How to build a datum reference frame using feature patterns

[[Ch...]]

Hi all. I'm up against a high hurdle and I'm hoping someone with more expertise can explain what to do and why.

I've attached a mock up of the relevant items from the print. Third angle projection.

Datum A is the combination of three planes. The two diagonal planes are at a 40.63° angle to center horizontal plane. Datum B is the pattern of the four holes in the center.

How in the world do I go about building a datum reference frame for A|B? I'm guessing it involves the "Geometry Best Fit" but I really don't know what I'm doing when I use that. Any help is greatly appreciated. Thanks.

Mock up1.png

[[SH...]]

I have not seen this like that datum in my machine shop life.... 😮 😮

[[De...]]

It looks like Datum A controls 5 degrees of freedom, then only direction B could control is the last one.

Datum A is a 3D feature and controls 5 degrees of freedom.

Assuming the bottom face in the 'V' is in the Z+ direction:
The bottom face controls translation in Z
The angled faces control translation along X
The bottom face controls rotation around X and Y
The angled faces control rotation around Z

That only leaves the holes to control translation along the Y axis.

The first issue is creating a Datum out of those three planes though, not really sure how you get there since there can be error in the angle in both directions of the two side planes. The only thing I can think of would be to create a symmetry construction between the two side planes.

Now make an alignment which uses the bottom face for Spatial orientation, then use the symmetry for planar rotation
Use the symmetry for X origin
Use the Bottom face for Z origin

Y origin remains undefined.

Create a geometry best fit using this alignment as its alignment, then choose your four holes and only allow translation in Y

That geometry best fit is now an alignment to A | B I am not sure A-B is a thing that is possible given the geometry.

Anyhow thats my take on it, but it is ugly - they should have chosen a better way to define the datum features.

[[Cl...]]

What about a "Best-fit" looped alignment. Would that even up any error in the two planes?

"The first issue is creating a Datum out of those three planes though, not really sure how you get there since there can be error in the angle in both directions of the two side planes. The only thing I can think of would be to create a symmetry construction between the two side planes."

[[De...]]

Well really even what I said has a bias toward the bottom face.

After thinking about it some more, I think you could make a geometry best fit alignment using the 3 faces and allowing it to best fit in every degree of freedom except the Y translation with the evaluation method being outer tangential on all three planes. That should emulate a fit of all three faces on a datum simulator.

You could then use that alignment for the starting point for the geometry best fit of the 4 holes and only allow it to best fit in Y.

Kind of convoluted but I think it might work. I honestly would have to try it to find out as it is pretty complex.