Profile of a Curve, Curve Form

[[Ma...]]

I have a customer who had their customer add a CPK of 1.67 to a Profile of a Curve, which is ridiculous. In addition to that I had a great conversation with Ryan Stauffer, with Zeiss.

I've attached a .pdf as the result of that conversation. If you have wondered which to use to report Profile, either Profile of a Curve or Curve Form, I hope this helps
Profile, Curve Form and Calypso.pdf
Mark

[[Ch...]]

thanks for the share Mark. Everything from Ryan is pure gold.

[[Ja...]]

Thanks Mark, that's a helpful document to have on file.

[[Ri...]]

This is an interesting conclusion - "Profile on a Curve should not have a CPK requirement assigned to it."

1st - Good luck getting the Automotive world to buy-off on that.

2nd - I'm not for sure if I agree with that myself. Sure there is no X,Y,Z location or centroid like True Position, but does that really even matter? How is that any different than any form measurement (Roundness, Cylindricity, Flatness, Straightness)? Profile is a compound measurement and covers a wide range of measurements depending on how its DRF is structured.

You cannot have CP on a unilateral tolerance like Profile - you can only have CPK. Saying though that you should not do CPK analysis on Profile because of the calculation being difficult isn't really a valid argument.

Also, I believe that Curve Form was only ever added because of the graphical representation that it gave you. That is negated now with PiWeb Reporting because you get the graphical representation with regular Line Profile now.

What I always suggest to people is to break Profile down into the simpler measurements that people can wrap their heads around - True Position, Roundness, Flatness, Diameter, etc. Doing this will allow you to focus on what is causing the Profile to show OOS (is it its Form, Location, Size, Orientation?). Doing that will allow you address the issues, and correct (hopefully) the issues.

I believe that Profile was designed for Fit/Function - to ensure that a part would assemble and function properly. It just essentially tells you if a part is good and if it will work - after that it isn't really a great tool to tell you why the part is bad.

Circling back to my first statement, if somehow you are able to get the automotive world to buy-off on this, you will have made a lot of manufacturing people happy.

Cheers.

[[Me...]]

Please sign in to view this quote.

Curve form just calculate the results differently than profile. But it's not going to give a green result on a point on the curve that is outside of the tolerance band regardless of the difference between min/max. Not sure I agree with this conclusion. I use curves all the time to check profile.

[[Ri...]]

Please sign in to view this quote.

I agree with you. If your DRF for the Line Profile is the exact same as the Alignment used to measure the Curve - the Curve Form (actual min/max results, not the delta) and Line Profile will exactly the same. I used to actually get the question of why the Curve Form would should Red, but the value was in the tolerance - that's because your max deviation can be out of specification, but your delta be within the tolerance frame. I just always told people to not worry about the delta value - the min/max values are where your focus is.

[[Ma...]]

A little more of the backstory on this part and customer. They sent us the CMM programs and want me to modify them to use in their U.S. facility. The people who wrote the programs used Curve Form exclusively to report profile, it gave them better results, obviously.

This what started the conversation. Perhaps the "Profile should not have a CPK assigned..." was too strong a statement but the intent of it was there is no hard X, Y, Z to target to as with True Position.

My customer is not using 6.6 yet so I'm glad to hear PiWeb deals allows the graph with Profile now.

Mark