How to measure a part without a model or drawing-not consistent...

[[Zo...]]

I want to measure an old template that has no drawing or model, so I clamped it on the table, recognized some features, top, side, end, radius, set up the base alignment and measured it. The actual was off from the nominal, so I took the actual values and put it in as "nominal", and remeasured it to see if all lights will be green, and to see if it confirms the values and if it repeats.
I measure the part 5 times without moving it, and got 5 different readings, ranging from R 16.1371 to R16.5483.
I am interested in the 2 radial dimensions and the center locations for the 2 radii. On one center the "X" ranges from 5.1737 to 5.2233and that is just too much variations. I can do better with a tape measure...
Is there a particular way to reverse engineer a part? To measure something that has no drawing? Am I missing something in the way I approach it? Not looking for super precision, a simple +-.005 will do, but +- 3/16 is too much
I attach a drawing showing what it looks like, a 0.06" flat template, approx 8" x 4" overall.

Thanks for any thoughts

TEMPLATE MEASURING.JPG

[[Mi...]]

That's a very typical experience when measuring such a small arc of a large radius. I've heard using touch points rather than scanning may help, but I can't vouch for that myself.

In any case, since you're reverse engineering, I dimension it as a basic "round number" radius (or maybe the average of all your runs?), and add a profile tolerance. That should repeat just fine.

[[Zo...]]

Please sign in to view this quote.

I ended up using constraint, separately for the radius and locations, and now my measurements seem to repeat within a few tenths.
I will use that method on other templates, hopefully they will work as well.