Runout equivalent for a non-cylindrical form

[BradVanasse]

I am wondering how best to dimension a part with the following characteristics. It is shaped like a short shaft with a bore down its length and a series of rectangular keyways cut full length into it. The keyways are equally spaced, so it is essentially a splined shaft with rectangular grooves. I want to separately control the distance from the keyway bottoms to the axis of the bore and the variation of that same distance; I want to have a tighter tolerance on the variation of the dimension than on the dimension itself. It would be akin to the runout tolerance except it's not a cylindrical form. How can I do this?

 

[pmarc]

Use composite profile of a surface callout.

(1) Make the bore datum feature, say A. (2) Use basic dimension for the distance between the axis of the bore and the keyway bottoms. (3) Apply the composite profile callout. The upper segment (with looser tolerance) referencing to A will control the distance between each bottom and the axis A. The lower segment (with tighter tolerance) containing no datum feature references will control variation within the pattern of bottoms.

 

[BradVanasse]

Thanks for your reply. However, it is my understanding that the composite profile callout as you describe would constrain the variation within the pattern to the tighter tolerance but allow the entire pattern to move within the larger tolerance. This isn't what I am looking for. I want the individual features to have the freedom to move radially within a larger tolerance provided they are all moving together so that the variation from the bore is within the smaller tolerance. In other words, if I were to mount the part on its bore, using that as the datum feature, and rotate it by indexing the angular distance of one groove at a time and use a dial indicator to detect the center of the groove (lifting the plunger between indexes) the FIM should be less than the smaller tolerance. If the bottom of the grooves were curved with a radius at the same center as the bore, I could call them out collectively as a continuous feature and use a runout tolerance that is tighter than the radius tolerance. But they are not curved, they are flat facets that form a polygon interrupted by the uncut shaft OD between them.

 

[pmarc]

I see what you are saying and agree that my proposal does not grasp your intent properly. To satisfy it, you would need something like a dynamic tolerance zone of the smaller value that could grow and shrink in "overall" size within the larger tolerance zone. This will be quite easly achievable in a next version of Y14.5 standard (when a new "dynamic profile tolerance zone" modifier is introduced), but unfortunately not now.

With that being said, I am afraid I don't have any decent solution at the moment other than the suggestion to use an additional drawing note (or even an extra view) and specify precisely what is acceptable and what is not (especially when speaking about the tighter requirement).

 

[pylfrm]

One possible hackish workaround for ASME Y14.5-2009: Apply a single profile of a surface tolerance to all of the flat keyway bottoms. Use the tighter allowable variation as the tolerance value, and the bore as the primary datum feature. Apply a basic dimension from the bore axis to the flat keyway bottoms, but give a range instead of the standard single value. Use a range that is equal to the difference between the total allowable error and the profile tolerance value. Apply a flag note stating something like "ANY SINGLE VALUE WITHIN RANGE SHOWN MAY BE USED FOR BASIC DIMENSION INDICATED".

It's a shame total runout is not defined in a more general way so it could be applied here. I think I'd favor such an expansion of definition over adding the new dynamic profile modifier. The current application would just become a special case, so you could have full backward (or is it forward?) compatibility.

Thoughts?


pylfrm

 

[3DDave]

For this problem, give all 4 surfaces an angularity tolerance that is desired relative to the axis. This will control the orientation of each flat to the axis while forcing them to be flat but not controlling the distance from the axis and still controlling (because of basic assumed 90 degree angles) their orientation to each other. Then put a 4X dimension to the center with a flag note indicating which center is to be used, such as the Datum Axis used in the angularity tolerance and, in that note, set the maximum variation among all 4 measurements. If there is some complaint about clocking, then pick one of the flats and use that as a datum reference for the other three flats or pick some item external to the set of flats and use that.

The part is chucked using the axis reference, and rotated until the the first flat meets the angularity to the axis using a dial indicator. Set the indicator zero at either the high or low spot. Then the part is rotated 90 degrees, do not reset the indicator, and the total variation is used to verify the angularity while the difference is used to verify the allowable difference. Repeat 2 more times ensuring that the total variation from the axis is not exceeded.

<soapbox>
How many drawings already have notes that describe acceptance of some part exactly the way a new modifier will?

If it aren't many, then is there a lot of value in adding more baggage versus getting users to think through the process and write out what they want to accomplish? Either way, someone will not know these edge cases and have to grab the reference, so placing that directly on the drawing seems best.

