Actual Size Changing During Program Run

[1CMM4ME]

I Just don't get it. I am manually programming this part. I took 6 points on a radius. I labeled them as points only. Then I chose to contruct a circle in regards to using these points. A prompt came up and told me that there was a conflict of value size ( trying to remember--I'm at home now), If I selected "yes" the circle was not on the centerline of the points. If I selected no, the circle was constructed right on center of the points. Now, the fun part begins. Once I chose no and the circle was presented in the graphics screen as being in the center of the points, I decided to measyre the dimension. I chose "R" for radius and gave the prompt box the correct tolerancing for the radius. I clicked o.k and the actual size dimension showed up fine in the program. So, I executed the program to run full. When it was finished, that stupid dimension was out .011 compared to what it read when I first wrote the program. Then, when I soomed in on the graphics screen, the constructed circle that was originally on center of my points was showing to the outside of the points taken and not on center of the points.
I am sure this is something simple, but I can not figure this one out. Any help would be appreciated. I am including the section of the program which I created the circle by construction and the measurement.
Thanks everyone. Note: This is what the dimension showed before i ran the program.After the program run the actual was like .411.

CIR4 =FEAT/CIRCLE,RECT,OUT,LEAST_SQR
THEO/0.00191,5.74824,1.17792,0,0,1,0.74422
ACTL/0.00562,5.74996,1.17791,0,0,1,0.74352
CONSTR/CIRCLE,BFRE,PNT49,PNT50,PNT51,PNT52,PNT53,PNT54,,
OUTLIER_REMOVAL/OFF,3
FILTER/OFF,UPR=0
DIM LOC3= LOCATION OF CIRCLE CIR4 UNITS=IN ,$
GRAPH=OFF TEXT=OFF MULT=10.00 OUTPUT=BOTH
AX MEAS NOMINAL +TOL -TOL DEV OUTTOL
R 0.37176 0.37211 0.00000 0.01000 -0.00035 0.00000 --------#
END OF DIMENSION LOC3

[John Kingston]

First thought is that points taken manually versus DCC could yield a different result. None of us can drive the CMM at the same speed and correct vector, while in DCC the CMM can. That said, are you correcting the vectors for those points that you are constructing the circle? If you "Learn" a circle, PC-DMIS will correct the vector for each point in that circle. BUT, the software cannot and doesn't do that if you take individual points. (Unless you are using Vector points or surface points.)

Second, have you tried measuring the feature using methods other than the CMM? Micrometers for OD or pins for ID. I am a big advocate of still using plate inspections. You can learn a lot about what the CMM is doing by comparing results.

I'm not sure, but probe compansation could affect this. Others here have far more knowledge on this than I do.

[1CMM4ME]

Hi John,
Long time no talk. Actually I am programming DCC mode. This stupid thing is pissing me off. All I did was take points around the end of the part. Then I decided to construct the points as a circle. Then I measured the circle ( but chose radius) because that's what it is. Everything looked good. ( I already checked the part on the comparator). Then when I ran the program, the stupid thing says the radius is bigger than what I got when I originally measured it during the programming. This kind of stuff drives me to drink. I don't know. It all seems giberish after 12 hours. Thanks ,atrick

[John Kingston]

I just noticed there is about .030 " difference........
Make sure you are using the same probe diameter as your probe file......
.030 " is really close to 1mm......
I have had that happen.......