need advice on measuring a radius

[[Am...]]

Hi guys, I want to see what you guys think of this dimension I need to measure.
[img]
[/img]

So the highlighted area is a cylinder (a very small part of the cylinder). The amount of the cylinder you can measure is roughly around .300" but the radius is almost 5" and the tolerance is ± .002". I honestly can't be confident with this im getting to scan less then 1% of that diameter, what do you guys think? I have done this using a variety of different methods and each once produces different results. The customer is not happy with the hold up and they are saying their previous supplier was measuring this with a CMM with no issues. I definitely don't want to just go with whatever gives me favorable results I want to be confident in my results.

The back side of the part is even worse, its a cone with similar dimensions asking for the radius at the edge. This is a molded plastic part and Im just not confident with making any adjustments to the mold based on these results.

[[De...]]

Assuming there is something that you can use a decent datum structure to define the nominal center point of the radius, you could create alignments to the center of the radius at different angles. Then create space points which use the alignments to get the radial distance from the centerline of the 5 inch radius to the points. By taking several points along the radius you can verify the radial distance at the ends as well as in the middle of the radius. This should allow you to give a radial measurement from the nominal center point of the 5 inch radius to the actual position of each point you measure. I think this is the best you will be able to do given such a small segment of the radius to work with.

[[Da...]]

This is very similar to a problem I am having. The rule of thumb is to scan a circle and constrain X and Y axis to get a radius reading, then constrain the radius to get X and Y coordinates. Yes it works , but I feel this is cheating somewhat. The way I prefer , but most of our customers can't quite wrap their heads around, is to section and apply a 3d curve to the surface. Then since the model is ( supposed to be ) at the mean of the model. apply the tolerance to the form and display your curve with the mean line and tolerance bands on either side , This will show that all the points you collected fall within tolerance. Unfortunately it doesn't produce a radius size or location in a numeric display for the inspection report. It works perfectly and clearly shows exactly where the form is in respect to the model. So that's not a solution that they like . Lets see what the forum comes up with , They are pretty good with this stuff 🙂 🙂 🙂

[[Am...]]

Dave that is what I was doing with that section you see in the image, but like you said as far as a report its not on the report the same as its called out in the drawing. This is a print from the 70s and there is zero chance I can get them to change the callout.

What Derek is suggesting is something ive been thinking about trying but I'm not really sure how to make it work what would I use to create that center?

[[Ja...]]

I had the same issue once. Constraining the location made a big difference but that was the nominal location based on the base system and the radius was still out of tolerance a little bit. I took a bunch of space points on the radius, used them in a 3D alignment, used the 3D alignment for the cylinder feature. Then when I constrained the XY and Normal Vector, it was in tolerance. I wouldn't feel any worse about using that method that they feel about asking for the radius of something like that. I created a file in excel that you can evaluate a 3 point radius, adjust one point until the radius falls out of tolerance and you can see what little deviation it takes to throw the radius off.

small_rad_seg.xlsx

[[Ro...]]

I know i'm a blaspheme, but whenever im dealing with a small portion of a feature, i use single points.
ive never been as low as 1% before, but ive had good luck with cones & spheres using 12 to 15 touch points.

12 points-2.png

[[Ne...]]

You could use formula in Result Element feature. In Result Element formula section pull up the minDev of the 3 D curve and add it
to the nominal . Do the the same for the maxDev. This would give them a number that matches the print.

[[Pa...]]

Measuring .3" of a circle with a circumference of 314" and holding a .002" radial tolerance is impossible by a magnitude of like 100 times. If the customer is taking this measurement seriously then they don't know beans.
PaulSchuyler

[[Am...]]

That's exactly what I said especially considering this print is from 1972, they claim their other supplier never brought this up and hasn't had an issues with it. Which is bs the previous guys just had no problem pencil whipping it in.

[[Er...]]

Getting accurate location is out of the question. So to me the most efficient way of measuring that would be to measure it as a cylinder, constraining the two location axis and the nominal vector, then either add a couple circle paths or take single points in your strategy and use the "radius measurement" characteristic. This will report each point taken as a radius from nominal centerline. It's basically like treating it as a profile of a surface type evaluation, this will ensure that at any point along that surface you are within ± 0.002" of nominal.