Interaction between Size , Circularity and Staightness 2

[mkcski]

The attached drawing is the portion of a shaft where a combination guide and thrust bearing interfaces. The bearing surface is Babbitt and is oil lubricated.

I am trying to determine where the conflicts, if any, are between the size, circularity and straightness tolerances. The sketch is fuzzy: the tolerance on the 290mm diameter is -0.291mm to -0.323mm, or 0.032mm total on the diameter. This drawing is from an European company and there is no ISO GPS standard referenced, so for discussion, assume Rule #1 is NOT in effect.

Note: Please ignore the problems with the "parallel to A" FCF - a datum feature "A" cannot be parallel to itself. Also I am not concerned with the runout and flatness FCF's - it should be total runout.

Issue 1 - the circularity tolerance of 0.06mm is radial error, so the diameter can vary twice this - 0.12mm - and still meet the form requirement. But this is 4 times the size tolerance of .032mm. From my understanding, even without Rule #1 the form error must be at least half of the size error. Comments please.

Issue 2 - Given the independent and two-dimensional nature of the circularity and straightness tolerance zones, the 290mm diameter could be a conical shape and still meet the form tolerances. But, given the 0.032mm size tolerance, the 0.60 circularity and the much tighter 0.01 straightness tolerance, how do these interact? I suspect a conflict. My understanding: given the (radial) circularity error of 0.06mm, a line on the surface can "jump" 0.06 from one cross section to the next adjacent one. This would allow a surface straightness of 0.06mm, violating the 0.01mm requirement. So ...how does the 0.032mm size tolerance overlay with the two form errors, which appear to conflict each other? Given the surface is a bearing journal, I am thinking of recommending cylindricity as a more appropriate control. Comments please.






Certified Sr. GD&T Professional

 

[3DDave]

Aren't the items being applied to the surface elements? If that's a different interpretation for ISO than ASME, then I'm done.

Otherwise the parallelism is for each surface element to be parallel to the axis of the feature of size RFS.

The surface jump is limited to .01 by straightness.

 

[mkcski]

3DDave:

Thanks for the review.

My ISO knowledge is limited. But as far as I know, ISO and ASME interpretations for form controls are the same. So in this case, I assume the form controls are definitely surface elements.

The parallelism could be surface or axis. If a diameter symbol preceded the 0.02 tolerance, I would assume tolerance zone is at the axis. But from my understanding, the words LINE ELEMENTS would be placed under the FCF to place tolerance zone at the surface - Y14.5 para 6.4.3 . It's not, so some clarification is needed from the customer. Alternatively: maybe with ISO, the lack of the diameter symbol implies surface control. However, from my understanding, in either case - surface or axis - the 290mm cylindrical feature cannot be parallel to itself - Datum A. Or, are you saying Datum axis A RFS (variable simulation) can be established from the 290mm feature and then line-elements on the same feature be controlled parallel to the axis. This seems to be a "chicken or the egg, which comes first" dichotomy. Can you clarify?



Certified Sr. GD&T Professional

 

[pmarc]

mkcski,

1. I am not sure if this is just my problem, but I am unable to open your attachment.

2. Issue 1 - I may have posted this picture before on the forum, but will do it once again:

http://files.engineering.com/getfil...995c68bf0&file=independency_circularity_2.JPG

It is a sketch created by me based on a figure from one of ISO GPS-based books I have. Since default principle in ISO is independency, there is no requirement (unless specifically stated) for the feature of size to stay within MMC envelope even if all actual local sizes of that feature equal MMC. As shown in the sketch, for 3-lobbed shaft all actual local (two-point) sizes can equal 80 (MMC size) yet the shaft violates the MMC envelope by quite a lot, that is by approximately 12 (the numbers shown were taken from the book).

The thing is that with no circularity tolerance directly or indirectly specified on ISO drawing, and for certain types of form errors, the actual circularity error can be much bigger than just the size tolerance, not to mention half of the size tolerance. In my example it is approx. 12 vs. 0.5 (and with additional assumption that actual sizes do not vary between 79.5 and 80.0, which would give even bigger circularity error). So this means that from theoretical point of view any circularity tolerance less than approx. 12 makes sense. Now, in your example it is just 0.06 (circularity tolerance) vs. 0.032 (total size tolerance), meaning that your case is far away from absolute limits.

