precision shaft tolerances 1

[cervantes]

Hi

i am a little bit confused about tolerances given for induction hardened and hard chrome plated linear shaft
(see attachment)

let's analyze diameter 4: for ISO h7 tolerance, default circularity at level 0.005 - ok for me
cylindricity (parallelism) at level 0.01 - ok for me
but straightness - 0.3/1000 = 0,0003 ????

straightness must be usually lower than circularity - but is it ever possible to machine such a good way?....
or maybe I miscalculated something here?

I reviewed also Bosch Rexroth specifications, they are using similar descriptions and values

but why not to use runout as it is a mix of circularity of straightness?

M

 

[pmarc]

cervantes,
I would not try to find a general relationship between straightness tolerance and circular runout tolerance, because there is no such. Circular runout, unlike total runout, is not capable of controlling straightness, and straightness by itself is not capable of controlling circular runout. Your example is very specific, because the maximum possible circular runout error is indeed two times the allowable straightness tolerance, but that is not always true. Just picture your shaft with both sides convex within 0.3 (assuming size tolerance allows for that), not bent in one direction. In that case the circular runout error will be 0. You can also imagine a shaft (but this time a stepped one) where the actual cylinder being controlled with circular runout is perfectly straight, yet has serious coaxiality error relative to a datum cylinder, meaning that actual straightness error is 0, but the circular runout error is huge.

I will intentionally stop from commenting in detail about applicability of general circular runout tolerances as defined in ISO 2768-2, because personally I firmly believe that usage of this standard should be limited to absolute minimum if one really cares about:
(1) unambiguity of technical documentation, and
(2) defining product geometrical requirements based on how the product functions.


mkcski,
I would agree that with the ASME Y14.5 specified on the drawing it is often much easier to think through different dillemmas similar to the ones cervantes is having. In most cases it is because the default Rule #1 simplifies a lot of things, and additionally because in ASME there is no general tolerances concept, which - instead of making drawing interpretation easier for everyone - almost always creates more confusion (especially part 2 of ISO 2768). Having all necessary rules in one book also help - no doubt about it.

Perhaps this will be slightly off topic, but my observation is that even in the ASME world many people have difficulties in proper interpretation of the runout concept (whether total or circular), especially when it comes to understanding how different types of runout tolerances affect form and location of controlled features. Maybe this is one of the reasons why in a public draft of the next revision of the Y14.5 the concept has been covered in more detail comparing to what is currently offered in 2009 edition.

And going even further off topic, one of the most interesting questions in that area I recently came across was following:
Is there any geometrical difference between two scenarios described above (Y14.5 in charge)?
(1) A dia. 1.000 +/- .002 cylinder controlled with total runout of .020 to a datum axis;
(2) A dia. 1.000 +/- .002 cylinder controlled with circular runout of .020 to a datum axis.
Anyone?

 

[Belanger]

pmarc,
Are you asking if there's any difference (per ASME) between circular and total runout applied with the same tolerance value to the same feature, referencing the same datum?

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

 

[cervantes]

pmarc, absolutely agree on your examples regarding straightness/runout relationship

taking Euro standard into consideration, I understand that I have just found classic confusion in this general tolerances

because in this particular case - if general tolerance runout is allowed to be 0,1 and straightness at level 0,2, then in case I presented on this drawing both are excluding themselves

 

[pmarc]

J-P,
I think its boils down to following question:
Can one imagine an as-produced dia. 1.000 +/- .002 cylinder that meets .020 circular runout requirement, but does not meet .020 total runout requirement?

 

[pmarc]

cervantes,
But that does not mean both tolerances are conflicting each other. It just means that you can't have this type (mode) of straightness error if you want to satisfy general circular runout tolerance. Like I tried to explain in my previous post, if both/all generating lines of toleranced cylinder are convex (or concave), it will be possible to have straightness error of 0,2 and circular runout error of 0,1 (assuming size tolerance allows for that).

 

[Belanger]

Can one imagine an as-produced dia. 1.000 +/- .002 cylinder that meets .020 circular runout requirement, but does not meet .020 total runout requirement?

