Interesting comments KENAT.
It still surprises me how different my instincts often are from those of you in the design field. How you reference the datums makes a lot of difference, especially to the inspector! And to the material review board tasked with dispositioning nonconforming parts, when this "non critical" outside profile somehow failed inspection even though the parts work fine. And to the quality manager, when the assemblies have less than the required .050" of overlap even though the detail parts passed inspection.
The functional datum features are what they are. You can specify something else if you choose, but that doesn't change how the DOF's will get constrained in the assembly. All it changes is how the inspector will be obligated to constrain the DOF's during inspection.
Even after all of this discussion, we never did give DSETERS a good answer for how to specify a good datum scheme for the functional situation as described. Jim, you said that the only way the rounded ends would come into play is if everything is at MMC and there's mega-slop. Well, that's what the situation actually is! These bosses interface with an as-cast channel with .050 profile tolerances on the surfaces. There is going to be slop! But you suggested changing the design to take the rounded ends out of play and perhaps put a press fit on the widths. Are we redesigning the part now? If we are, why not just make the bosses round? I'm not a design engineer, but I can't see how the widths can constrain the left/right translation more efficiently than the round ends would given the part geometry. If there was a .005" gap between the boss widths and the channel, the cover could move left or right by about .015". Let me know if I'm completely out to lunch here.
I guess I'm more used to working within the confines of a given situation as opposed to changing it. This mindset probably comes from years in the CMM room, trying to reverse-engineer the real functional datum features of a part in order to make sense of nonsensical inspection results.
Sorry about the bitterness and ranting. I just remember that whenever there was a disagreement over part conformance or inspection results, the three most common causes were:
1. Datum Reference Frames
2. Datum Reference Frames
3. Datum Reference Frames
Evan Janeshewski
Axymetrix Quality Engineering Inc.