I expect it will get as much usage as datum modifiers for customized datum reference frames do, which is not much.

(Can someone, for goodness sake, do a replace of 'invoke' with 'use' I know someone there has a thesaurus and they should have it taken from them. Invoke, verb: cite or appeal to (someone or something) as an authority for an action or in support of an argument. Using stilted language isn't the same as using it correctly.

"a customized datum reference frame may be invoked." No, it cannot be invoked. Stop it. From

https://www.vocabulary.com/dictionary/invoke

: To invoke is to call up something such as a law, a higher power, or even a ghost. In court, you might invoke the Fifth Amendment (the right not to say something that will make you look bad) if you don't want to talk.)

</soapbox>

 

[Belanger]

I would say that having new modifiers in the alphabet at least gets people to think about it, and realize that they have a lot of old drawings which should have had extra notes.

For example, how many times did you think about the "continuous feature" idea (along a shaft with grooves cut into it) before you saw the 2009 standard?
How many times after?

That's the value (assuming the modifier is explained clearly when first introduced).

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

 

[pmarc]

I agree with J-P and would like to add that the standard is like a toolbox - it offers a set of tools (modifiers, rules, concepts), but this does not mean that each and every tool has to be regularly used by each and every user. Just this month I had an opportunity to apply customized datum reference frame concept at least 4 times. Only since the beginning of 2016 I can remember of at least a couple discussions on this and other GD&T forums where possibility of "dynamic profile tolerance zone" application would really help to meet design intent (and I can recall similar discussion in past years too).

In my previous job (I was there for 6 years) I never had a chance to use a runout tolerance (for example), because items I was dealing with had no cylindrical features. I never had a chance to use 80% of things defined in Y14.5. So does that mean runout tolerances and the 80% of other things should be gone from the standard? Of course not.

Going back to the original question, both ideas (pylfrm's and 3DDave's) still require some kind of a note to clearly express design intent. In the lack of standardized tool, I would lean towards mixing them, that is to apply 'nX' directly toleranced dimension with looser toletance from the axis to the flat (a drawing note would have to really precisely explain interpretation of this dimension) and also use profile of a surface wrt A with the tighter tolerance value with 'nX' prefix. In absence of a basic linear dimension between the datum axis and the flat this would mean (at least on 2D drawing) that the pattern of n profile tolerance zones would have to stay a polygon, but its overall size (across flats) would be free to shrink/grow.

3DDave,
In my opinion angularity tolerance will not work because as an orientation control it does not fall under simultaneous requirement rule. In other words, with angularity tolerance used 4 flats may be perpendicular to each other but do not necessarily have to form a square - each flat may be at different distance from the datum axis, and the variation of that distance will be controlled by directly toleranced dimension with looser tolerance, which is not what OP wants.

 

[3DDave]

Are the four surfaces shown at nominally 90 degrees? Then this falls under '2009 2.1.1.4 Implied 90° or 0° Basic Angle, and these angles are considered basic. No need for simultaneous requirement.

<soapbox>
pmarc, I didn't come close to suggesting that every one needed to be used by every user, just that it should be frequent enough that the general population would see some benefit.
</soapbox>

 

[pmarc]

3DDave,
Of course it (4 flats nominally perpendicular to each other controlled by angularity tolerance) falls under '2009 2.1.1.4 Implied 90° or 0° Basic Angle. But it does not mean that 4 angularity tolerance zones must/will create a square.

Picture an as-produced part that has each pair of two adjacent flats perfectly perpendicular to each other, but the distance between opposed flats #1 and #3 is 1, while the distance between opposed flats #2 and #4 is 10. This will also meet the angularity tolerance suggested by you.

In OP's case, of course, such difference in distances between opposed flats will not be possible because that distance will be controlled by directly toleranced dimension with a looser tolerance, but this still will not be enough, as OP is trying to keep the variaton in distances within a tighter tolerance.

<soapbox>
Well, "frequent enough" is a very subjective term. To me a couple opportunities per month is pretty frequent.

In this thread

http://www.eng-tips.com/viewthread.cfm?qid=415186

you seemed to disagree with the committee's decision about removing concentricity and symmetry from the future version of the standard. May I ask how often have you seen true functional need to use these two characteristics? If I remember correctly, you once gave us really great real life example of concentricity tolerance application, but you probably must have seen much more valid applications since in your opinion concentricity and symmetry deserve to stay, right? (BTW, I agree with you that they should stay).
</soapbox>

 

[3DDave]

Which is why I said there needed to be a note about locating the four surfaces from the center datum with a smaller variation than the overall width; pretty much the verbiage required to describe the same situation as might be added to a standard.

