Using a Auto line (Scanning) to detect 0.001" steps on the OD of a Cylinder

[acgarcia]

Good Morning,

I was told to scan a line along the body (parallel to the cylinder axis) of a cylinder to detect a defect from a CNC machine. The defect is 0.001in step about every 0.100in along the body of the cylinder. They also want to see a graphical analysis. Not familiar with scanning lines in general or showing profiles.

I'm using an auto line with 1.0 pnts per mm with an LSP probe. I tried using profile dimension but the graphic comes out all wacky. I tried straightness as well which made the line look a little better but it doesn't represent the defect on the part.

I was changing the points from 6 down to 1 per mm to make the graph look a little better. What they want to see is a more or less straight line with a dip representing the 0.001 step and the repeat all the way down the body of the cylinder.

Can someone offer some advice?

[acgarcia]

This is what my graph looks like, its not what they want to see or how'd they imagine the line would look.

Looking at the part, you can visually see the 0.001 steps in the part. its stainless steel so the steps pop out so to speak. It doesn't match what the CMM is showing.

[JEFMAN]

I'm not at the cmm, you have to dimension the t value or the scan.

[thomas pilc]

i would take at least 6 to 10 points per mm and i would use auto cylinder line scans and slow it down and then use profile on only one of the lines and see what you get do like 5 or six line scan and use profile on each and see if your profile shows view shows anything also might have to turn up the scale on the profile dimension to like 50% or 100% to see what you want

[louisd]

Within the autofeature dialog, there's a "Measurement Strategies" tab. make sure you select an adaptive cylinder line scan and utilize the settings within the little sub-tabs of that strategy.

[Mike Ruff]

Your stylus could be too big to detect such small grooves also

[acgarcia]

thanks for the tips, I will try them out now.

[Ego Murphy]

Tiny stylus - smallest possible.
Accurate machine - a tenth or better uncertainty
Triple layered alignment to get right on the OD centerline
Slow linear open scan with high density
Copy paste w/pattern a bunch of times around the cylinder
Profile dimensions
Analysis Views
Color printer
High-fives and tasty beverages of your choice.

[acgarcia]

I changed from a 5mm ball to a 1.5mm. Scan speed set to 5.0 mm/s and 10 pnts/mm. I think my problem is not really understanding the results. The print doesn’t have have any gd&t other than the OD of this cylinder is +/- 0.0005in. If I scan an auto line and make a profile with analysis view, what am I looking at? I through in a straightness dimension and it shows a straight lines with arrows above and below the line in a uniform pattern. Ok don’t think that represent the part either. I’m not looking for a dimensional value, the thing I need is the graphical analysis to be clear.

the part is turned in a lathe so what it I’m seeing every line from the cutting insert in my graph? Could that be why the graph doesn’t show a straight line and a dip for these 0.001in steps?

[louisd]

what you showed in the pic is the straightness of a line, which would not be relevant to the cylinder(cone) axis line. If the bad cutting was consistent as you describe, the straightness would still be rather good for that line... But the cylindricity (or radially speaking, angularity of the line to the center axis of the bore), will show the conical steps you are trying to graph.

Also, by default, graphs will show deviation arrows relevant from the LSQ (Least SQuares average) measured value. If you want all the deviation arrows to show stepping in the same vector direction, you would need to construct the cylinder as Max Inscribed/Min Circumscribed (depending on if its a bore/boss).

[JEFMAN]

If you dimension a line, pc-dmis construct a ls line, and then dimension the straightness.
What you're looking for is an evolution of the radius, so a distance perp to the axis.