Question on inspection frequency and methods...

[quizzical1]

Hi All,

We're trying to become more efficient with inspection of both machined in-house and bought out parts. I'm curious as to what other companies do to accomplish this.

The idea that was brought up was to only inspect critical dim's on a machined part print, but aren't all dim's important to some degree? Otherwise why have them on there in the first place?

Also, it was mentioned to perform 100% dimensional inspection per MIL-STD-105D Sampling Procedure and that would be good enough. Is this true?

As for rubber o-rings and sheet stock, is a C of C and Material Cert good enough? Some have suggested having a sample from each new batch order sent out for analysis to verify it is the correct material. Is this over-kill?

I love to hear what other's do to satisfy this problem.

Thanks,

Quiz

 

[KENAT]

Depending what it comes to mean, the idea of critical dimensions is indeed a slippery slope in my opinion to be avoided.

However, if it means paying more attention to difficult to meet tolerances (be it because they are fundamentally tight or because they are on the limits of the process capability etc.) it may have some merit.

Posting guidelines faq731-376

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

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

 

[redpicker]

How ciritically of a dimension is ulitmately dependent on the applicaiton.

Take the dimensions of engraved markings on a scale, for example. The distance dimension of each mark from the zero datum is highly critical; if this dimension is not correct (within tolerance), there is no way the scale will function as intended. The width dimension of each mark is also critical, but not as much. While it is true that if these marks are too wide or varries excessively, the functionality of the scale is decreased, this dimension is not nearly as critical. Finally, the length of the mark is important as the staggered lengths alow one to read the scale quickly and accurately, but this dimension is not nearly as critical as the others.

Now, part of this is addressed by the tolerances applied. But even with pratical and appropiate tolerances, a scale with the length of some of the markings being 20% over tolerance is essentially as good as one whose markings are all within tolerance. But a scale that has the distance of the markings from the zero datum out of tolerance by 20% can be worthless.

So, in this example, a reasonable inspection could be to inspect a first article scale from each supplier to 100% of all diemsnions, just to document the fact that the supplier understands and is capable of meeting all requirements, then a 10% random sample on receipt, with the sample inspection verifying the distance from the zero datum and just a visual inspection on the width and length dimensions of the markings.

With in-house inspections, you can depend on your process control to reduce the number of inspections. You would want to still perform a first article to document 100% conformance, but after that, depdning on the process capability limits, you might be able to only inspect 1 per 100 and if it is acceptable, you have a high confidnece that all the pieces since the last inpsection are also good. This can't really be done with an outside supplier, though, since you don't have any way to tell which one is the 100th, 200th, etc..., piece.

If you don't want to address the criticallity of the dimensions (this is usually done by the Engineering group, not the QC group), then a statistical sampling plan is always safe.

As far as whether or not you can depend on the vendor supplied C of C, again, it depends on the critically of the component. The C of C only identifies what standards the components were produced to meet; they don't even show any actual testing to demonstrate that the product was checked and found to meet the requirements. If all you are needing to do is be able to document that your Shore 90 O-rings were, in fact, ordered to be Shore 90 O-rings, it should be fine. If you need to show, however, that the O-rings used actually met the requirements, then sending a sample to a lab for testing would be called for.

rp