Jump to content

Line profile and best fit


---
 Share

Recommended Posts

I have a Line profile which have no datum reference, so I have created a 2d best fit alignment and put the feature of the curve into it.
But I don't think that it actually does the best fit cause on the printout it says that there is no translation or rotation. And I would say that if I look at the plot i would be able to move the profile in the y-axis and get it within the tolerance band.

Can anyone help me out with this issue.
1022_bf5deace0ae01e7b8bc391c30ef3c27b.png
1022_00d6435b29b488cd9cf8462a9d771b4b.png
1022_e7a86ce6e71d39e64486c26d1ef71e77.png
1022_1b2d758f01b4b2ded7160e68324ca1c8.png
1022_d84fc0164637a1942871a5f535d17d55.png
1022_b79da8f2b07868c869ea3988eb5ec20c.png
Link to comment
Share on other sites

First thing I see looking at this is that you have setup your curve to use best fit, which should best fit the points for the curve. If you were to create a curve form characteristic, then the best fit in the curve itself would be used, I am not sure if the same is true for a line profile.

You also use a best fit alignment based on the original curve which probably has already had its actuals modified by the best fit within the curve feature.

I am not sure if this is causing your problem or not, however not having two separate best fits seems like a good place to start.

I would suggest unchecking the best fit in the feature, then creating your profile characteristic and choosing 'no datum reference frame' which should give you the profile independent of any datum reference frame.

It probably would help to make an alignment that locates where the curve should be scanned as closely as possible.

I have found when doing curves like this that one of the biggest issues is scanning the feature in the wrong location in relation to the actual part. By creating an alignment to the feature to be scanned you can verify that the curve is measured as accurately as possible. From the look of your picture, in your case you could probably create an alignment by taking a cylinder from the OD and then a self centering point in the groove, then have the curve be in that alignment.
Link to comment
Share on other sites

Please sign in to view this quote.

When you click the best fit within the feature like he does, it automatically generates that curve BF alignment. I think that's what hes trying to show. But I agree, he should try the "No Datum Reference Frame" option. If that doesn't change it, try using the Tschebyscheff fit instead of the Gauss and see if that gets a better fit.
Link to comment
Share on other sites

Hm....

I use this method a lot with no issues. Typically I don't have all 6 degrees of freedom on though.

You might try something as simple only letting it translate in the X, and Y axis - turn off everything else and see if it behaves normally that way.

I was told long ago by a man whom I would consider the guru of Curve at Zeiss to never allow the Curve to Translate and Rotate in all axis as it can cause issues with the software. d
Link to comment
Share on other sites

So tried changing a few things and i actually got the best fit and translation onto my plot which I am very happy about. But then another question came to me.

The red line on the plot is the tolerance band and that is great. If you are looking my attached image there is a tolerance band in both ends of the y axis. In my mind this should not be there cause then translation in y-axis is limited and looking at my plot translation in y-axis would make my profile within specs.

Anything I can do, or any comments?
1022_986583b7152a2bcc3e3b7df25299ba4f.png
Link to comment
Share on other sites

Please sign in to view this quote.

I think that's just a graphical thing. By the looks of it, its not actually looking at that as a violation of the tolerance zone in those areas. The two little circles you see are your maximum and minimum values within the profile. It doesn't seem to be concerned about the the little piece of profile at the top. Also, that best fit you do is done before the tolerance is set. Calypso has no idea what the tolerance on the curve will be, so it shouldn't take that into consideration at the time the best fit is created.
Link to comment
Share on other sites

Please sign in to view this quote.

I think I understand what it is actually doing.

But is there somehow any option that is capable of doing the translation and rotation to fit my actual curve within a tolerance zone?
Link to comment
Share on other sites

Please sign in to view this quote.

Yes I tried it in on a program and it didn't turn out right, so I thought it wasn't right.
Think I will try doing the program from scratch. Could have been a check mark somewhere that I have forgot.

Actually today I had a chat with a guy from Zeiss telling me that I might benefit from doing a 3D curve instead of a 2D. He talked about vectors could act strange on a turned part like mine.
So I thouht I could just do a 3D curve in the same way as 2D. But I don't think that the procedure is the same.
Is there any guidence materiale on 3d curves. I know Zeiss has. And I will attend their training, but it will be in the fall.
Link to comment
Share on other sites

 Share

×
×
  • Create New...