Issue with radii corners in Curve

[[Ja...]]

When I run a curve profile, the corners are looking very wonky (see pic below).  I have tried different filtering methods, but nothing seems to work.  Any ideas?

 

Using a 1mm sphere probe on a Calypso G2 with RDSXXT.

Edited February 10
More pics

[[Cl...]]

What size is the radius being scanned there? That 1mm styli might be too large? In addition, have you tried adjusting the speed and point density?

Edited February 10

[[Ja...]]

They are .060" radii.  I also tried dropping the points down to 2000 and still having the same issue

Edited February 10

[[Cl...]]

I would try measuring them separately as radii. If that's an option. Single points and constraining location which ever work-plane it's in. I'm assuming in XY? The other thing I'd look at is the surface finish of that edge. If it's too rough the styli might be "skipping" along the surface.

Edited February 10

[[Cl...]]

Constrain location for size, size for location.

[[Ja...]]

Its a limited dimension drawing, so a lot of these features are .010 profile to |A|B|C|.

Yes, its in the XY plane.

This part has several areas like this, and I am trying to get away from doing circles (using TP) and planes (using Profile) for everything.

 

[[Ja...]]

I also have a similar issue on a slot feature.  The radius on the end of each slot is .050.  I used a 1mm dia probe and ran it slow (.05 for speed) and taking a point every .025".

 

Here is the result

[[Ja...]]

OK, got another one doing the same.  Running a profile around a boss, the corners are modeled sharp, are probed as sharp, but shows chamfers on plot, and is out of tolerance.

 

Any help would be appreciated. 

 

 

[[Ma...]]

In that case you should measure them as individual curves, model that as a small radii or use more segments.

Problem is scanning around sharp corners and evaluation also - similar with tolerance band.

You can also mask some of problematic points around corner.

[[Ja...]]

Please sign in to view this quote.

should I remove the evaluation then?

How do I mask the points on the curve?

[[Ma...]]

Masking -> on Standard tab "Nom. Vector Direction" ( 3rd window on your screenshot - in the middle )

[[Ri...]]

Jake - a few things:

• The recommendation is 3:1 of actual to nominal points - so in a lot of your cases you have 2000 nominal points, so you should be collecting 6000 actual points. 
• Sharp corners are extremely difficult to measure correctly because 99.9% of the time there are no nominal points on the sharp corner, so the software doesn't have a great idea on how to round the edge. Very rarely do true sharp corners exist, so sometimes you can "cheat" with an unknown curve to get it to wrap around the edge correctly. Most of the time I would recommend to just measure the segments independently (say you have a 200 point nominal curve that point 1-100 are the first line of the wall, and 101-200 are the second line of the adjacent wall; you can tell the software to measure 1-100; and then have another segment inside of that curve to measure 101-200). You also might have to limit evaluation (Nom. Vector Direction) right at the edge. 
• You are possibly defining too many nominal points. My general recommendation is when you build them, tell it 25 points per segment. 

[[Ja...]]

Thanks Martin and Richard.

Another question I have is about the differences between the cad model and the plot.  In the pictures below, you can see the black line representing the original geometry shows a angle to the radii but in the actual model there is no angle.

 

 

[[Ma...]]

That black line is your nominal curve, which will be made out of nominal points. Not a real spline as on CAD.

You can enable "show all nominal points" to see every point. This can help you to understand.
To enable it, just open curve feature and right mouse button on CAD view -> settings -> check all nominal points.