Reporting Inspection Results

Trust this helps:

Significant figures - Wikipedia, the free encyclopedia

en.wikipedia.org/wiki/Significant_figures - Cached - Similarleading and trailing zeros which are merely placeholders to indicate the scale of
the ... Trailing zeros in a number containing a decimal point are significant. ... as
many decimal places as the measured number with the smallest number of
decimal ... ASTM E29-06b, Standard Practice for Using Significant Digits in Test
Data to ...

Your boss is right.

The 4 is meaningless. Don't report it.

OK, zero means zero, but your boss might still be right, you both have to go to engineering and ask about the positive tolerance manthisa, which in the case of the negative margin, only gets to the third digit.

Without seeing the documentation requirements, it's anybody's guess, but why report insignificant numbers? They add nothing but confusion.

Confusion or noise? _{Just askin'.}

I thought I offered quite a lucid explanation, I don't really have much to add.
unless you'd like a little light sarcasm? *

Imagine there is a tolerancing issue at some later point and they want to apply a new go/no go limit. Or maybe some other part can't be made to tolerance and they have to select specific parts to match with others? I can't see how the extra figure can do any harm, if you don't want it just put your paw over it.

It may be useful later for analysis of machine wear, or even just to prove it's 'noise' and random variation.
Heck those spare digits can come in handy for constructing bankers bonusses
Del

*If an extra digit adds confusion I'd submit the person reading the data should be making the tea instead (other hot beverages are also available).

They must do things differently in England.

If I can measure out to 10 digits and my tolerance is to 3 digits, the other digits are meaningless, and add to the confusion.

To be fair, this isn't my field, and I'm happy to be wrong.

It's my interpretion of 'Report'.
For production or my own use, I'd just work to the tolerance, but for a 'report' I'd simply report exactly what I read, no more no less.

Probably the usual case of language getting in the way of communication.
Del
(I shall go and measure my secret cat nest to the nearest paw width)

I'm sure the OP make will make clear his intentions.

I'm taking my ignorant self out of this one.

Happy Holidays.

The process of inspection is separate from the decision of rejection or acceptance. What you are reporting in the inspection is the full capability of the test equipment.

With out a tolerance listed unless it's dead on it's rejected. What your boss is doing by wanting you to round off is making a implied tolerance. So he can increase productivity. Do as he says but have him write off on it.

I disagree with a lot of the other commenters.

You say the key dimension is " 1.500, and our measurement finds it at 1.5004, that is the number we show on our report. Throughout my extensive training in many different fields including metrology labs, this was a common practice. Applying a tolerace to the above example such as +0/-.001, normally we would reject the part."

NO, you would not. A measured value of 1.5004 is within the +/- 0.001 tolerance on a 1.500 dimension. Your tolerance allows a part as large as 1.5010, so this part 'passes' with 0.0006 to spare.

If, however, your part was toleranced at 1.500 +0/-0.002, then reporting the measured dimension is necessary for the cause for rejection. In this alternative case the part is too large by 0.0004.

In the case you are citing, though, your boss is correct (or, at least, not incorrect) since the part meets the tolerance you cite. There is no need to state the measured size to 4 decimal places if it meets the criterion at 3 decimal places. In some companies a simple 'Pass' criterion is all that is needed in a case like this.

D'oh. I really need to wait til I've had my morning coffee. Just ignore this^.

It depends on wether you want to take attribute data or variables data. With attribute data, each measurement is less powerfull for making a judgement. Varaibles data is much more powerfull to monitor your process. I would use the 4 digit data to determine the standard deviation from print nominal. By using only 3 digits of data, you are essentially doing an attribute inspection with Pass or fail indications. This means your boss wants to see that a unit is perfect or fails with no in between. You know the parts are not all perfect or failing. Of course some of the parts are perfect and some fail, but most are somewhere in the middle of these two extremes. You will know what your process is capable of with actual measurements as opposed to pass fail criteria.

Most of you who've commented have made some very valid points with only one person directing me to a written document from a reliable source. However, it in itself refers to testing results and not dimensional measurements. Other documents I've reviewed are from Y14.5, and ISO 8015. I'm currently have asked people at NIST, and other doctorate degreed individuals in which I'm still awaiting responses. There are many pro's and con's to this but has anyone really nailed it in an accepted document as an accepted practice?

The man who sighs your paycheck should get what he wants.