"Then put a 4X dimension to the center with a flag note indicating which center is to be used, such as the Datum Axis used in the angularity tolerance and, in that note, set the maximum variation among all 4 measurements."

Again - it is the need of the general population. I've had need to represent a tolerance zone in the form of a complex equation, but I don't expect that method to be added to the standard, even though it would make most interpretation problems go away in favor of inspectors using scientific programmable calculators.

Adding rotational symmetry to the standard would also fix this particular problem.

 

[BradVanasse]

Thanks for all your input. Pmarc, if I understand your post of 29 Dec 16 20:27 correctly, either of the two illustrations shown attached would work and be equivalent to each other?

Regarding the comments on customized datum reference frames: I was not familiar with the concept, so I studied up in the ASME Y14.5M-2009 standard and it struck me that if one could set up a customized datum reference frame using a cylindrical coordinate system, then doing what I wanted without notes would be possible. But I suppose that would be too esoteric of an extrapolated application of the concept to be useful.

 

[dgallup]

If you have an even number of these features so they have a mirror image feature across the centerline, you could use the much hated and maligned symmetry control to accomplish what you want.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[BradVanasse]

No, dgallup, symmetry controls the location of the mid-plane between the opposing surfaces, not the surfaces themselves. Borrowing pmarc's illustration using "4 flats nominally perpendicular to each other" from the posting of 30 Dec 16 16:30, "Picture an as-produced part that has each pair of two adjacent flats perfectly perpendicular to each other, but the distance between opposed flats #1 and #3 is 1, while the distance between opposed flats #2 and #4 is 10." In such a case, the mid-planes could meet the symmetry requirement while exceeding the desired variation in the flats distance to the ID's center.

 

[pmarc]

BradVanasse,
Answering to your previous post with the attachment, I would say that equivalency between the two illustrations is at least debatable.

Personally, I see potential issues with the |prof|0.05|A| callout on the illustration on the right. It may be unclear to some readers (including me if I did not know the intent of the callout) as to whether basic linear dimension 34.29 locates this 0.05 profile tolerance zones or not. That is why I suggested getting rid of this basic dimension and go with the solution on the left.

But now, as I think of it more, I would see another option - modification of the picture on the right by just adding 'BASIC 34.29 DOES NOT APPLY' at the end of 'VARIATION AMONG ALL LIKE FEATURES' notation. I think that would solve the problem and would be much more GD&Tish than applying directly toleranced dimension 9X 34.29+/-0.10, as shown on the left illustration.

 

[BradVanasse]

pmarc,
Yes, of course you are correct. The fact of the basic dimension also applying to the tighter surface profile tolerance slipped by me. As an alternative to your 'BASIC 34.29 DOES NOT APPLY' idea, would the use of a customized reference frame that shows only rotational constraints in the tighter tolerance work? See my illustration, attached. Did I correctly apply the constraints to the degrees of freedom in this example?

 

[pmarc]

BradVanasse,
I would say that the customized datum reference frame concept, as shown in your latest attachment, will not produce equivalent requirement to the 'BASIC 34.29 DOES NOT APPLY' idea. Since A constrains only two rotational degrees of freedom, there is no requirement for the pattern of nine tighter profile tolerance zones to be centered as a group on datum axis A. Your idea, in combination with the other profile callout, actually creates a requirement pretty similar in meaning to my first proposal in this thread - composite profile tolerance. The difference is that I initially suggested using no datum feature references in the lower segment of the callout.

In general, I don't think that the customized datum reference frame concept will be useful in this particular case. The intent is to have the pattern of tighter tolerance zones centered on datum axis with additional ability to shrink and grow freely within the larger profile tolerances while staying regular polygon. That additional shrink/grow ability is not what the customized datum reference frame concept was designed for.

Some additional comments/remarks:
- if the tolerance for 34.29 dimension is +/-0.10, then the profile tolerance value defining nine larger tolerance zones should be 0.2 not 0.1.
- when the customized datum reference frame concept is used, there is no need to explicitly list degrees of freedom in those feature control frames that actually don't override any DOFs (see geometric tolerances applied to datum features B in figs. 4-45 and 4-46).