3. Issue 2 - I don't see a conflict between the considered tolerances (size, straightness, circularity). With no straightness tolerance specified (either surface straightness or extracted median line straightness, as they call it in ISO) form of the cylinder along its axis stays completely uncontrolled. The 0.01 straightness tolerance defines that form limit. In my example there is no such limitation, and that is why presented shaft, even with such a huge circularity error, could be a banana or a sine wave of any amplitude. Any straightness tolerance would simply limit that error to certain finite amount.

 

[3DDave]

I was looking at fig 5.1 of the 2009 version, which does not mention line elements. On the surface of a cylinder there can't be anything sensible but line elements for orientation, if the control is not applied to the diameter.

In the orientation examples it was necessary to distinguish from a surface bounding zone.

The advice of 6.4.3 was not applied to Figure 5.1, so it's not clear if it was thought through or if cylinders are considered exceptional and exempt.

 

[pylfrm]

mkcski,

What exactly do you mean by conflicts between tolerances? It seems to me that each tolerance places a limit on a distinct characteristic, each of which can be verified independently, so I fail to see an opportunity for conflict.


pmarc,

I have very little experience with the ISO system, so perhaps you can confirm whether I am understanding this correctly: Based on your post, it sounds like there is a drastic difference between an ISO size tolerance (with default independency) and an ASME size tolerance (with explicit independency) when applied to a cylindrical feature. Per ASME Y15.4.1M-1994, the surface must fall within an envelope generated by sweeping a ball (having diameter equal to the upper size limit) along a spine. If we assume the curvature of the spine is not too extreme (as I suspect is the intention of the standard, despite it not being discussed), then geometry like that shown in you image will not meet the ASME tolerance.


pylfrm

 

[mkcski]

All:

Thanks for the prompt responses. I am "snowed" right now with work and I will digest your info and reply ASAP

I do not know why the drawing will not open. I will attempt the resend



Certified Sr. GD&T Professional

 

[SeasonLee]

pmarc

I did unable to open the attachment, but I can see it on my ipad, here you are.

Thanks for your posted picture and interpretations on ISO default principle, a really big difference between ASME and ISO. A simple question for you: shall we accept or reject it if the actual part made with circularity error on Ø92 as shown on the picture? I will reject it, am I right?

Just for curiosity, normally there is NC modifier on flatness tolerance, I know NC means Not Convex, but OP’s drawing with an indication “Not Concave” on the flatness FCF, so Convex or Concave?

And furthermore, why “Not Convex” always followed with flatness callout I/O “Not Concave”? Are there any technical reasons on the application?

Season

 

[mkcski]

All:

Here is a resend of the file I attached in the OP. Let me know if it does not open.

Certified Sr. GD&T Professional

 

[3DDave]

mkcski, If there is a problem with the file it is likely the name, which includes a comma. I opened it without trouble originally, so there wasn't anything wrong with the contents.

 

[mkcski]

3DDave:

Thanks for the tip.

Certified Sr. GD&T Professional

 

[pmarc]

First of all, thank you for attaching the file once again, mkcski and SeasonLee. I am able to open Season's file, but still no success with mkcski's. As 3DDave suggested, this may have something to do with special characters in the filename. I would also like to thank J-P who sent me this file offline.

Going back to the questions addressed to me...

pylfrm,
Yes, there is a drastic difference between definition of size in ISO and ASME. And yes, you are right - the geometry shown in my image will not pass ASME requirement.

One interesting question someone might ask here could be:
So is there currently a way in ASME Y14.5 to intentionally allow such thing to happen?

The other interesting question (somewhat loosely related to the topic being discussed) could be:
Is there currently a way in ASME Y14.5 to define derived median line straightness tolerance without overriding Rule #1?


SeasonLee,
I think the answer to your first question as to whether the part shown in my image (with such a significant circularity/roundness) should be accepted or not is yes, it should be accepted. If an ISO-based drawing specifies size requirement only (as is in my example), the drawing is simply lacking complete definition of form of considered feature. Therefore QC inspector does not have a requirement/criterion based on which she/he could make meaningful decision that the feature is non-conforming. Not to mention that because the circularity requirement has not been specified, the inspector does not even have to verify or know the actual circularity error of the feature. In many cases all she/he will do will be simple two-point size check with a caliper. With this check chances to detect a 3-lobbed-like circularity error are rather small.