The comment was meant to be tongue and cheek, but I'll stand by it. You live and die by the print and the process sheet. The customer sets your tolerance, you and everyone in your organization must strive to meet that. Yes, we have SPC to log and show trends. The print will tell you how many points out the customer desires.

If you produce the part within that spec, you've done your job, and it's time to move on to the next piece. Remember, time is money, and the lowest bidder stuff. From Wiki:

Statistical process control (SPC) is a method of quality control which uses statistical methods. SPC is applied in order to monitor and control a process. Monitoring and controlling the process ensures that it operates at its full potential. At its full potential, the process can make as much conforming product as possible with a minimum (if not an elimination) of waste (rework or trash). SPC can be applied to any process where the "conforming product" (product meeting specifications) output can be measured. Key tools used in SPC include control charts; a focus on continuous improvement; and the design of experiments. An example of a process where SPC is applied is manufacturing lines.

Contents

• 1 Overview
• 1.1 Objective analysis of variation
• 1.2 Emphasis on early detection
• 1.3 Increasing rate of production

Manufactured parts vary, and manufacturable designs specify tolerances. Useful measurements are sufficiently precise to show variation within tolerable limits. Suppressing deviations makes measured values indistinguishable from pass-anything, rubber-stamp "measurements." Rounding measurements may move them farther as well as nearer to specifications. Trends in deviations may provide early warning of the development of potential problems. Limiting reporting of deviations to one digit is sensible.

You may have a hard time finding an authoritative reference for these assertions, because they are so fundamental. My guess is that your best bet might be a textbook used in a freshman course in an industrial engineering curriculum.

This is from the horse's mouth:

Code requirements for measurements of significant digits?

The horse's mouth is apparently ASTM E 29. This has become an essentially legal document, incorporated by reference into innumerable contracts and specifications. As such, it can be downloaded for free from legal resource sites on the Web. It raises several questions. My (lay) opinion, after reading it, is (1) if the report in question is a "measurement report," all available precision should be used; so that rounding can be done only once; and (2) if the report is a "conformance to specification report," measurements should be rounded to match the specification, with or without tolerance limits according to the specification.

ASME Y14.5M-1994 2.4 Interpretation of limits

To determine conformance within limits, the measured value is compared directly with the specified value and any deviation outside the specified limiting value signifies non-conformance with the limits.

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

With 1.500 (+0/-.001) most would probably consider 1.5004 to be out - not me.

I feel if .0004 is a problem with fit/form/function then it should have a tolerance of ( +.0000 / -.0010 ).

I would call it good with a 3 place tolerance rounded down to 1.500.

What if it's .000063158 on a CMM? Do you kill it for .0001?

Thanks Halb1t, you understand my dilemma correctly and I couldn't agree with you more. I appreciate your intelligent comments!

It's been my experience that this issue goes away if you stick to the print.

1.500 ( +.000 / -.001 )

With a 3 place nominal and tolerance the .0004 material condition is no more of and issue than .00006 would be.

Our engineering group has questioned my department the same way your currently being questioned and they were right.

We now report print value - nothing more/nothing less.

With 1.500 (+0/-.001) most would probably consider 1.5004 to be out..
No they wouldn't. Nobody would consider it out, because it isn't out!

Why do you believe the 1.5004 measurement? How are you sure that is correct?

Don't tell me you know it is correct because the measurement instrument is calibrated. That info means very little except in the Cal Lab itself. The Cal Lab is a controlled environment. If you were to take that same instrument (properly calibrated) and set it in the bed of a pick-up truck driving down a dirt road, would you still trust the measurements made on it? Your manufacturing or incoming receiving areas are not that bad, but they do influence your results.

Someone earlier asked "confusion or noise?" This is a basic question for any measurement system. How much variation is in your measurement system? To answer that, you need to know how much variation is introduced by the equipment, the operator, the environment, and the procedure. If you can't answer each part of that question, I would assert that the last digit(s) of your reading is(are) unknown.

Look up Gage Repeatability & reproducibility (GR&R). AIAG has an excellent document, "Measurement System Analysis", which will take you through the how and why and even gives you a report form.

I applaud you for wanting to know the answer to your question. No data should ever be taken unless you know exactly what you are going to do with it. Apparently, pass/fail criteria are not well defined. Clearing this up now comes under the heading of continuous process improvement.

Finally, in my opinion, a QC/QA document which has only "pass" or "fail" written on it is useless. I recommend the inspection record clearly identify what is being inspected, what procedure and equipment is used, who is inspecting, what the acceptable range of measurement is, the actual reading taken, then the pass/fail determination.