Determining What to use for measurements

[Quadra]

We are looking to purchase a 3d scanner to help speed up our inspection time. We have many dimensions that have XXX, which I Know should be measured by a factor of 10 beyond that. Unfortunately, purchasing a scanner that is accurate to .0001" is nearly impossible and would be cost prohibitive if it was. The scanners we are looking at are accurate to .001", if we verified any dimension that was within .001" of the tolerance limit with a more accurate device, would that be an acceptable industry practice?

 

[JNieman]

The first thing you should do is evaluate whether or not all measurements are only accurate to within +/-.001 or if the expected deviation is .001" over the entire volumetric capacity. That's a big difference.

As to what is "good enough" - that depends on your parts, your customers, and what your acceptable limitations are.

Before looking at machines at all, you should decide what degree of precision is required over 'x' inches, on a small scale, and what degree of precision is required over 'y' inches over a long distance. You should be involving your machine reps on this, and them be able to demonstrate it. Before we bought our Romer Arm with RS3 scanner, our dealer came out and duplicated some inspections we'd already performed on our Zeiss Contura bridge CMMs. We were able to compare/contrast a "known quantity" with the results from the demonstrated arm.

 

[geesamand]

First, all accuracy specs should be tested. Completely ignore any precision / repeatability specs. I have dealt with inspection machinery that did not have a clearly published accuracy spec and even then it could not achieve that.

The 10x rule of thumb works well for hand tools but not for digitally controlled equipment such as 3d scanners and CMMs. 5x or 4x is much more practical.

Lastly, for accuracy to .0001" I would expect the need for a bridge type CMM system. Thermal stabilization time of the samples, climate controlled inspection room as well. I haven't kept up with 3D scanning solutions much but in the realm of accuracy a .0001" or .0002" accurate system seems quite optimistic.

As you shop for 3D scanning solutions, look very hard at how the system takes in the data and derives the characteristics on the drawing. 3D scanning can give you relative surface errors without much trouble but the real magic comes from how it processes the data and generates actual cylinder/plane/etc features and measures the drawing tolerances.

 

[KENAT]

Surely you need to worry what the tolerances you are checking are, not how many decimals the dimensions have. Then look at using up to 10% of that for your inspection allowance.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[3DDave]

10% is a convention, but if you are using equipment to accept parts that are 10% out of tolerance you have a problem. Just be sure to allow for whatever error you end up with as part of the tolerance stack calculations.

 

[Quadra]

Thanks for all the great answers! I was planning on taking the error factor into account, I was just not sure if their was standard out their that prevented from using something that was less than 10x. Looking for scanners has proved a bit more tricky than I thought at first. I was hoping the price had come down more. To find one that will measure the part and compare it to a cad file is pushing over $300,000. Our budget is less than $100,000, any recommendations on something that might work?

 

[JNieman]

We have a Hexagon Romer Arm with an RS3 scanner attached to a laptop running PC-DMIS CAD++ 2015 and while I'm personally a bit amateurish, I've been rather impressed with the results of measuring some of our masters and gages during periodic calibration checks.