maximum Material Boundary in 2009), there is an effect called datum feature shift that provides a type of additional tolerance. The alignment can be optimized (rotated and/or translated) within the clearance between the as-produced datum feature and its simulator (gage element).

But this is not the same as optimizing within the tolerance applied to the datum feature, as shown in the sketch. It sounds like your CMM programmer has been misled.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Fuutgut]

Axym,

Thanks, that was the same conclusion I kept arriving at after pouring over ASME Y14.5.

We have many customer supplied drawings where datum features have TP call-outs (such as the Datum C in my example).
What would the meaning of these call-outs be, since by definition the position of the datum should always be perfect?

Thanks,
Chris

 

[pmarc]

Fuutgut,
I agree with Evan and with the Tec-Ease tip.

In this thread, however, you used two different examples. First - fig. 4-6 where datum features B and C in positional FCF for pattern of holes were referenced at MMC (per '94 nomenclature). Second - a figure where datum features B and C are referenced RFS. These are two different stories. In the first case the best-fit alignment is possible at certain conditions, in the second you are not allowed to do the best-fitting. Are you sure the situation you are dealing with (shown in the sketch) is for datum features B and C referenced RFS?

 

[pmarc]

Fuutgut,
The position callouts you are describing do not control datums, they control datum features. No definition says that datum features should always be perfect. On the contrary, datum features, as features existing on a real part, are not perfect, and per the standard should be controlled relative to each other (directly or indirectly) in order to avoid incomplete drawing specification.

 

[Fuutgut]

Pmarc,

Please disregard the first example, I understand how the best-fitting is allowed when the datum feature is referenced at MMC/MMB in Figure 4-6.
My second example (sketch) is like what I'm dealing with (datums referenced at RFS).

So what you're essentially saying in your second post is that the TP call-out is really controlling the form of the slot?

 

[axym]

Fuutgut,

The Position tolerances on the datum features still have meaning. As pmarc mentioned, datums are perfect but datum features are not. Datum features are imperfect as-produced surfaces on the part, and datums are theoretically exact planes, lines, and points that are extracted from datum features. This is one of the main concepts of the datum reference frame section of the Y14.5 standard.

In your sketch example, the Position tolerance on the slot controls the slot's centerplane within a tolerance zone that is two parallel planes 0.5 mm apart. The tolerance zone is exactly perpendicular to datum plane A, and exactly in line (centered) with datum axis B. So the Position tolerance controls the slot's centerplane to be perpendicular to datum A within 0.5 mm and centered on datum B within 0.5 mm. The Position tolerance does not control the form of the slot - the size tolerance would do this (because of Rule #1).

The datum extracted from the slot would be a theoretically exact (i.e. perfect) plane that is, by definition, exactly perpendicular to datum A and exactly in line with datum B.

When a feature is used as a datum feature, one could say that certain aspects of the feature become perfect. For datum feature A, the height (say, the Z coordinate) of the high points becomes perfect because we're creating a reference frame that is "zeroed" on a plane that goes through these high points. This plane becomes the origin for measurements in the Z direction. If you created the ABC alignment on the CMM and then went back and measured points on datum feature A in that alignment, the Z coordinate of the high points should read zero (i.e. perfect).

By the same token, the clocking rotation of datum feature C (strictly speaking, its related actual mating envelope) should be perfect in the ABC alignment. The orientation to datum A and the centering on datum B will not be perfect, but the clocking of the AME will. This is because the clocking rotation is the degree of freedom that is constrained by datum feature C.

Sorry about all the wordy text descriptions - if I had time to make a couple of sketches it would be much more clear.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Belanger]

It doesn't really matter whether we're talking about verifying a position tolerance or creating a datum from that same feature: in both cases the best-fit idea is not to be used, because what's being derived is based on the actual mating envelope. So I stand by my original post, whose idea was simply fleshed out further by Evan and pmarc (with the additional sidebar of datum shift due to MMC/MMB).

The bottom line is that wherever datum C is, that's where datum C is. The holes must be oriented to that (here assuming no MMB modifier). I suppose that you could look at a machined part where the holes are not oriented correctly, and after the fact you could re-cut the slot to be different ... such as the orange line ... and as long as that angled center of the AME is within position tolerance it might rotate the disc slightly, making those holes pass their position tolerance.
And if it wasn't answered yet, position never controls form -- just location and orientation.

To CH: "if we specify tight flatness requirement for datum feature A it will reduce the difference between 'high point' and 'best fit' datums" [answer=yes] "and make the drawing more CMM friendly" [maybe, but since when are drawings driven by whether they're CMM friendly or not?]

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

 

[pmarc]

Fuutgut,
Did our responses regarding datums vs. datum features make the story any clearer? If not, here is the link to a figure for 2009 version of Y14.5 standard (there is no corresponding figure in '94) which I think should be helpful.

http://files.engineering.com/getfil...-a0a4-4e98-b985-7123959e5240&file=fig.1-1.png

Notice at least 3 things:
1. Perpendicularity tolerance applied to datum feature B controls orientation of axis of unrelated actual mating envelope (UAME) of the cylinder. This control makes sense because UAME is derived from actual as-produced feature that has certain measurable perpendicularity error.
2. Form of the cylinder B is not controlled by perpendicularity tolerance, but by size tolerance and Rule #1. And if a cylinder was controlled by position tolerance relative to a DRF, the position tolerance would not control form of the cylinder too.
3. Datum axis B is the axis of related actual mating envelope of the cylinder (RAME). In this case "related" means perfectly perpendicular to datum plane A. So as you see, datum axis B is by nature perfectly oriented to datum plane A, but datum features A and B are not.

===================

Perhaps I should start new thread, but I believe the question I am going to ask is closely related to the initial topic in this discussion. I think it is especially for Evan, as he is a member of the Y14.5.1 subcommittee, although I will really appreciate other opinions too.

In the most current version of Y14.5.1M (math standard for Y14.5), reaffirmed in 2012, there is a statement [para. 4.3.3(b)] and a figure (fig. 4-2) showing tertiary datum basically oriented, but not located to a higher order of precedence datums. Why is that? Per Y14.5-2009 this is true only if translation modifier is used for datum feature reference C in the positional callout for pattern of 2 holes. But the figure 4-2 (again, reaffirmed in 2012, so 3 years after '09 version of the Y14.5 had been issued) does not show the translation modifier. Is this part of the Y14.5.1 standard going to be updated at some point in a future?

 

[axym]

pmarc,

The Y14.5.1 standard will will be updated in the (hopefully not too distant) future. I can't give a specific timetable for release, but we're working on it!

The 2012 re-affirmed Y14.5.1 version has the same content as the original 1994 version - there were no updates. So it doesn't incorporate Y14.5-2009's clarifications and additional tolerancing tools, such as the translation modifier.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Fuutgut]

Thanks everyone! That made things much more clear, especially that long post from axym.
Pmarc, I haven't had a chance to review your new example yet, but I will soon.
I'm eager to get a more firm grasp on GD&T and whip our inspection department into shape.

I'm gonna have a talk with our cmm programmer tomorrow (I guess today since it's after midnight already). Hopefully we can get this confusion settled and move on.

Thanks again!