Per ISO 1101 'NC' stands for 'Not Convex'. And since there is no modifier for 'Not Concave' I think that explains why in mkcski's drawing someone explicitly wrote 'Not concave'. The other side of the story is that in this very case the 'Not concave' callout is still pretty ambiguous. Since the flatness tolerance frame has been associated with size (width) dimension, it (together with the 'Not concave' note) applies to the extracted median surface of the 300,5 width. Question: for extracted median surface a bow like this ) means convex or concave?

One reason that comes to my mind for having NC (Not convex) modifier often added to the flatness tolerance frame is that especially for primary datum features it is much better if the feature does not rock/wobble when brought in contact with its counterpart or inspection equipment that simulates datum plane. When the datum feature is concave (not convex) the rocking/wobbling is much less likely to happen.

 

[SeasonLee]

Thanks for your detailed interpretations, pmarc.

For an ASME user, I need some time to adapt to these ISO concepts.

Season

 

[mkcski]

All:

It appears not everyone can open my original file. Given the earlier "commas" advise, I renamed it and have reattached it.

I have attempted to digest all comments to this point and I think there is a general redirection to discuss "I" Independency vs "E" Envelope as it applies TO ISO vs ASME. I understand the difference. So allow me to bring the discussion back and rephrase my OP:

My question focus' on the two form controls with Rule #1 not in effect. Given the 0.06 circularity and the 0.01 straightness, it appears the straightness tolerance limits the circularity error. For example: it may NOT be possible to meet the 0.01 straightness tolerance at every line element if the circularity error is maxed-out at 0.06 at every circular element. As I see it, the circularity error is limited to the smaller straightness error to get controls simultaneously acceptable at every line-element and at every circular element. Because the design not a conical shape, where circularity and straightness can be independent, but is a cylinder, the two form controls are interdependent - one or the other can be applied but not both simultaneously. True, there may be line elements where the points on each circle fall on a line within 0.01; but every line and every circle must be within tolerance. To meet both limits, the zones must equate at 0.01 - this "smells" of cylindricity.

Regarding the parallelism requirement to Datum A: As I posted earlier, a cylindrical feature - surface line elements or axis - cannot be parallel to itself - Datum A is the 290mm diameter. So there is a relationship issue to resolve with the customer. Functionally the surface needs to be positioned to the other features on the shaft, which are not shown in this sketch.

Regarding the NOT CONCAVE note. The FCF's are attached to the 300.5 width FOS and as such, the tolerance zones are by convention at the center plane. But here the symbols mean they applied to each face independently - runout and flatness are surface not "center" controls. Function: The inside faces are thrust surfaces controlling axial movement in both directions. For some functional reason unknown to me, the crown of individual surfaces must be higher near the center(closer to together in near the 290 diameter). This may have to do with maintaining oil film thickness, making sure the opening between the two is wider at the outside edges to allow the mating part to engage between the two, etc. I would combine the two and change the control to total runout with Not Concave note.



Certified Sr. GD&T Professional

 

[pmarc]

Given the 0.06 circularity and the 0.01 straightness, it appears the straightness tolerance limits the circularity error. For example: it may NOT be possible to meet the 0.01 straightness tolerance at every line element if the circularity error is maxed-out at 0.06 at every circular element. As I see it, the circularity error is limited to the smaller straightness error to get controls simultaneously acceptable at every line-element and at every circular element.

I am afraid I don't agree with you, mkcski. Straightness tolerance and circularity tolerance control two different aspects of feature's form, therefore one does not have any impact on the other. Take a look at my image again and imagine that this 3-lobbed shaft (with significant circularity error) is perfectly straight. In that case the straightness error is 0 and the circularity error is 12. If what you are saying was true, it would be never possible to have a shaft produced with circularity error greater than straightness error.

