hexagon logo

position of a welded cylinder

Good morning,

I have recently set up a program for a welded flange that has a position callout from one welded cylinder to a perpendicular plane(A) and a cylinder (B) located  on  the flange beyond the welded cylinder .

it has a position of .025 and the program seems to repeat itself well.

my problem is, if you rotate the part 180° the position can come in tolerance or OOT, I am unsure which is most correct, I usually error on the side of caution in these situations but would hate to fail parts that are actually conforming. the difference usually is about .005 or so.

Datum A is established as a plane using 4 points, Datum B is established as a cylinder using concentric circle scans, feature is measured same as datum B. I know perhaps I am not providing enough information, but is there  any advice on how to make it so I am getting the most accuracy?

my current running thought is that perhaps rotating the part has the machine picking up B's perpendicularity (as a welded cylinder, it's not going to be perfectly perpendicular.) differently and throwing off the numbers. 

  • Screen shots help with this stuff.

    If the part is supposedly symmetrical, and you can influence the results by flipping it around you need to look for the differences in your setup. Are your probes perfectly vertical, are you hitting some weld one way but not the other, is the part distorted from the heat. Is the part being held stable one way, but rocking when you flip it around because it's not flat. Many variables come into play. 

  • Another element is manufacturing process.  If the part is symmetrical for example and that "A datum" plane could be one of two opposing planes on either side of the part... You might be seeing the difference as a result of datuming to the opposing side of what the machinist used as the datum surface, when he/she cut, welded, etc the part.