Help with explaining to the Machinist how there part is out of tolerance

[[Ky...]]

Please bare with me this might take a little to explain. I can not post the whole print but I have snips
We are producing a large hinge weldment. The weldment consists of 2 side arms, 1 cross member and a shaft goes thru the top of the arms and sticks out of both sides of the arms.
Here is the problem we have found to make the parts in tolerance we have to put the 4 Datum B spot faces in after welding. Datum B is the primary datum ( 1 spot face goes in each arm and 2 spot faces go in the cross member.
Datum C is the outside of the left arm. Datum A is the cylinder on the shaft that sticks out of the left arm
We have to put a .531 hole in the center of each spot face with a true position callout of .014 MMC to B C A
Here is how I am doing this so please tell me if I am doing it wrong.
I have made each spot face a Plane, then recalled all points from those planes and made my Datum B plane.
I have taken over a dozen points on Datum C . And I have 4 circles on the Datum A and made those a Cylinder Datum A
So when I run my program I use my Datums for my base alignment. Datum B is Spatial Rotation, Datum C is Planar Rotation, Datum A Cylinder A is X and Z origin and Datum C is Y origin. This also puts my X,Y and Z in the proper place per the print.
Here comes the problem. We have run over 200 of these parts with only 2 failures. Not a bad average. The failures are in true position in the 2 holes ( 1 in cross member , 1 in the arm ) on the far side of the of the part. The sticking point is in the report the True position in hole 4 is .045 . X dimension should be 19.06 and the actual measurement shows to be 19.0376 and Y dimension should be 17.52 but measures 17.5227 using the Datums to constrain the part as called out.
But when the machinist measures he can not find the 19.060 being 19.0376 so the CMM is measuring the part wrong.
I tried to tell him that he could not constrain the part to the Datums properly so there is no way he can measure it properly.
I'm trying to get some other insight so when the machinist goes to the owner and tells him the we in QC don't know what we are doing that I can give a proper explanation of something other then that is what the CMM says.
If you read all of this and can give me some insight thank you so much.
Kyle

Datum B Spot Faces.pngBCA DRF.png

[[Ia...]]

Do you use your base alignment for the true positions? What happens when you plug in the actual datum features for your data within the position characteristic?

If your reference frame is [ B ][ C ] [A ], Datum B should be controlling a translation degree of freedom. Unless I'm reading this wrong, which could very well be true.

[[Ky...]]

That is how a I am doing the True Position B C A. Not just using base alignment

[[De...]]

Without seeing a print or having a better idea of what you are attempting to explain, it is hard to give any definitive answers.

It strikes me as a bit ridiculous that Datum B would allow 0.01 profile when it is the primary datum and would appear to control rotation about two axes and translation along one axis. That much error in the primary datum used in the true position callout would seem to almost certainly be the cause of a substantial amount of error in the true position of the holes.

I would want to look at how manufacturing was aligning the part when measuring it. It has been my experience that many engineers / machinists and even plate inspectors do not realize the amount of influence that order of precedence in the datum reference frame can contribute.

I would be curious what the result for profile is on Datum B (which could be checked with a form characteristic I would think btw)

[[Ky...]]

Profile on the 4 pads was .0013, .0005, .0019, .0011
Even the Shaft to pad measurement are 5.0315, 5.0313, 5.0319 5.0316
We are checking all parts and profile of the pad profiles are one of the items called out to be checked.

The machinist is constraining the part in the center of the shaft not on the Datum A cylinder on the outside. Also his y stop on datum C is just 1 point. So he is actually flipping his Secondary Datum and his tertiary Datum (and not even using the proper part of the shaft) I know it is all the same shaft but it has had hot ears put on it because it is a press fit so we heat the arms up to 300 deg so they will slip on. Then we drive dowel pins into the part. Then we weld it all together. So we are at a disadvantage on to where Datum A is. He says he is finding the center of datum A bit that is only in one place on the entire datum a cylinder

[[De...]]

My next step would be to emulate their check on the CMM.

So probe where they are, and make new characteristics that perform the same measurement they are.

If after doing so your measurements and theirs match one another, then they will be far more likely to believe your original measurements and theirs differ due to how the datums are being constrained.

[[Ky...]]

Thanks Derek,
I did just what your said and you were correct. I can get the exact same numbers the floor is getting measuring how they are. But when I constrain to the Datums per the DRF the numbers turn bad. I have both sets of numbers in the program and on the printout.
So now all I have to do is get everyone together and show them that just because he measures in the machine and says it is right, unless you constrain it properly like you said it still isn't correct per the print

Thanks
Kyle