Alignment Rotation Question

Suggest to engineers that your secondary datum become a ray line starting at datum A and radiating out to a single midpoint of B datum slot. 
It would be defined in feature control frames like this [Control][TOL][A][A-B][C].

--If you are physically clocking the part with high reproducibility (IE in a fixture),
--and if you aren't using the points for a size measurement
--and if you mirror the vector points to be identical for each side of the part... 
You can change your hit vector for those points even beyond a 45° angle so you can get a clear prehit/retract. 

Yep! I have recommended that datum structure ([A][A-B][C]) to engineering on a different part. Do you agree that many parts assemble in that manner? Honestly, I have no clue how this particular part assembles, so maybe I should stick to the print 

Yes, I would physically clock the part as best as I can to the CMM axis,

I'm not sure exactly what you mean by a size measurement. I would not dimension the distance between the points. I would just use gage pins to measure the size.

Yes, I would input the same Y and Z value for the points on each side of the slot (see my pics for trihedron),

I'm not sure what you mean by a 45° hit vector. Do you mean wrist angle?

I have to think about this a bit more before I start my program. This is one of the trickiest alignments I have encountered!

Make the probe hit NON-NORMAL (not square to the feature plane, IE your IJK vectors would be altered from 1,0,0 to like 0.7,0.7,0 for a 45° hit). 

This would make it so you can prehit/retract outside of the slot's narrow width if you can't safely probe these points within the slot.

Hmm interesting - I didn’t know that was possible. However, the part/slot is not small. The probe can easily fit inside the slot, even with a modest prehit/retract.

I’m still not sure how to align to this thing. I guess just measure 2 planes (one on each side of the slot), create a mid plane, and rotate to the mid plane. I don’t love it, but I guess that’s really my only option *shrugs* I suppose the accuracy will be decent since it’s rotating about the center axis.

I was assuming since you said you didn't have room for planes that the part was very small?

I was assuming since you said you didn't have room for planes that the part was very small?

I assumed the same thing in my response.

Can you just measure the OD of the part,  the ID of the part, and take hits along the tab surfaces? Then construct lines, and create pierce/intersect points to that outer circle and then construct reverse lines with the first 2 lines you made originally, and create pierce/intersect points to that inner circle. THEN with pierce points you can construct midpoints so they will be in center of the slot walls. Then with those 2 midpoints create a midpoint. And then with that third midpoint you can create a line from the origin circle to it for rotation? Alternatively you can probably create 2 more lines with the 4 pierce points, then make a midline of those 2 lines

I will throw this method into the ring... It will be pretty close, and repeatable, definitely will want to repeat the same alignment a couple times to get it refined. 
We had the advantage of having a vision system before we had a CMM, this is how we did it on the vision system and translated that experience to the CMM.

Dang, we both posted practically the same idea. What are the odds of that?

I saw you typing too lol. 

I think I get it. Thanks for your help!

I tried peircing ,and the points showed up in the graphic interface on the opposite side of the part. Even if I reversed the line and tried peircing it again. I think pericing is the way to go too...that was my initial idea.

By opposite side of the part do you mean the side opposite of the slot wall? Was the vector of your line going in the right direction? That should work.

Sorry, the point showed up on the opposite side of the OD. My lines were running along the Y axis. I measuring a cylinder on the OD and I also tried measuring a circle on the OD. And I tried reversing the lines. The pierce points kept showing up on the Y+ side of the OD (bottom of the picture)

What was the vector of those lines created? If the opposite side of the part is Y+ at the bottom, then your line would want to be a Y- line to be able to pierce the circle correctly. So if the line is created by points, point order matters for the direction of the line constructed.