Jump to content

Rules of thumb for 2d curve evaluations


---
 Share

Recommended Posts

Hey all,

I am looking for some input on general 2d curve evaluation settings. I know this may be generic, but I figured I would put out a post. 

 

For some background. I have came across a few instances in my company where the same feature in different programs are being filtered differently.

the examples below show a 2d curve evaluated with spline with 50 upr and a second instance where it is evaluated with Gauss with Lc 2.6.

image.png.89467e104f7fc48dfd105e1be2c2541d.png image.png.9514cb874057200651a908d56ac22972.png

 

Appreciate the help!

Link to comment
Share on other sites

Hey Tom

problematic is always inside radius.

1st What's the best stylus radius

2nd probe center path defines step distance, minimum 18 points at 90° arc

3rd Determine the break even Lc for spline filter and compare it to Gaussian

I found no background inside the standardisation and the used parameters for Spline-Math is also a mystery

The result must be plausible. It is best, if I have similar results, e.g. CMM versus form tester.

 

  • Like! 1
Link to comment
Share on other sites

Thank you

Please sign in to view this username.

! That is very helpful. 

Based on the best practice note. For my example, I am measuring a spiral 2d curve. Since the slope tightens towards the center, I should be using Gaussian. Am I interpreting that correctly?

Link to comment
Share on other sites

Hey Joseph

the gradient growes up for inner to outer, means not constant.

I prefer Gaussian, because it only can get problems at start and end. An adequate overlap (cutoff) solves this.

Additional compare the result Gaussian & Spline with same Lc to get a feeling for the behaviour.

In order to make more precise statements, I need to know the exact geometry
Please look at inspection
Klaus

Spiral.zip

Edited
  • Like! 1
Link to comment
Share on other sites

Thanks Klause,

So my print calls for the below Lambda filtering settings in regards to profilometer measurements. Should I be using 0.8mm for my filtering method when it comes to CMM filtering?

image.png.e2752ceab2c5b8b1b524abb561d6e212.png

Link to comment
Share on other sites

Hey Joseph

The typical specification of a roughness

Evaluation length 4,000 mm

Lc = 0.800 mm

Ls = 0.0025 mm

Samples = 5

Standard CutOff = 1

Measuring length = 4,800 mm

There are no specification evaluate the total profile of the spiral

Use the spec above, and the number of samplings = MeasuredLenght/Lc

Link to comment
Share on other sites

 Share

×
×
  • Create New...