Gage R&R concept - Nearly 50% of the QC engineers rejected the part

[anc1984]

A component manufactured with finish process like surface grinding was handed over to a team of QC engineers. Each one made an as accurate measurement of dimension as possible, but it turned out after comparing notes, that nearly 50% of the QC engineers indicated the part to be out-of-tolerance range and hence classified it as 'Reject', while another set of QC engineers sweared that their measurements are correct and the part confirmed to dimensions and is a 'Pass'. How can I solve this problem in a technical manner that can be convincing to both the team of QC engineers. Assuming that a measuring instrument has defined measurement capability and the equipment is a standard measuring tool, explain how you would address and resolve this problem

[Wes Bucey]

I have never before witnessed a situation where two separate camps of equally talented people have taken a completely polar view on whether a characteristic was in spec when both have access to the same instrumentation (which is how I interpret the situation you posit.)

I will follow this thread with interest. Is there anything more you can add so we might have a clue where the discrepancy between the two groups comes from?

[Arvind]

Your question is general hence let me answer in a general way. If you have specifics, we will discuss these as well.
1) Imagine a precision engine component which has a very tight tolerance range of few microns on a nominal dimension.
In such cases, stabilizing of temperatures of part is very essential, otherwise the component dimension will change from morning to afternoon to evening.
Make sure that all measurement conditions are identical.

2) In above case where tolerance band is very narrow, selection of measuring equipment becomes very important. Generally CMM machines have a resolution of 1 micron. A general rule of thumb is that resolution of the gage should be atleast 1/10 the tolerance band.
Sometimes tolerance specification may not exist for a process output. Then, resolution of the gage should be 0.6 times standard deviation of the process or better.

Arvind

[qualitytrec]

Are these QEs using an inspection instruction or procedure to help insure that they are all measuring the same way. For example is one group checking it while holding it, thus transfering body heat to it, whlie the other has it held by some sort of restraint that will not transfer the heat? Or is one checking it when the sun is shining through the window on it and another at a different time when the window is shaded? Something must be different. is one applying more presure to the instrument? Is the insrument shielded from body heat? What type of measurement is it and what tool are you using?

Also, how many QEs are we talking about here? And did they in fact use the same instrument?

I would be and am asking lots of questions.

I am interested in what might be found. Please keep us in the loop.

Mark

[wmarhel]

Originally Posted by Wes Bucey

I will follow this thread with interest. Is there anything more you can add so we might have a clue where the discrepancy between the two groups comes from?

I'm with Wes on this, and my curiousity is peaked.

The original post was vague and I hope some additional information can be provided.

Wayne

[howste]

Welcome to the Cove!

I'm guessing from the forum that this is posted in and the information in the poster's profile that this is a purely hypothetical question.

That being said, I would complete a gage R&R study (which should have been done before measuring to begin with). It may well show that the variation in the measurement system exceeds the tolerance on the part.

[Vicman]

You may want to verify the setup of the length.
Normally the reading of the surface testers is an average of certain stroke length. You may want to check if is the same all the time.
Is the direction of grinding perpendicular or parallel to the gage sensor? This will affect the reading.
With so general information that my guessing considering there is an R & R already on this gage.