How would you do it ? Brain teaser ?

[[ZE...]]

So, this part is a production part several thousand pcs. The top chamfer is actually a spherical radius. Tolerance for the rad. is +/-.002 ( inches ). There is a gage line height ( .177 diam at .560 +/-.002 height) The sph. is 20 degrees of arc on each side of center line for a total of 40 degrees of arc.

We are running this in a cnc od grinder. So , it dresses the radius into the wheel. Diamond wear and shape can affect the roundness of the profile and has to be monitored. Also because its grinding only on one side we have to watch the spherosity of the finished part ( in other words we can produce a football shape if we go past center, or a doughnut shape if we fall short of center ).

So, we have this running really well, but I would like to see how you gentlemen would go after this.

One last note, I am using a Micura, XTR head and a 2mm probe

[[ZE...]]

An optical comparator?

[[ZE...]]

With a +/- .002 radius tolerance , We ran 100 pcs with the comarator, they were rejected for incorrect radius . We also tried a line reader and a quadracheck. Repeatability was not very good. We think we were producing footballs.

[[ZE...]]

Whoops, I need to correct something, the +/-.002 was on the sphere not the radius. That would need to be +/-.001.

[[ZE...]]

Given the nature of what you're trying to accomplish, I would not measure it as a sphere. I would do a series of radial circles--say, eight. Each circle covers only 20 degrees of arc. Some would say take single points, but I would simply scan each segment twice, in opposing directions, and not too fast.

I wouldn't constrain anything. Look what it gives for the size and locations of each circle. If it's a football, they will all be shifted towards the opposite side.

Does that make sense?

[[ZE...]]

I had to do something similar not too long ago.

I used 4 circle paths horizontally and eight radii along the axis like what Aaron mentioned, but in a sphere.

I wish someone had asked this at the time, because I did end up having to reverse the directions on them due to the short length and the amount of points that were filtered out.

I didn't figure it out until I couldn't get decent results, and realized Calypso was filtering a lot of my points.

Anyhow by measuring them in both directions I was able to maintain good data for the entire length short though it was, combined with the four circle paths it gave very consistent results, and since the entire part was dimensioned off the c/l of the spherical it had to be good.

[[ZE...]]

The designer's duty.
See attached.

Sphere_Champfer_1.jpg

[[ZE...]]

I hope you're not suggesting that the designer should use gauge points beyond the extents of the spherical surface to set the tolerances.

Profile of a surface, anyone?

[[ZE...]]

The designer defines the priority and
sometimes the measurement strategy.
See attached.

Sphere_Champfer_2.pdf

[[ZE...]]

This is awesome !!

First of all thank you Guys. It's hard to get this kind of validation here at work. Most don't understand what we do... Thanks again.

This is how I ran this.

I used a sphere ( 5 circles ) to establish X&Y zero. Recalled two circles on the cylinder into a 3d line for spatial. Scanned the mag for Z. Looped to .0001

Ran a circle on y axis, 30 single points, no filter, outlier set for only OL and skewed to 1.0 inside 3.0 outside. (The diameter next to the sphere is not ground and sometimes runs out to the sphere, so setting OL this way takes out low edge points better than if I leave it at 3.0 3.0) NO constraints.

So this is repeated on the other side. ( I did 4 Circles to begin with but found very good results just using 2, That was after I used the sphere to align X&Y ).

Now all the points are recalled into a 3rd circle.

We monitor the radius of all three circles. The closer they are to each other the better...Typ .0003 dev and we make adjustments.

Center X and Z coordinates also vary a bit.

For the gage line height, I made a theo cylinder and intersected it with the sphere I use for alignment. This matches the drop gage and cupped tip we use at the machine.

We get occasional whacked out numbers for one or the other circles and usually rotate the part and check again or get the next part to see what's going on before making machine offsets. We can also adjust to correct the football/ doughnut condition in usually one offset.

Customer called and said they look like ball bearings now !

Before me, they would check the radius and the gage line only. This could produce all kinds of strange geometry. We still have to keep some of the old timers from going back to that method, But they're leaning.

[[ZE...]]

Andeas Your second illustration Shows exactly what we're up against. Now add the potential radius variation and things get really weird. You can have an O/S radius, bad X & Z coordinates and still use sphere to check it and it will show insize !! Same with curve.

[[ZE...]]

Putting a poorly selected straw man profile tolerance down and then crossing it off doesn't make profile a bad approach, nor does it make gauge points beyond the physical part a good one.

If the form is more critical than the position, then the appropriate thing would be multiple profiles or a composite profile.

And it is NOT the responsibility of the designer to determine the measurement strategy.

[[ZE...]]

Just my two cents.
See attached.

[[ZE...]]

Andeas Your second illustration Shows exactly what we're up against. Now add the potential radius variation and things get really weird. You can have an O/S radius, bad X & Z coordinates and still use sphere to check it and it will show insize !! Same with curve.

Dave Scott! | 03-23-2018 03:27 PM |

[[ZE...]]

Putting a poorly selected straw man profile tolerance down and then crossing it off doesn't make profile a bad approach, nor does it make gauge points beyond the physical part a good one.

If the form is more critical than the position, then the appropriate thing would be multiple profiles or a composite profile.

And it is NOT the responsibility of the designer to determine the measurement strategy.

Aaron Gerber | 03-23-2018 05:30 PM |

[[ZE...]]

Something about dimensioning from a CALYPSO user.
See attached.


Andreas Binder | 03-26-2018 11:46 AM |

Sphere_Champfer_3.pdf

[[ZE...]]

Something about measurement strategy from a CALYPSO user.
See attached.

Andreas Binder | 03-26-2018 12:07 PM |

Sphere_Champfer_5.pdf

[[ZE...]]

There should be profile call out instead of Radius. If arc is less than 180 Degree result vary too much.

Dilipkumar Shah | 03-26-2018 01:43 PM |

[[ZE...]]

Dilipkumar, That's true if I was just checking the parts for conformance. I need to direct the machinist HOW to correct the part.

The Micura that I am using is very tight. I can repeat the readings on individual parts within .0002" on this program. We need to sometimes realize that a particular feature may check higher or lower due to partial clean ups or the occasional point that hits off the sphere. So the reports on this part are NOT all pretty, but the end results are a happy customer and pretty good production rates. That's what its all about, Make a good (enough) product at a reasonable price.

Dave Scott! | 03-26-2018 03:03 PM |