Freeform surface from faces gives unreasonable value

[[Be...]]

Hello everyone! I am relatively new to working with CMMs and I have encountered a strange behavior when creating a "Freeform Surface" feature using the "From faces" option.

My co-workers have experience with this behavior and have devised a work around by creating another "Freeform Surface" using the same faces to recall or reteach the points and correct the unreasonable form. Before creating the auxiliary feature, the form can be as bad as 1.05 but then becomes 0.05 with this work around applied.

One other method that I found to correct this, but that I'm unsure of, is to turn on "Tschebyscheff" under "Best Fit" in the evaluation for the original feature. This would prevent the need to create an auxiliary feature to recall or reteach the points and appears to give the same results in form correction.

Can someone give me a better understanding of why the auxiliary feature method is needed in some "Freeform Surface" cases, but not all? Furthermore, would the "Tschebyscheff" method be a satisfactory method or are there any issues with it that I am not aware of?

Thank you for any input!

[[Ri...]]

Look at each individual point in the set as displayed on your model,.
You will see a yellow arrow with an attached yellow line, on the model surface of each point there should be a turquoise "x" indicating where the point is visually.
If you cant see the turquoise "x", it may be embedded into the model. As the scan proceeds it will evaluate with any errant point and could cause strange values.

Notice Point10 in the attached image.

[[Be...]]

Hey Rick! Thank you for your response!

I should also mention that I am using multiple faces that wrap around the outside of this part in this case. I have a similar problem for a freeform surface made from multiple faces on the inside of the piece.

We have it set up to scan the surface, so there are many points on each face. The gaps between point sets allow for avoiding the fixture.

Below is a snapshot of one of the freeform surfaces that I've seen this behavior on. However, I don't see any of the "X"s at the base of the arrows going beneath the surface of the model.

[[Ma...]]

I wonder how it looks if you place it to PROFILE characteristic and enable visual evaluation. May be then you can find something.

I had similar problem. When i used on freeform fitting for XY and Rot Z and checked Gauss or Tchebyschev, then i got strange numbers on profile char. but on graphic eval, there were no such things. I don't get it.

[[Ro...]]

Please sign in to view this quote.

You shouldn't need to create another surface or best fit to fix "unreasonable form" of a surface. That's a red flag that you have a problem. Best fitting isn't fixing anything. It's giving you a false result by moving the geometry.

Surface issues like you're seeing are often caused by the wrong nominals being used to check the surface against. Freeform uses the actual CAD geometry so I suspect your issue has to do with combining surfaces together and not having the correct nominals. Have you checked each surface separately instead of combining them? How did you create the freeform surface itself and get the nominals for it? Did you edit any of the surface points after you created the surface?

[[Be...]]

.

[[Be...]]

Please sign in to view this quote.

Hey Martin! Thank you for your reply!

The profile characteristic shows a proportionally "bad" or "shifted" result before I correct it with one of the methods that I mentioned.

What confuses me at this point: The "Minimum" claims to be -1.03 according to the Profile window, but the heatmap does not imply that it varies anywhere near that much. How could the minimum not agree with the heatmap until I apply a corrective method?

[[Be...]]

Please sign in to view this quote.

Hey Robert! Thank you for your reply!

I do agree that this all implies that I have a problem somewhere, but I'm hoping to better understand it using any info from this post. The best fit is not an accepted method for sure, but I thought it had an interesting affect that might be worth consideration. We apply an alignment to this feature in the characteristics, so I believe that may override any best fit regardless.

I have not checked each surface individually yet, but that certainly would be something worth trying.

The freeform surface itself was created thru "CAD"->"Creating features"->"Point Set" with the "Freeform Surface" and "From Face(s)" options. From there, I ctrl+click each face and click "Create" then "Create Feature" after verifying that I get the intended result. Finally, I open up the feature and modify the "Strategy" to selectively delete points that would cause the probe path to collide with the fixturing. Speed is set to 3.0 and step width is 0.125 for the separate point sets of this feature with clearance data between. This results in reasonable actuals with no indication of bad readings.

[[Ma...]]

As far as I know, it's not needed to select faces on model. I select them only in certain situations.
Try to deselect faces, or turn off Point elimination.