Skewed/Outlier Point in Profile

[[Jo...]]

We have had an intermittent problem with skewed points in freeform. I have checked everything I know to look at, so I would like some support from you guys. I apologize that I am not able to send better pictures due to customer restrictions.

We get a bad point that throws our profile way out. As shown in the screenshots below, there is one point that is way worse than the others. Also, it appears to be "stacked" on the adjacent point. NOTE: Values in report are in inches.

When you look at the points in the feature, you can see a similar skew. This seems to correlate with the "stacked" points theory above. Most importantly, the form in the feature is nowhere near the .0999" shown in the profile report. It usually lands around .010"-.015" which is typical for these parts.

I have verified that there is no such defect on the part, and this has happened on many various occasions in the past. It does not repeat consistently in the same spot. Sometimes it jumps from one point set to another. Sometimes there are multiple outlier points. And sometimes, it doesn't happen at all.

Freeform settings:

• Strategy
  • Multiple point sets from curves
  • .100" step width
  • .750" speed
• Evaluation
  • Best Fit: Off
  • Point Elimination: On
  • Outlier Elimination: On (3)

Have any of you seen this before? Do any of you guys have any idea what causes this, and how we can prevent it?

[[Ma...]]

Seems like you have same number of nominal and measured points - as many states on this forum - you should have 3 measured points for 1 nominal point.

You either reduce nominal points or increase measured points - that should help.

[[Jo...]]

Please sign in to view this quote.

Thank you Martin, this seems to be consistent with all other occurrences of this issue too. Do you know why this is the case? Is it simply a best practice for feature calculation, or is it based on some other reason?

[[Ma...]]

Maybe it's for calculation ( to be more accurate ) or for cases where more nominals would take same actual.

I also don't understand when you have curve with adaptive nominal distance ( on straight path it would be less point density than on curved path ), because now you have somewhere nominals 2mm apart and then you have nominals 0.5mm apart.