Flatness Measurement Errors from a Datum - Test Method

[cslauenwhite]

Hi all....



I have been racking my brains for this for quite some time and I have come up with nothing... So time to ask the experts..

We have a part which is approx the size of a piece of paper and approx 3mm thick. On this part in the beginning was a flatness callout of 0.020mm. The 1st problem came from the question that the engineer wanted to see direction of any flatness errors from a datum.. After much discussion this was changed to a profile of a surface...

The issue I am running into is it is now listed as a offset tolerance of +0.005mm/-0.035mm.

To boot this measurement is also listed as a "restrained state" measurement with 17 lock down points.

Right now we are doing this on a touch trigger cmm with a clamping fixture. Cycle time to complete is 6min. takt time needed is approx 6 sec to do the measurement complete with restraining. Also the GRR (TV) of the CMM method is approx 30%

Any ideas would be greatly help full.

 

[Claes Gefvenberg]

The 1st problem came from the question that the engineer wanted to see direction of any flatness errors from a datum..

That, I can imagine, as flatness can have no datums: It only concerns how flat a surface is, regardless of any other datum’s or features.

To boot this measurement is also listed as a "restrained state" measurement with 17 lock down points.

Which I assume corresponds to how it is assembled? What is the function of this surface?

 

[Ron Rompen]

Cory: Have you considered using an optical flat, rather than a CMM? The GR&R will (probably) not be much worse than the CMM, hopefully better. The throughput time should be significantly faster, which I understand to be one of your significant constraints.

 

[ncwalker]

The flatness question: If this is a primary datum, I don't see how you can control it with a profile because there is nothing to call it back to. So having a non-centered tolerance on the object doesn't make sense. But only if this flat surface is a primary datum. If it is a primary datum, than flatness is your callout and it only has one direction - away from dead flat.

As to your throughput - I don't know the part (other than it's the size of a piece of paper) but 17 restraining points seems excessive. That's a whole lot of opportunity for bolt torque order and amount to warp the thing. And if that is actually mimicking installation points, I'd gather that with that many of them, you could take a part considerably out of specification and draw it up with all the bolts. To increase throughput I would:
a) fight with engineering to get the number of required restraint points reduced
b) fight with your boss/tooling to make you a fixture that would clamp all the points automatically either with air or hydraulics. That's way cheaper than buying more CMMs to keep up with through put. In fact, get two of them. So you can be clamping up the front part while the CMM measures the back one and vice=versa. OR get a specialized machine with clamps and transducers and do the entire measurement one shot at a faster cycle time.