Hi All,
Wow, this has become quite a thread. The discussion is similar to previous discussions over the years in other forums, and in the standards committees ;^/. Sigh. I've also had to think about how plus/minus tolerances affect profile tolerances for committee work fairly recently, so this is unfortunately familiar ground. When plus/minus tolerances are involved, the discussion inevitably grinds down into opinion-based interpretation because plus/minus tolerances are ambiguous and don't have rules.
With regards to the original question, I would say that the two drawings comply with Y14.5 and would apply the same tolerance requirements. I wish that it were otherwise and I don't think that combining +/- with profile is a good practice, but Y14.5 does currently allow it. CH's references are correct, and there are figures showing examples.
There is a very mixed message in the standard. Profile is mainly described in terms of true profiles being defined using basic dimensions, and I would say that this is definitely the best practice. Nevertheless, there are figures showing profile tolerances combined with plus/minus size tolerances. In Y14.5-2009 see the following:
-Fig. 8-17 where profile controls "conicity" but not size. This can be extended to apply to JP's initial cylindricity/profile example, since a cylinder can be thought of as a cone with an included angle of zero.
-Fig. 8-18 where profile controls form, orientation, and location (but not size). I would say that the appropriate characteristic for this type of control would have been total runout, not profile. But Y14.5-2009 does not allow (or at least does not show examples of) total runout on conical features. Part of the reason for this is lack of agreement on how plus/minus tolerances would be interpreted (such as the 45 +/- 2 degree cone angle in Fig. 9-2 for circular runout). Some argued that the nominal angle represents the "true geometric shape" and would be different if the tolerance was unilateral (43 +4 -0), others argued that no true geometric shape is defined, others argued that the cone angle would have to be basic. This is a major example of how the ambiguities of old-school plus/minus tolerancing clash with geometric tolerancing.
-Fig. 8-27 where profile controls form and orientation, but not location. In this case the location is height, and is controlled by a directly toleranced size dimension. I've ranted about this example before, on this forum, probably more than once. The only "explanation" given is that "In this application, the datum references only orient the profile of a line tolerance". We are left to wonder exactly how this happens.
These figures are problematic for several reasons. One reason is that they are example-based and case-specific. Without a general description of what a directly toleranced dimension does to a profile tolerance zone, we are on our own with examples that are not shown. Another reason is that these figures undermine Y14.5's own directive of using geometric tolerances to locate features. The plus/minus tolerances introduce non-rigorous and ambiguous aspects such as the orientation relationship between the size dimension and the geometric tolerance zone and DRF (if present). Another problem I have with 8-17 and 8-18 is that the size dimension is on a surface in which the points are not fully opposed - this would introduce additional errors and uncertainty (especially if the included angle was significantly larger than the 15 degrees in the figure).
Regarding whether or not profile can be applied with a directly toleranced radius, who knows? There is no figure showing this, so we really can't say for sure.
Evan Janeshewski
Axymetrix Quality Engineering Inc.