ASME Y14.5M 1994 Good examples for coaxial GD&T? 1

[AHartman]

Hi all,

I'm working on a project where we've been tasked w/ updating drawings for a customer and making recommendations as to nominal and tolerance changes on parts to improve function. This is my first introduction to GD&T and Tolerance Stack Analysis, so excuse any ignorance my post conveys.

The drawings have minimal GD&T, and much of what's there seems arbitrary and inconsistent for mating/related features. Much of the design is based on maintaining coaxiality between several nested/stacked round parts. The typical spec for an overall assembly would be +/- 0.002" axial tolerance.

My research (Krulikowski's Fundamentals of GD&T, Fischer's Mechanical Tolerance Stackup and Analysis, and others) leads me to believe that MMC is generally not recommended for features needing axial alignment, and that True Position or Runout are the correct types of GD&T to use.

Question 1: Am I correct that MMC is not advisable, and that axial alignment is typically done w/ True Position and Runout?

Question 2: Can anyone point me to some examples specifically dealing with axial alignment?

Thanks in advance for any help or advice!

-Adam

 

[KENAT]

ASME Y14.5M-1994 B6 gives an example of coaxial features, the calculations to used for finding tolerances and shows use of MMC with them.

Also, define better what you mean by axial alignment.

For stepped diameters which 'nest' where the driving requirement is that they always mate, Position with MMC would generally be my first choice.

For a situation where one part runs on the diameter of another part, I might consider run-out.

Do you have a copy of ASME Y14.5, and if so what year are you working too?

While the kind of reference books you state have their place, they all to often get contaminated by the preferences/habits of the author - not necessarily strictly what the standard says. Same is true of advice here and is unavoidable as on some topics even 14.5 committee members have slightly different ideas.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[PeterStock]

If the parts are rotating in use then you may need to consider the ballance while spinning. Then I would use runout. In addition notes limiting the unballance and or providing means of adjusting the ballance may be needed.

Peter Stockhausen
Senior Design Analyst (Checker)
Infotech Aerospace Services

http://www.infotechpr.net

 

[AHartman]

Kenat,

We do have a copy of the standard, though it's the '82 version. We're supposed to be working toward the '94 version, though.

As for better defining axial alignment, let me try. It's required that the axes of several holes be parallel and located identically such that line-of-sight from one end of the assembly to another is not obstructed. I can't upload specific drawings/parts, but think of it as a series of washers stacked one on top of the other. If an imaginary particle were to come into the assembly along a perfectly oriented datum axis, and were constrained to follow a smooth curve from one hole axis to the next, the final position and direction of the particle would not deviate from the datum axis by more than a cylindrical zone with diameter of 0.002"

All parts in the design are static, just bolted together to provide a series of holes for particles to fly through. The reason I independently concluded that MMC wasn't useful for my requirements was that, when bonus tolerance kicked in, I found it practically impossible to have a stack of more than 2 or 3 parts meet the spec and have reasonable tolerance values. Maybe I was just applying the concepts incorrectly?

-Adam

 

[axym]

Adam,

1. Position referenced at RFS directly controls the axis of the feature within a tolerance zone. This is appropriate if the feature has a press fit or other self-centering fit with its mating feature.

Position referenced at MMC controls the surface of the feature and ensures that it does not violate a certain boundary. This is appropriate if the feature has a clearance fit with its mating feature.

The runout tolerances directly control the surface of the feature, and indirectly control the axis. The runout tolerances also control the form (shape) of the feature, where the position tolerances do not.

2. I agree with Kenat that third-party books reflect the author's biases, but they are usually better than trying to figure things out from just the standard itself. Krulikowski's Fundamentals of GD&T book is very good if you're just getting started with GD&T. Bryan Fischer's tolerance stackup and analysis book is much more sophisticated and will serve you better when you get into analyzing assemblies. Both books are good for their intended purpose.

Krulikowski's Fundamentals book has several examples and figures showing the different controls for nominally coaxial features. The position section has some examples of position with a nominally coaxial datum feature, but they're mostly at MMC. Part II of the position section shows how to calculate "part distances", which are very simple cases of stackup analysis. Both the RFS and MMC cases are covered. The runout section shows a simple calculation for the worst-case axial misalignment that results from a runout tolerance. I don't have the book in front of me right now, so I can't quote page numbers.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[KENAT]

Not sure if 82 versions differences impact what I'm about to say but...

Try to apply what axym said about if you really care about axis, or the surface.

Not claiming I fully understand but I'd guess from what you say that you care more about the surface, in which case MMC should be beneficial for you.

Simplistically, as the hole gets bigger (within it's tolerance range) can it afford to get slightly off center and still function? This is what MMC on holes boils down to. If the goal is to put a shaft through the holes, then MMC would be usefull, since as the holes get bigger, they don't need to be as well aligned.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[AHartman]

Axym,

Thanks for the points to ponder. All of the parts are clearance fit to their respective mating parts. Most are actually like stacked washers, with no "fit" between parts at all.