Most certainly. I can only suppose that you ask such a question to re-emphasize your earlier statement that "circular runout, unlike total runout, is not capable of controlling straightness."

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

 

[pmarc]

J-P,
I did not doubt even for a moment that you will know the answer to that question ;-).

Actually, I asked it to emphasize the other statement: "[...] even in the ASME world many people have difficulties in proper interpretation of the runout concept (whether total or circular), especially when it comes to understanding how different types of runout tolerances affect form and location of controlled features."

I do not know why, but there is this common paradigm according to which a runout tolerance greater than the size tolerance of the considered cylinder does not control form of that cylinder. This even seems to be supported by the draft of the Y14.5 mentioned by me before (see fig. 12-5). The thing is that if this was true, we would have to say that both scenarios in my example did not differ at all in terms of geometrical requirements. Do you see my point?

 

[3DDave]

Pmarc,

This is the first I've heard about a new draft, but I see that the period is closed already. For that I thank every committee member here for keeping it quiet. Apparently this time it was free to those who emailed the otherwise secret address, unlike the 200x version they charged $120 or so for. It's almost like there is no interest in public comment.

What problems with the existing version are being addressed?

---

My favorite announcement, based on the short time between the announcement and the closing period:

Posted on February 9, 2016 by John Evans • 6 Comments

You can get a draft copy of ASME Y14.5 free and mailed to you. American Society of Mechanical Engineers (ASME) is completing their overhaul of ASME Y14.5 standard for Dimensioning and Tolerancing. If you want a draft copy, you can request it by emailing your name and address to:

Mayra at ANSIBox@asme.org

They will in turn send out a preliminary hardcopy by mail.

Note: This is not the final revision of the standard, but should serve as a very close copy for those that are either not required to adhere to the standard strictly, or those that want to be part of the review process.

The deadline for requests is 23 Feb 2016. After that they will close the public review process,

 

[Belanger]

pmarc,
I agree that there are plenty of misconceptions about ASME (or any standard, really). A runout tolerance can be greater than size, but I disagree that the runout will still control form. Since runout doesn't nullify Rule #1, the size would control form. Runout would merely do its other duties: orientation and location.

So back to your scenarios: If both circular and total runout were greater than size, then neither runout would control circularity. However, they would still be different because circular runout would allow longitudinal deviations (straightness) so that the part may look barrel-shaped and still pass. Total runout would not allow the barrel shape.

So when discussing form I think we need to isolate circularity form vs. straightness form.

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

 

[pmarc]

J-P,
I do not really have time now for a longer reply, but after reading your latest comment I have to ask this:
Why are you saying in one place: "Since runout doesn't nullify Rule #1, the size would control form", and in the other place: "Total runout would not allow the barrel shape"?

 

[3DDave]

I'd go with banana shaped and not barrel shaped to describe a possible runout condition in excess of local size. A perfect barrel would have no runout.

 

[Belanger]

pmarc,
Right -- I have to differentiate between form of circularity and form of straightness! When it comes to circularity, it boils down to this: the circularity aspect is controlled by whichever is smaller: the runout spec or the diameter tolerance. That's why I partially disagreed with you, because runout fails to control that aspect of form.
When it comes to straightness, there's a different factor to consider: whether it's circular runout or total runout. Sorry if I confused things.

3DDave,
I don't know what you mean by your last word, "runout," because the very point I was making is to distinguish the two types of runout: circular or total.
A banana shape would (most likely) fail both circular and total runout. But a barrel shape could have every cross-section perfectly coaxial with the datum, and perfectly circular, thus passing circular runout. If controlled by total runout that same part would most likely fail because of the bulge in the middle of the span.

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

 

[pmarc]

Okay, after re-reading my statement about the paradigm a few more times, I think I need to clarify my point, because the wording I used was not too fortunate.

What I am really trying to say is that since runout tolerances are composite controls of form, orientation and location, there is a mutual relationship between those characteristics, or putting it differently, they are not independent of each other.

In case #2 of my example, in order to have actual dia. 1.000+/-.002 cylinder produced with maximum possible coaxiality error along its entire length, which is .010 (half of .020), the cylinder needs to be perfectly round in every cross-section. Any circularity error will be reducing allowable amount of coaxiality error.

