Tolerance of Form on a Width

[flash3780]

I know a width dimension has to have perfect form at maximum material condition per Rule #1, but if I need to apply a refinement to control the form of two faces defined by a width dimension, what's the best way to do that?

I know that I could establish a datum from one of the faces and then use a parallelism control, but that seems messy (and a little bit contentious in my particular situation). Is there a better way?

If I have already established a centralizing datum from those two faces, would I be out of line to establish a parallelism on both sides relative to the centralizing datum (see attached)?

ASME Y14.5-1994 and 2009 seem to indicate that straightness can control the form of a cylinder, but I couldn't find anything to show that it will control the form of a width.

 

[drawoh]

flash3780,

Parallelism is applied with respect to a datum. In your drawing, you have applied datum[ ]A to the width. You need to clearly apply it to the side opposite the parallel specification.

Otherwise, your parallel specification makes perfect sense if the parallel value is tighter than the width tolerance.

Your datum[ ]A, as applied, is a "centralizing datum". It does not allow you to control parallelism on the dimensioned faces. I would be very reluctant to use a Feature Of Size (FOS) datum to control parallelism, even of a third feature.

--
JHG

 

[flash3780]

drawoh,
Thanks for the feedback! The centralizing datum is functional for the rest of the part, so it is appropriate in this case.

However, when it comes to controlling the form of the width itself, it sounds like you recommend creating a new datum on one face and referencing it on the other face. I concluded that this would work as well, but I was hoping that there might be a cleaner way to control the form of a width dimension... it sounds like there is not.

ASME Y14-5-1994, Fig 6-1 (or Fig 5-1 in the 2009 spec) gives an example of using straightness on the surface of a cylinder to control waisting and barreling (though, I'm not sure about the difference between this and cylindricity). I basically need the same control on a width as is shown in Fig 6-1. Hence, I initially thought that straightness might provide the appropriate control, but I couldn't find any examples of that usage.

If the above doesn't do the trick, then I think the only way to accomplish a form control on a width is to establish a datum plane from one of the faces and limit the parallelism error on the other face. Seems ugly, but if it does the job, I suppose it's fine.

-C

 

[drawoh]

flash3780,

Straightness, as per Fig[ ]5-1, is a tolerance of form, and it does not require a datum. Parallelism is a tolerance of orientation, and it does require a datum.

Are you worried about parallelism or straightness?

--
JHG

 

[flash3780]

I made an image explaining what type of form control I'm trying to achieve. I think that I can control the form properly if I use a flatness control on one face, establish a datum from that face, and apply a parallelism to the opposite face. It just seems like there should be a simpler way to do this, though.

-C

 

[cbrf23]

For what you've sketched as rejectable I would recommend profile of a line or profile of a surface.

 

[flash3780]

cbrf23,
Hmm.... I can't envision how profile of a surface would simplify the callout any more than the "form_tolerance_gdnt.png" sketch that I made. Can you elaborate on what you're thinking?

-C

 

[pmarc]

flash3780,
Technically, there is nothing wrong with your original tolerancing scheme IF you are interested in controlling form of the faces as well as their orientation to the datum center plane (so indirectly to each other).

The confusing part (at least to me and drawoh) is that you keep saying you are interested in controlling form of the faces, while the pictures you posted seem to be saying that mutual orientation of the faces is a concern too. So if the latter, you can stay with your original scheme (provided that both tolerances specified as .XXX do not somehow conflict with each other), but if the former, all you need is two flatness callouts applied to both faces of the width.

 

[flash3780]

pmarc,
Thanks for the feedback. I was considering the width to be a single feature (aka feature of size), so my intention is to control the two faces in a way similar to how cylindricity would control a cylindrical feature.

I apologize if this was causing confusion. In fact, I would like to control orientation of the face features relative to one-another.

I wasn't sure if the original sketch would be problematic since I'm establishing a datum from the width dimension, and then referencing that datum with the paralellism requirement. However, I can't think of another simple way to get the control that I want if that's not kosher.

-C