True Position: Datum Shift Discrepancy

[bfire85]

I have an angled brackets. Two holes on each surface, angle is at 125°. Datum A is a plane, Datum B is a hole through Datum A, which is perpendicular to A. There is a hole on a 125° angled surface in line with Datum B in the X axis. Datum C is a line created between Datum B and the angled hole. Part is positioned parallel to the Y-axis on the CMM. When I apply the MMB to Datum B, the position gets worse than if I measure it against it's physical Y & Z locations. Why would the datum shift move the two holes closer together (see shift of -0.379 in Y). How is this possible?

Running 2019 R1 SP1.

PCD2019R1SP1_TP.PNG

[Sora5]

Is Line C 2D or 3D? If you drop Datum C from both reports, does the numbers make more sense? The few times I ran across this same issue all had something weird with my rotation lock. I think I ended up making some intersection circles between the holes(cylinder) and datum plane and then creating a 2D line from that for my Datum.

[bfire85]

2D line. If I drop Datum C from the equation, same thing. Datum A is a plane, Datum B is a circle, Angle_Hole is a circle.

[Mike Ruff]

Can you measure the angled hole as a cylinder? I would try that and check the perp to CL checkbox. It looks to me like your deviation is along the axis of the angled hole. If you calculate the angle based off the deviations you get ATAN(.466/.326)=55.024° and 180-55.024=124.975. Circle positions calculate based off X, Y, and Z. You need to use a cylinder to ignore deviation along the axis and only measure the deviation perpendicular to the axis.

[Sora5]

^ Makes sense. PCDMIS probably went bonkers trying to fit a 2d circle into a 3d calculation using the circle's vector which is not perp to the primary plane. If there's not enough surface for a cyclinder, would it be correct to project the angled circle to the leveling plane to negate the deviation along the axis?

[bfire85]

Cylinder Time Mike Ruff Sora5 Leaving PC-DMIS up to it's own devises. Still doesn't make sense.

PCD2019R1SP1_TPcyl.PNG

[Mike Ruff]

The numbers it shows may or may not make a whole lot of sense, but the measured value makes sense now at least. I would trust it since the angle calculated from the deviations is still close to 125°, most of the .426 and -.300 are along the axis of the hole. It is essentially telling you that your hole is "deep" but on center.

[bfire85]

Mike Ruff I can only hope PC-DMIS is correct. Usually I can back up the TP value from the deviations. However, it is really hard for me on this one. Usually I don't deal with TPs on an angle like this, especially with MMB allowed.

[bfire85]

I am still struggling with the math behind the way PC-DMIS is calculating the TP value. Anyone else have any insight? JEFMAN KIRBSTER269 vpt.se VinniUSMC AndersI sealevel

[bfire85]

Yeah I get a -0.002 in X

[KIRBSTER269]

Don't be fooled by what you see in X, Y or Z. If you truly want to see what the true numbers are, move your trihedron to the center of the feature your calling out. You know, by moving it according to your basics. I had a position of a hole that called back to A,B,and C. that had 3 different basic angles. When you looked at X and Y the numbers didn't make any since. The other guy that runs the CMM had the machinist moving X, never getting a good part.(When in fact it was a slight movement in Y) I told him and the machinist you can't go by those numbers. PC-DMIS hides everything but does the calculations with the shifts and rotations. If you want to truly understand the numbers. Plug in your basics in an alignment to get your trihedron in the center of your feature. 0,0,0 Then you will see the true numbers you seek.

[KIRBSTER269]

Kinda like JEFMAN only difference is he shows all his work, PC-DMIS does not.

[JEFMAN]

Did you think about a design error ? (so PC-DMIS confused)
On a classical reference system, you find first a plane (or a cylinder), second a line and third a point.
With 3 planes, the first datum is a tangent plane (L2 constraint or minimax), the second datum is the intersection between this plane and a plane perp to it and tangent to the COP of second plane, and the third is the max point along the line vector of the intersection.
In your case, the first is a plane, the second is a point (even if you measure it as a cylinder, because it's constrained to be perp to A), and the third is a line perp to A and B, but it's a little hard to be perp to a point !!!!!
If you'd construct the reference system as an alignment, I think you would first level and origin on A, then rotate on C and origin on B.
Maybe I'm wrong (I didn't have my first coffee !).

Rob Jensen : any idea ?

[AndersI]

I am still struggling with the math behind the way PC-DMIS is calculating the TP value. Anyone else have any insight?

You have DEV PERPEN CENTERLINE=ON. That means that the TP is calculated only from the deviations perpendicular to the theoretical circle/cylinder center line. Any deviation along that center line is ignored. This also means that there is no intuitive relation between the TP value, and the coordinate system XYZ you see in the summary.

See the PC-DMIS Help, Available Help Systems -> PC-DMIS Core Help File -> Using Legacy Dimensions -> Dimensioning Position -> Deviation -> Perpendicular to Centerline

[VinniUSMC]

To add to this, if the surface that the angled hole goes through is high or low, that changes where the circle is, and the cylinder ignores that.

What I think is weird is that your Datum B has 5 times the tolerance of the hole you're measuring.

[bfire85]

VinniUSMC Lol! Oh the discussions we have had with this part! Believe it or not, it's fallen on deaf ears. Can you believe that???

[vpt.se]

I would display all three coordinates in the summary since all three values are interesting, especially if (when) the alignment isn't parallel to the toleranced features centerline.