It's important that the hole surfaces be mostly round, but the priority is that each hole's axis is nicely in-line with the next. A good analogy would be an optical lens assembly (tubes, focus rings, lenses, apertures, etc, all needing to line up). Tolerance Stack Analysis is driving the GD&T, so I'm kinda stuck in limbo between Krulikowski and Fischer

Mostly, I'm hoping to find some other examples beyond the resources I have, to see solutions from multiple viewpoints and make sure I really understand the overall approach and best practices.

-Adam

 

[Belanger]

For what you're doing, I would also vote for the MMC usage. In the ASME standard (not sure about '82) look at Appendix B -- I think the examples called "floating fasteners" might be close to your scenario. That shows how to apply position tolerances and calculate their numbers when the line-of-sight idea is desired.

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

 

[KENAT]

Reading your 4 Nov 10 11:33 again I think I may have misunderstood, MMC might not make much sense for you.

Ø.002 is pretty tight. In practice is this achieved by fundamental geometry of the parts, or is an alignment fixture or something used?

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[AHartman]

Kenat,

Yeah, the 0.002" is tight. The application is in the field of chemical detection, so high precision is needed to keep the electrical fields and flying molecules interacting in well controlled ways.

The customer's hope is that these tolerances will be approached by fundamental geometry, though we keep coming back to ideas for alignment fixtures. Still, I'm working under the assumption that we have to get as close as possible with just geometry.

-Adam

 

[PeterStock]

I am thinking that an inner sleeve that acts as a pilot for all the pieces in the stack would be one option. If this would interfere with function, then use a shaft as a fixture to assemble the parts over while fastening them together with sufficient force to keep the alignment during operation. If you have any parts with two or more features on the axis, these should be located to each other to something like .001 RFS. The remaining features could be located to a looser tolerance at MMC as the fixture forces the alignment. All featurs should be made perpendicular to a plane normal to the axis within .001 RFS or better. Assuming a part has two planes normeal to the axis used in a linear stackup, I would make the parallel within .001 at most.

Peter Stockhausen
Senior Design Analyst (Checker)
Infotech Aerospace Services

http://www.infotechpr.net

 

[drawoh]

AHartman,

I hope you are looking at an assembly drawing and have a clear understanding of how your assembly works. The assembly drawing also helps you verify your system can be put together.

If your tolerance stack does not work, you have two optiojns.
[ol]
[li]You can reduce the number of parts in the stack.[/li]
[li]You can tighten tolerances.[/li]
[/ol]

Look into reducing the number of components. If some central or external round thing does your centreing, you will have a maximum of three pieces between any two features that must line up.

Rather than just tightening tolerances, carefully examine your geometry. This is where good GD&T practise will help you. Work out good locating procedures.

Do you need accuracy, or repeatability?

Your MMC question means very little to me without seeing your assembly. When I apply positional and diameter tolerances to screw clearance holes, the MMC affects the minimum clearance. The maximum allowable clearance is a design requirement that I must meet, MMC or no MMC. In this context, calling up MMC allows me to meet requirements with a looser tolerance. This all comes from undestanding the geometry.

JHG

 

[AHartman]

PeterStock,

Can you clarify your last sentence a bit? Each part is more or less a disk or cylinder, with end faces perpendicular to the axis of interest. Are you saying that whenever the face of one part becomes the mating surface for the next part in the assembly, that 1st part's face should be parallel to 0.001"?

-Adam

 

[axym]

Adam,

Another thing to keep in mind is exactly how the axes are defined. In Y14.5 an axis is defined using the actual mating envelope (maximum inscribed cylinder in this case) which is rooted in physical fit and assembly considerations. The axes in your line-of-sight application may require a different definition, that reflects the path that the particle would follow.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[AHartman]

Axym,

I more or less understand your point, but really don't have any good ideas how to re-define. More specifically, unless I write out all the inspection steps explicitly, how would a redefined axis help ensure parts met spec? Won't I still have to default to the understood meaning?

-Adam

 

[PeterStock]

If you have a series of disks placed on top of each other as you have in your assembly, any errors due to one side of a disk being thicker than the other could, if assembled in the worst case, accumulate resulting in the last disk being out of the line established by the first part. The first part would have a flat plane A (make it flat within .0005) and an axis B for the center hole. The top surface of the first part is held parallel to A w/in .001. The next part is set up the same, it's Datum A surface mates to the top surface of the first, etc. As you add pieces, the parallelism errors add up. This can make the line thru your centerholes deviate from the ideal straight line. It is possible that the assembly method I suggested may make that a moot point, in which case the flatness and parallelism tolerances may be relaxed.

Peter Stockhausen
Senior Design Analyst (Checker)
Infotech Aerospace Services

http://www.infotechpr.net