3d best fit alignment problem

[zox20 Junior Member Find all posts View Profile Close zox20]

I want to do a 3d best fit alignment using 8 constructed points I manually inputted with no CAD and no existing axis system in place. 4 are nominal values and 4 are actual values previously taken. When I try new alignment and best fit, select the 8 points, 3 Dimensional, Least Squares, weight of 1.000 to all, the result is no rotation or translation. Prior to trying best fit, the largest 3d deviation from a nominal point and it's actual partner is approximately .023". After best fit, I'm hoping to see an almost equal 3d deviation between each nominal and partner. What am I doing wrong? Thanks for any help.

[vpt.se PC-DMIS Ultra Guru Find all posts View Profile Close vpt.se]

Did you allow any rotation/translation in the 3D-alignment or not?

[sealevel PC-DMIS Guru Find all posts View Profile Close sealevel]

in addition to vpt.se's suggestion, for 3D aln you may wish to use vector instead of least squares.

[DJAMS PC-DMIS demi-Guru Find all posts View Profile Close DJAMS]

Prior to trying best fit, the largest 3d deviation from a nominal point and it's actual partner is approximately .023". After best fit, I'm hoping to see an almost equal 3d deviation between each nominal and partner. What am I doing wrong?

As far as the best-fit not fitting - it sounds like you only have 4 points, but you have entered the nominals into 4 point features, and the actual into 4 other point features. This creates 8 points with no measured deviation, so there is no deviation to fit. You should create 4 generic point features. Enter the nominals into the THEO line and the measured into the ACTL line. This will give you 4 points with measured deviation. Then the best fit will fit.

[zox20 Junior Member Find all posts View Profile Close zox20]

Thanks for the help. I did allow rotation and translation. DJAMs got me thinking. Edit the 4 constructed points of actual values and plug in their nominal values into their theoreticals. Then best fit again only using the 4 edited actual points. Now when I open the alignment to see the values, I have something other than zero. I'll verify the deviation lengths tomorrow at work. Thank you.