Rounding dimensions for part inspection 9

[TommD]

Looking for feedback on rounding during part inspection. I've read some older posts directed to the drafting standards ( which is what this forum is for) but I cannot find much re inspection & rounding. Is a +/- .005" a hard limit with no rounding ( +/- .005 = +/- .005000) or does the next digit get rounded? (+/- .0054 = +/- .005") I hope I've been able to express this clearly.

 

[pmarc]

Per paragraph 2.4 of ASME Y14.5-2009 all limits are absolute:

That is:
12.2 means 12.00..............0
12.0 means 12.00..............0
12.01 menas 12.010............0

 

[pmarc]

I am pretty sure you noticed a typo in the first line after "That is:"
12.2 means 12.20.................0

 

[Enginerd9]

Are you suggesting you have an inspection machine capable of measuring something to a greater precision than the dimensions call for on the print? IE... print says 3 decimal places, but your machine can measure to 4?

If so... what an interesting question. I would assume standard rounding / significant figures practices would apply.

 

[axym]

Hi All,

I agree with pmarc that there is no rounding of dimensions or tolerance values. There is also no rounding of inspection data based on how many digits of precision are in the specification - all significant digits are kept.

Here's an example. Let's say that the specification is 12.200 +/- 0.005. If we have an inspection machine that is accurate to the nearest 0.0001, then the fourth decimal place is used (it's not rounded off). So if the measurement is 12.2053, the feature is out of tolerance.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[KENAT]

The requirement stated on the drawing is as pmarc says and sets absolute limits with no allowance for typical rounding laws so indeed ( +/- .005 = +/- .0050000000...).

Remember that tolerance also needs to allow for the gauge tolerance. So if your inspection set up is only accurate to +-.0005 then you need to make sure the parts measure within (+-.004500000000...).

Posting guidelines faq731-376

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

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

 

[pmarc]

In addition to Evan's comment, I think it is even better visible when a dimension for a feature on a metric drawing (where no trailing zeros are specified) is for example 10+/-1? If inspection measured 8.9 or 10.1, would we say that the feature is ok? I don't think so.

 

[pmarc]

There should be 11.1 not 10.1. Again, typo.

 

[TommD]

We use optical measurement that gives us a 4 place readout. I'm not confident in the repeatability of that last digit but Inspection will write a non-conformance based on it, i.e a part that is 1" + /- .005 will fail at 1.0051. I'm pretty confident a gage R&R will not support that last digit. But doing a gage R&R for each measurement is not likely either.

So there are really two pieces:

1) Do you round inspection results > answer, No.
2) What is the precision of the inspection equipment / process? > ought to be 10x the most significant digit. Reality is that is not always feasible. We probably have 2 or 3x, not including the 4th digit & depending on the operator.. Just having another digit on the gage does not make that real.

 

[Enginerd9]

I would expect to pass / fail a part based partly on manufacturability as well. If the operation used to create that feature holds tolerances to 3 decimal places, why wouldn't a 4 decimal place inspection gadget be rounded? Especially considering gage R&R issues.

Or, use the right tool for the right job? No need for a fancy video system to measure a hole if you have gage pins? Etc.

Experience: accumulated knowledge over time.

Talent: the ability to use experience.

Which is more valuable?

 

[TommD]

Fancy optical? Have you never worn out a gage pin? I have had grinders destroy carbide pins on a single job checking parts.. and do you want to talk about an argument ?? YOU gave me this pin... and no, we do not continue to do business there, but I don't need more arguments in my life either.

Its also a little hard to do Cpk studies on a pass / fail from a pin. There are plenty of reasons to use a measuring device isntead fo a pass /fail.

 

[Belanger]

I agree with the others, but that's all based on the print's numbers. Maybe the OP has a different angle on the question: What exact number to use from the readout of the inspection equipment in order to compare it to the print?

Example: If the print specifies 12.5 ± 0.5, we all know that means a max allowed of 13.00000000000000000. But would a measurement of 13.000000000001 be acceptable?
One could say yes, because you don't really have any business measuring that many digits out, when the print only goes to the tenths place.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[steveapathy]

We require our vendors to measure to one additional decimal place of accuracy when qualifying parts, and it certainly is not rounded. For a plastic molded part with tolerance of ±0.005", we may need to reject parts that measure ±0.0053".

In some applications, half a thou can cause unintended and noticeable consequences.

 

[axym]

J-P,

I would say that the measurement of 13.000000000001 would not be acceptable (assuming that the 1 in the 12th decimal place is significant). This is one of the consequences of the dimensional limits being absolute. We can compare any measurement (no matter how precise) to the dimensional limit, without rounding. If it's out of tolerance, even by a trillionth of an inch, it's out of tolerance.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Belanger]

Evan -- is it acceptable practice to measure something to 12 digits' accuracy when the print only calls out one digit?

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[Enginerd9]

What are we building which requires anything beyond 3-4 decimal places? Let me rephrase... what are we building, which will remain on Earth and not take pictures of stars many, many light-years away?

Experience: accumulated knowledge over time.

Talent: the ability to use experience.

Which is more valuable?

 

[axym]

J-P,

Keep in mind that we're trying to illustrate a principle here, not comment on acceptable practice. I would say that measuring to a far higher level of accuracy than the print calls out would generally be inefficient and unnecessary, but there is nothing fundamentally wrong with doing that.

Here's a question. If we used a measurement device that was accurate to two decimal places and the reading was 13.00, would you consider the feature acceptable?

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Belanger]

Yes, Evan, you're getting to my precise point.
A measurement of 13.00 would be acceptable.
Now, has the part changed from when it was rejected at 13.000000001? No. But the measurement we go with has changed, suddenly making that part pass.

So I was pointing out that the Y14.5 rule about dimensional limits doesn't fully answer the question. The answer has to do with the number we hold up for comparison with the standard's clear-cut requirement.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[3DDave]

If someone can reliably make parts that are .000000000001 oversize, and they do, then I think they are purposely antagonistic because they should also be able to make parts that are in spec.

Once you start accepting parts that are .000000000001 out of spec, it's the first step to accepting parts that are .1 out.

I think what Evan was suggesting was that the meter only reported 2 places, leaving a +/-0.005 potential error in the measurement. Presumably anything from 12.95000000 to 13.049999999 is reported as 13.00 even. It leaves the question, is a part acceptable if the measurement is of sufficiently low quality. I'd say no, but those who are in the business of accepting parts can do as they please.

In any case, the promise made on behalf of 'Y14.5 is that values are not to be rounded and then compared to the limits, e.g. round(12.4) <> 12.

 

[dgallup]

If inspection gave me a measurement of 13.000000000001 I would ask them to repeat it. Fat chance of duplicating that number. We have some gauges that are accurate to sub-micron levels but nothing to a femtometer (assuming mm is the unit of measure). Accepting parts that are right at the limit of a dimension may be OK for a one off part but if we are talking volume production then you know some of the parts will be out of specification. If our QC is finding parts near the high limit I'm going to ask production to adjust their process back toward nominal. So really all this debate over the 12th significant figure is moot. You can't have acceptable process capability and make any parts near the limits.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.