Greetings all!

One of our customers reports decimals out to three significant digits after the decimal (.XXX) on their drawings, but will reject the lot if they find any measurement that is not less...but rounds to .XXX. These are very tight tolerances being met with very good but barely capable measurement systems (i.e. the process is capable but the parts can't be measured reliably with current tolerance window).

I am torn on this...as I think the customer should almost always be accommodated...but the other side of me cries ASME Y14.5 standards (called out on their spec) require that any basic dimensions and related feature control blocks be the same number of decimal places. I.E. if a tolerance is set at .XXX, it must be reported at .XXX....I can't ignore the existence of significant digits...my 10th grade science teacher will haunt me in nightmares! And that just isn't healthy. Am I right in sticking to the standard or do I ignore scientific convention and go with the customer?

:thanx:

Greetings all!

One of our customers reports decimals out to three significant digits after the decimal (.XXX) on their drawings, but will reject the lot if they find any measurement that is not less...but rounds to .XXX. These are very tight tolerances being met with very good but barely capable measurement systems (i.e. the process is capable but the parts can't be measured reliably with current tolerance window).

I am torn on this...as I think the customer should almost always be accommodated...but the other side of me cries ASME Y14.5 standards (called out on their spec) require that any basic dimensions and related feature control blocks be the same number of decimal places. I.E. if a tolerance is set at .XXX, it must be reported at .XXX....I can't ignore the existence of significant digits...my 10th grade science teacher will haunt me in nightmares! And that just isn't healthy. Am I right in sticking to the standard or do I ignore scientific convention and go with the customer?

:thanx:

Answered my own question. ASME Y14.5M-1994 2.4 Interpretation of Limits

"All limits are absolute. Dimensional limits, regardless of the number of decimal places, a re used as if they were continued with zeros....the measured value is compared directly with the specified value and any deviation outside the specified limiting value signifies nonconformance with the limits."

Answered my own question. ASME Y14.5M-1994 2.4 Interpretation of Limits

"All limits are absolute. Dimensional limits, regardless of the number of decimal places, a re used as if they were continued with zeros....the measured value is compared directly with the specified value and any deviation outside the specified limiting value signifies nonconformance with the limits."

There's been discussion here before about the rounding issue. In lieu of an agreed-upon standard such as Y14.5M, it should be assumed that you can't round your way out of an out-of-tolerance situation. If the upper limit is .200, and your measurement result is .2001, it's out of tolerance unless there is explicit understanding to the contrary.

There's been discussion here before about the rounding issue. In lieu of an agreed-upon standard such as Y14.5M, it should be assumed that you can't round your way out of an out-of-tolerance situation. If the upper limit is .200, and your measurement result is .2001, it's out of tolerance unless there is explicit understanding to the contrary.

That's exactly what ASME Y14.5M-1994 says.

Re: 2.4 Interpretation of limits. (Page 25)
"To determine conformance within limits, the measured value is compared directly with the specified value and any deviation outside the specified limiting value signifies nonconformance with the limits."

Stijloor

That's exactly what ASME Y14.5M-1994 says.

Re: 2.4 Interpretation of limits. (Page 25)
"To determine conformance within limits, the measured value is compared directly with the specified value and any deviation outside the specified limiting value signifies nonconformance with the limits."

Stijloor

Yes, as the OP pointed out. In many cases there's no reference to Y14.5M (or any other standard), so it has no standing in those cases. It should be assumed that any number greater than the upper limit (or smaller than the lower limit) is out of tolerance, regardless of the number of decimal places used in the specification.

Yes, as the OP pointed out. In many cases there's no reference to Y14.5M (or any other standard), so it has no standing in those cases. It should be assumed that any number greater than the upper limit (or smaller than the lower limit) is out of tolerance, regardless of the number of decimal places used in the specification.

In such cases, the outcome of (heated) negotiations (Re: MRB) will likely prevail...:( regardless of Y14.5M).

Here is a very common scenario:

Question one: "Who is the customer?"
Question two: "When does it have to ship?"
I'm sure you've been there...

Stijloor.

In such cases, the outcome of (heated) negotiations (Re: MRB) will likely prevail...:( regardless of Y14.5M).

Here is a very common scenario:

Question one: "Who is the customer?"
Question two: "When does it have to ship?"I'm sure you've been there...

Stijloor.

Yes. I was speaking in theoretical terms. The fact is that "nudging" of out-of-tolerance findings happens all the time. If you're not sure of the integrity of a supplier's inspection reports, it's always a good idea to verify until you are, especially when findings are close to the limits.

Yes. I was speaking in theoretical terms. The fact is that "nudging" of out-of-tolerance findings happens all the time. If you're not sure of the integrity of a supplier's inspection reports, it's always a good idea to verify until you are, especially when findings are close to the limits.

If you graph supplier inspection results (of COURSE this would happen in any Cover's organization :notme:), it is astounding how often there is a "bump up" departure from normality just inside the spec limt and a "dip down" departure just outside the limit. Pure coinsidence with no statistical significance, of course. :sarcasm:

Geoff Withnell

If you graph supplier inspection results (of COURSE this would happen in any Cover's organization :notme:), it is astounding how often there is a "bump up" departure from normality just inside the spec limt and a "dip down" departure just outside the limit. Pure coinsidence with no statistical significance, of course. :sarcasm:

Geoff Withnell

That's called statistical artwork...:)

The company I used to work for had a sister company as one of our major suppliers. They managed to provide us parts showing distributions on the outside of both the spec limits. We already figured out who were getting the "good" stuff...

Stijloor.

This discussion comes up frequently in our shop, particularly with the CMM which will report dimensions out to 5 places. Is there an ISO spec that adresses theoretical absolutes of measurement or rounding? Does measurement uncertainty have any effect here? I often get stymied when trying to explain exactly why we can't simply round down when a measurement is "just a little bit" out of tolerance. Thanks!:thanx:

This is an old thread and refers to ANSI Y14.5M. We are now with ASME Y14.5-2009 and the rather definitive paragraph has been dropped from 2.4 INTERPRETATION OF LIMITS:

"For the purpose of determining conformance within limits. the measured value is compared directly with the speicified value and any deviation outside the specified limiting value signifies non-conformance with the limits."

We have one customer who requests that if the call-out is 269.5 +/- 0.4 we should report to one significant digit.
Being able to ship product that is .044 above the upper limit works to our favor and makes production happy but is completely contrary to my thinking.

Should one always inspect and report exactly as called out?

I actually found a good way to explain it to my management... Think of it as a dial indicator. The needle has passed the limit, but is not yet to the next tick mark. Is the part over? Obviously.... It worked to explain it to all but the most obtuse production personnel