Filtering Free Form Surfaces

[[Cl...]]

I have a six sided (OD) hex with a .001 surface profile (.001 U .000 A B). I measure the six planes and recall them into a FFS. Without any filter applied, I get .0018, if I apply the Gauss filter, I get .001. Not sure if it's actually a good part? I have never used filters in a FFS? Can anyone elaborate on this?

[[Ke...]]

Are the same areas filtered if you are just looking at the individual components of the hex, with the same filter settings?
If so, is there an obvious reason why those areas are filtered out?

[[Jo...]]

More specifically on this subject. Filtering a freeform object or point cloud needs associated geometry to apply those filters ? In individual features like a plane, there is a "target" of sorts knowing the nominal shape. Recall a single plane and see the data discrepancy in evaluation.

[[Cl...]]

Per my general practices, I applied the default spline filter in all six planes, as they are used for other things. I then applied both the Gauss and the Tchebysheff filters to the FFS and saw no difference between the two. However, they both give me a good part at .0008 over max. I don't know exactly what these filters are doing?

[[Cl...]]

Please sign in to view this quote.

I saw no difference when filtering just the plane, or the FFS with that same plane recalled, both Gauss.

[[Ri...]]

Please sign in to view this quote.

I don't know if this is true. Filtering is a wave-function equation. Outlier Elimination is just a statistical analysis of the raw data points (maybe this one is impacted by the intended geometry, but possibly not).

Evaluation algorithms are directly impacted by the intended geometry certainly though.

I would assume that the filtering in a FF would work the same as the filtering in a Curve.

[[Jo...]]

Please sign in to view this quote.

So outlier elimination is filtering based upon neighboring data point ?

[[Cl...]]

Please sign in to view this quote.

outlier:

A data point in a set of results that is very much bigger or smaller than the next nearest data point. Based on this definition, I would say yes.

[[Cl...]]

Is this a generic and simple way of explaining what filters and outliers are?

Filters remove/ignore surface finish imperfections. Outliers remove/ignore
something out of the ordinary, like a burr, dirt, pit, etc?

[[Ri...]]

Filtering is a smoothing of the data. Depending on what frequency you select will dictate how much smoothing you apply. Its intent is to remove machine harmonics, and yes surface finish.

Outlier is intended to remove abnormalities in the measurement. Its original intent was to save re-measurement time from unclean parts. Outlier elimination is the largest grey area tool there is as you can unintentionally remove real data (like a gouge, etc) that should be included in the form characteristic.

[[Cl...]]

Please sign in to view this quote.

With that in mind, should outlier not be ticked when measuring form?

[[Cl...]]

Please sign in to view this quote.

[[SH...]]

My new company have free form license, feeling over the moon.. 😃 😃 😃 .l have to fire in all cylinders to master free form .....

[[Cl...]]

That's great. This will give you much more flexibility.

[[Ri...]]

Please sign in to view this quote.

I don't want to upset anyone saying "should" again. Lol.

[[Ke...]]

Please sign in to view this quote.

🤣
But, one definitely needs to use caution, as selecting "outliers" can easily discount relevant data.
I would rather *not* use outliers, and review "failed" characteristics; but then, I typically only deal with benchmarks, prototypes, and supplier PPAP part validation.

[[Ri...]]

In high production environments, Outlier Elimination is extremely helpful. For FAI, PPAP, testing, etc. I would say to use it with extreme caution.

[[Cl...]]

Here's the explanation I got from Zeiss:

The Gauss checkbox is actually doing a best fit and here is what those three checkboxes do along the top of the window:

• Checking JUST the “Best Fit” checkbox alone, and neither of the other two: Performs a Gauss-method fit using the actual data and the “real,” full/original CAD model surface(s). Compared to…
• Checking Best Fit + Gauss (or just Gauss): Performs a Gauss-method fit using the actual data and a simplified, triangulated representation of the CAD surface(s).
• Checking Best Fit + Tschebycheff: Performs a Tschebycheff-method fit using the actual data and a simplified, triangulated representation of the CAD surface(s).

Am I wrong, or does this indicate that filtering a FFS can only be done when there is a DRF?

[[Ri...]]

That doesn't sound like filtering. That sounds like the algorithm being used to calculate the Best-Fit. Kind of like the different algorithms you can select when doing a Best Fit Position.