Geometric tolerancing of bearing holes to avoid bearing misalignment

[ibrown44]

BACKGROUND: We are in the process of updating our CAD drawings to enable us to out-source production of various components. Many of our components have a similar issue: we have a shaft that runs through two steel ball-bearings pressed into a machined piece of aluminum. As per various bearing manufacturers specs, it is important that the inner and outer rings of the bearings are aligned (i.e. bearing manufacturers typically specify a maximum misalignment between inner and outer rings of < 10 arcmin). Assuming a perfectly straight shaft, then the location of the two bearing holes relative to each other in the part will control the orientation of the inner rings.

QUESTION: What tolerances on the bearing holes are necessary/best-practice to control the orientation of the outer rings of the bearing after assembly? Should the orientation of the short walls of the housing bore be controlled, (i.e. parallel to the main axis of the shaft) or should the orientation of the housing shoulders on which the outer race sits be controlled (i.e. perpendicular to the main axis of the shaft). Or both?

POSSIBLY RELEVANT MATERIALS/SPECS. Typical bearing sizes we use are for 10-25 mm diameter shafts (e.g. bearings 6900, 6205). Bearings are typically spaced 50-100 mm apart. We typically use interference fits (e.g. M7). Ball bearings are deep-groove, steel. Material of machined parts is aluminum.

INITIAL THOUGHTS: From a theoretical perspective, my assumption is that the orientation of the housing shoulders is more important than the orientation of the housing bore simply because that surface is much larger. For example, with a 6900 bearing the housing shoulders present a 22 mm diameter surface, whereas the housing bore is only 6 mm long. Does that make sense? Even if it does make sense, from a practical perspective, our machinist suggested that it didn’t matter whether the tolerance on the orientation of the housing bore or housing shoulder was specified, because it would be impossible to make the housing shoulders non-perpendicular if the housing bore was parallel, and also that it would be easier to measure/confirm of the housing bore was parallel, rather than the housing shoulders. Again – does that make sense?

 

[3DDave]

There are competing constraints which require elastic deformation to match up and which are not part of geometric tolerancing standards.

The orientation of the bearings will be controlled by the shoulders; the bore and the bearing housing will deflect by whatever amount they need to if one shoves the bearing hard enough.

The location of the bearing will be controlled by the bore. No matter how parallel the shoulders are they cannot ensure the bores will line up.

Fortunately there is clearance in the bearing.

If one uses both bores as a datum feature then a perpendicularity for the shoulders to that datum feature can be used to limit the orientation change to an amount small enough for the clearance to accommodate - for certain that number is never zero because the bores will never perfectly align.

 

[ibrown44]

3DDave – thanks for the clear answer.

With that answer in mind, the follow-up question is for geometric tolerancing of these bearing holes (following ASME Y14.5 - 2018 ). Given the above, I think I need to specify:
• Tolerance on bearing hole diameters
• Tolerance on location of bearing hole axes with respect to each other
• Tolerance on orientation of the bearing shoulders

Is the attached drawing annotated correctly to specify those items? In particular, I am not confident about the way that I specified the location of the bearing hole axes WRT each other. Does what I have annotated for the position of the bearing holes (i.e. 0.025mm WRT datum A) apply to the pattern as a whole (which I think is what I want), or to each hole independently? I have tried to find a clear example/explanation of this scenario in ASME 14.5, but have not found one and would appreciate any guidance. The closest example I could find is figure 10-45, which is referenced in subclause 10.5.8.1, which clearly indicates that the first segment of a composite tolerancing is PLTZF, but I was not clear if a non-composite tolerancing (e.g. single segment as I have used) is also treated as PLTZF. I feel like I have just missed something in the standard…

Additionally, is my definition of datum C the correct way to use both bore holes as a single datum feature to specify the perpendicularity of the bearing shoulders?

 

[3DDave]

It is correct - discuss this with the inspection department to make sure they understand how it functions.

 

[Tmoose]

Is this a mature, successful design ?

Most every bearing catalog with an engineering section discusses the seats and the shoulder tolerances in some detail.
AN example is attached.

That same catalog will information about shoulder abutment heights, maximum radiuses in the "corner" between the bearing seats on shaft and housing and the abutment/shoulders

What rotates, the shaft or the housing ?
What is the speed of rotation, and the bearing seal design, and lubrication?
Is the radial load consistently in one direction, like a belt drive, or gravity on a heavy rotor?
What is the axial load ?

What is the shaft-to-bearing fit? How is the shaft retained in the bearings?
What are the shaft details that will positively control bearing preload?
Example, a controlled length spacer between the bearing inner rings etc to go along with a controlled housing shoulder length between the bearings ?

 

[ibrown44]

3DDave – thanks the reviewing the drawing.

Tmoose – This is a mature design. We have a very talented machinist who always makes things correctly, so the issue has arisen because we are going to be out-sourcing manufacturing of these parts now, and therefore we need to ensure that the parts still function as expected.
In terms of your comments/questions and the example datasheet from SKF, I think that all of those relate to the size of the bearing hole. We are confident in that aspect of the design. My question was more specifically related to ensuring that we have the correct orientation of the bearings in the final assembled part.

 

[Tmoose]

Hi ibrown44,

I said "Most every bearing catalog with an engineering section discusses the seats and the shoulder tolerances in some detail.
......... That same catalog will information about shoulder abutment heights, maximum radiuses in the "corner" between the bearing seats on shaft and housing and the abutment/shoulders."

Please spend some time reviewing first tier bearing manufacturer's catalogs. A call to the tech support of your bearing manufacturer, once you are familiar with some of the jargon, would likely be mighty informative too.

You are looking for something like this. -

https://www.gmnbt.com/modx/assets/images/tolerance-shaft-pic.jpg

And this -

https://dg30iru7ycmng.cloudfront.net/wp-content/uploads/2019/04/bearing-eng25.png

pages 22 ff here-

https://www.schaeffler.com/remoteme...ownloads_6/mh1_technical_principles_de_en.pdf

 

[ibrown44]

Tmoose - thanks for sending those links. The catalogs I had been looking at previously were missing this level of detail.

I am curious that GMN and INA use runout for tolerancing the shoulder surface, whereas I noticed SKF uses perpendicularity and total runout for the same surface. Why the difference?

 

[geesamand]

I am curious that GMN and INA use runout for tolerancing the shoulder surface, whereas I noticed SKF uses perpendicularity and total runout for the same surface. Why the difference?

Personally I wouldn't trust anyone who uses runout for keeping a machined feature square to an axis. That's perpendicularity all day long.

Once you have a functional tolerance scheme, the real question is what angular misalignment will the bearing withstand before a degradation in performance. This varies based on many variables such as the bearing type, fixed/floating race fit, internal clearance, shrink fit on the race(s), and vibration requirements. The bearing manufacturer should be able to give at least a basic limit for misalignment for the style of bearing.