In case #1, in order to have actual dia. 1.000+/-.002 cylinder produced with maximum possible coaxiality error along its entire length, the cylinder not only needs to be perfectly round in every cross-section, but also perfectly straight longitudinally.

Reversing this logic, the dia. 1.000+/-.002 cylinder produced with .010 coaxiality error along its entire length would have to be perfectly round in every cross-section in case #2, and perfecty round and straight in case #1 to be able to meet corresponding runout requirements.

So, I agree (and never wanted to imply otherwise) that in case of runout tolerance greater than feature's size tolerance, the magnitude of form error is controlled by the size tolerance, and the magnitude of orientation and location error is controlled by the runout tolerance, but since extreme values of these errors can never happen at the same time, I wanted to emphasize that there is a relationship between them to recognize and consider.

Have I now straightened up the confusion I created at least a bit?

 

[CheckerHater]

Besides the fact that I've never seen Runout without a datum before, the rest is perfectly clear.

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

 

[greenimi]

Ch,
Please forgive my ignorance, but where did you see the runout without a datum ?
Are you seeing the runout with no datum specified/recommanded or even depicted/ shown anywhere in this thread?
Thank you

 

[CheckerHater]

@greenimi: The bitter irony is exactly that: yes, indeed, there was absolutely no reason to discuss Runout in this thread, because the shaft in question has no suitable datum feature.

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

 

[mkcski]

@greenimi: I may be off base, but let me add my two cents with some "basics". As I understand it , ISO GPS allows Datum Feature Symbols can be attached directly to the centerline (axis) of a shaft (a part with cylindrical features). A drawing pictorially depicts the "perfect" part. So a part with multiple cylindrical features - a shaft - has multiple centerlines, one on "top" of the other, and the drawing shows a single centerline. So the question then becomes, if the Datum Feature Symbol is attached to the centerline, which one of the physical feature does the axis come from, i.e, which one do you set-up to to establish the axis to rotate around. This lack of specificity causes one to guess which feature is the datum. With Y14.5, Datum Feature Symbol must be attached to a physical feature.

As CheckerHater states: there are no datums on the drawing and, per Y14.5, datums are required when applying Runout to simultaneously control the form and position of a feature. So you cannot interpret Runout without a specified datum axis.

I my GDT travels, many confuse "runout" readings with dial indicator readings because dial-indicators are used to evaluate Runout error AND many other types of geometric errors. Dial indicator readings are only "runout" readings when Runout is being evaluated. I recommend being careful with terminology.

Additionally, dial indicator readings present radial errors - the distance between the surface and the axis of rotation. Form and/or position error(s) will cause the radial distance to vary. Runout control is a composite of both form and position errors (and orientation if Runout is applied to a surface perpendicular to the axis of rotation).

A pmarc mentioned: when the Runout tolerance is larger than the size tolerance, Rule #1 limits the form error and position can only be the remaining larger portion of the "composite" Runout error. At my place of work we manufacture large shafts for hydro turbine installations. The functionally critical features have very small size and position tolerance. We avoid the use of Runout. Rule #1 controls form errors. We specify datums - many times a pair of datums (with a dash) - and apply Position to control coaxial relationships. This "separates" the geometric errors and avoids any confusion as to what the requirement is.

 

[pmarc]

mkcski,
In ISO GPS system it is also not allowed to attach datum feature symbols to centerlines. It was legitimate practice in the past, but got prohibited in 2004 when the second edition of ISO 1101 was published.

I agree that if one wants to separate form and location errors, position tolerance is a good choice. Thing to remember, however - it is only true in ASME, because in ISO position (or coaxiality) tolerance indirectly controls form (straightness) of what they call extracted median line of a cylinder.

 

[mkcski]

Thanks for the correction and the update. I have not studied ISO standards in detail and rarely run into them at my place of work. I would love to have a complete set of ISO standards to review but the cost is too much for my bosses' budget.

 

[mkcski]

Pmarc: I have Alex Krulikoswkis' 2010 treatise on ISO GPS. A very good reference. Have you seen any other materials that describe ISO and the differences between Y14.5 and ISO?