Bad EDGE Point Comp

[J Temmen]

Just a heads up to everyone that uses edgepoints. As you can see in the attachment I have two identical points, edgepoints without sample hits, the only difference is the surface vector which should be irrelevant, but look at the actual Z value, it comps to the wrong side of the 2mm styli if I use a -1 I vector. It took me a couple of hours to find,hope it helps someone out in the future.

[kbotta]

By using auto - can we assume you were in DCC??

[Matthew D. Hoedeman]

Version?

[J Temmen]

Sorry, All measured in DCC using 3.7mr3.

[Matthew D. Hoedeman]

Question:

Did you use a paste w/pattern to get the second point? I found in V3.5 (haven't tried in V3/7 yet) that when mirroring edge points, all kinds of problems crop up. The biggest one was that the surface vector didn't mirror even though it ran 'as if it did', but then the results were totally whacked and messed up. I found this on a part shaped like a bowl that had 8 notches in the edge. I used paster w/ pattern to get all 8 notches, but then no matter how I indexed the part, the results were exactly the same for the '12 o'clock' notch, no matter which notch I used as 'master' (as well as the other 6 'patterened' notches). I ended up having to delete the 'patterned' notches and checking them using alignment rotations and NO paste w/ pattern.

[J Temmen]

Question:

Did you use a paste w/pattern to get the second point? I found in V3.5 (haven't tried in V3/7 yet) that when mirroring edge points, all kinds of problems crop up. The biggest one was that the surface vector didn't mirror even though it ran 'as if it did', but then the results were totally whacked and messed up. I found this on a part shaped like a bowl that had 8 notches in the edge. I used paster w/ pattern to get all 8 notches, but then no matter how I indexed the part, the results were exactly the same for the '12 o'clock' notch, no matter which notch I used as 'master' (as well as the other 6 'patterened' notches). I ended up having to delete the 'patterned' notches and checking them using alignment rotations and NO paste w/ pattern.

No, once I discovered the bad comp I went back and switched the vector back and forth on the same point more then once and get the same deviation each time. I just posted both points for visual purposes.

[DaddyBluto]

How would that run? I see you have 0 initial and 0 permanent sample points. If you run an edge point with sample points and a flipped surface vector it will crash into the part, since it would be trying to measure from the inside. I R corn fused.

Jaime

[Matthew D. Hoedeman]

No, the effect of the wrong surface vector would be to make the automove go to the wrong direction (down and left instead of up and left, for example) as well as the sample hit would be on the wrong side, the tip will still try to hit the edge point from the same direction.

Are those measured deviations or is it changing the 'actual' when you change the surface vector? Also, are you changing the surface vector in the edit window or are you doing an F9 on the feature? That is/was/has been a bug-a-boo with edge points for some time now as well, it won't change everything correctly if you do it in the edit window instead of F9. It will LOOK as though it did, but NOT!

[J Temmen]

I don't use sample hits because I don't need them for this hit. I only need to check the EDGE of the part. Thats why I didn't care what the surface vector was as long as it was in an X direction. It "shouldn't" have affected the reading,