measurement strategy

[[mi...]]

Hi. I have a little argue with my co-worker about dimensioning. We have to measure distance as per attached image. Calypso did not want to measure intersection of middle cylinder and circles on both sides. When asked for caliper distance between two intersections it was coming out with Reference error and did not give result. My mate created plane on middle cylinder one end and  intersection of one side circle with that plane, on other side circle intersected with cylinder. Is using plane correct approach? Would results be accurate ?

[[Cl...]]

I'd scan a radius on all three surfaces. Then construct two tangents. left to middle and

right to middle. That should give you two tangent points to use in caliper distance.

[[Ma...]]

Well there can be many solutions which can give you similar results. But presented strategie with plane is surelly wrong - using circle/circle will sometimes come with no intersection.
Also seems neither one of screenshots shows correct intersection point. That "x" mark should be on line from CAD which divide between middle cylinder and radius corner.

I would use circles - using 3d line connecting centers of middle circle and one side circle. Then instersection of šd line with middle circle. This will give you one point.
Repeat that for second side circle to obtain second point.

[[Cl...]]

I just tried my method in simulation, and it works fine.

Edited April 24

[[mi...]]

This have to be measured pre-rad. I experienced issues with circle intersection before but my main question is if using plane as shown on second picture (on cylinder) would produce accurate result  for pre rad distance ?

[[Da...]]

I don't think you can get a "caliper dimension" from the intersections of cylindrical features, since the intersections would be 2 lines with no guarantee of parallelism. Measure between 2 intersection points calculated from radii measured at the same setback, not lines in 3 dimensions.

[[Cl...]]

Use the center radius as XY zero and a 3d line constructed from the left/right radii for the planar.

Constrain size on each radius. Intersect center with right/center with left. Caliper distance. I just ran this in simulation, works fine.

[[Pe...]]

 

I'm pretty sure this is what Martin was describing above.  I don't like circle tangents or intersecting circles with circles since you get zero results or multiple results more often than not.  This has always been the simplest solution that has always worked for me.

[[Ma...]]

Please sign in to view this quote.

That's what i meant. This is a geometrical solution.

But i wonder what if those radiuses won't be tangential. If you machine those radiuses on correct spots, but with different R value - then any solution will give not accurate result.
This dimension is bad - it says nothing about a part and it's questionable result. Much better would be profile callout or ignoring side radiuses and using only mid cylinder and plane to get that intersection. You will get result combined from real radius and position relative to used plane.