The lack of a relationship between straightness and circularity tolerances is true in ASME world as well (where Rule #1 is default condition). Picture a shaft with total size tolerance of say 0.1. There is nothing wrong in having straightness tolerance of 0.01 and circularity tolerance of 0.06 applied in addition to the size requirement. As long as both geometric tolerance values are less than the total size tolerance everything is fine.

Did I miss something in your question?

BTW: This time your attachment works fine.

 

[mkcski]

pmarc:

Thanks for the prompt response. I need to ponder this somemore before I respond. I might have to unlearn something I thought I understood hahaha

Certified Sr. GD&T Professional

 

[pylfrm]

mkcski,

You might find thread1103-418322 to be of some interest. One of the posts contains a drawing very similar to yours. I wonder if datum feature A in your drawing is supposed to be an opposed generator line instead of the entire cylindrical feature. Of course that doesn't make sense when it's referenced in the runout tolerance though.

If datum feature A is indeed the entire cylindrical feature, then I would see no problem with applying a parallelism tolerance to line elements of its surface.


pmarc,

Thanks for confirming about the ISO size tolerances. I hadn't really realized that until now.

For the first interesting question, I can't think of anything other than explaining the requirement in a note.

For the second interesting question, perhaps you could apply a straightness tolerance of 0 at MMC in addition to the derived median line straightness tolerance.


pylfrm

 

[CheckerHater]

I knew I've seen that picture before

"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[mkcski]

pamrc:

Before I agree with your statement that circularity and straightness are independent, let me offer the following scenario for discussion:

Consider the "triangle" sketch you posted on 4/6. I tried to repost but no success.

BTW..the .12 circularity error stated on the sketch is not correct. It should be 6 not 12, because the circularity tolerance zone is a radial deviation and not diametrical.

Anyway...back to my scenario....the sketch shows the circularity error at ONE cross section. But suppose the triangular lobe (form error) in the next cross section is rotated 120 degrees, so the "highs and lows" out of phase. The measured straightness between the two would then equal the 6 circularity error. As I see it, the only way the straightness error could be less than 6 is IF , and a big IF, the high (or low) lobes at ALL cross sections align along the entire length of the part (at least within a lesser unspecified straightness tolerance). Forcing the lobes to align to obtain a lesser straightness error seems "unconventional". I cannot understand how the independency resolves this scenario. Can you enlighten me?

Certified Sr. GD&T Professional

 

[pmarc]

BTW..the .12 circularity error stated on the sketch is not correct. It should be 6 not 12, because the circularity tolerance zone is a radial deviation and not diametrical.

Allow me to again disagree with you. Please take a look at what happens along the vertical center line (y direction) in my sketch. The size dimension 80 is between the highest point of dia. 92 red circle and the lowest point of the smaller red circle. This means the radial difference between two red circles, thus actual circularity error, is 12 not 6.

Anyway...back to my scenario....the sketch shows the circularity error at ONE cross section. But suppose the triangular lobe (form error) in the next cross section is rotated 120 degrees, so the "highs and lows" out of phase. The measured straightness between the two would then equal the 6 circularity error. As I see it, the only way the straightness error could be less than 6 is IF , and a big IF, the high (or low) lobes at ALL cross sections align along the entire length of the part (at least within a lesser unspecified straightness tolerance). Forcing the lobes to align to obtain a lesser straightness error seems "unconventional". I cannot understand how the independency resolves this scenario. Can you enlighten me?

First, if the next cross section of the 3-lobed shaft is rotated 120 degrees, the "highs" and "lows" will stay in-phase. I believe you meant 60 degrees rotation, right?

Second, I agree that if the sections are rotated this way (by 60 degrees) the measured straightness error would be equal the circularity error. However, I am not sure I understand why you find "unconventional" the forcing of the lobes to align to obtain a lesser straightness error. For this specific (as described by you) "incarnation" of actual circularity error certain additional conditions must be met to satisfy straightness tolerance. But this is no different to what can happen in ASME. Using one more time the example of a shaft with total size tolerance of 0.1, straightness tolerance of 0.01 and circularity tolerance of 0.06, you would also have to satisfy pretty much the same conditions for the shaft produced like an egg (within the 0.06 circularity tolerance). In other words, it also would not be possible to have a shaft looking like a vertically oriented egg in one cross section and like a horizontally oriented egg in the other cross section.