ASME Y14.5.1M-1994 Mathmatical definition - was location double worst or min vs max

[sendithard]

Quick question...

I believe ASME Y14.5.1M 2019 changed profile measurement value to be double the worst as 1994 was deviation in min/max.

My question was in 1994 did this same min/max approach apply to location or was this double the worst?

Thanks

 

[pmarc]

The method B data for profile directly tells users if the actual feature conforms to the drawing specification or not. That's the main purpose of it and in many cases it's all that people care about (especially when the feature is conforming).

If you don't care about the underlying cause for the total runout reading, then I don't understand why you would want to distill the size characteristic from the profile error in the cylindrical feature scenario and compare these two inherently different characteristics with each other.

The blink icon I used wasn't of course to joke about the two 737 crashes. I used it in attempt to soften the consequence of me pointing at the mistake you made in converting the +/- tolerance to profile - something that for sure isn't the key thing in this conversation, therefore deserves a doze of humor.

Also, it's not true that Y14.45 doesn't allow for other method C data for profile than the one called default. When you go to para. 4.5.3.2, you will quickly see that the paragraph starts with "Unless otherwise specified", which indicates that other options that produce non-useless answer are allowed as well.

 

[3DDave]

Method A tells if it conforms. The calculation for Method B requires a measurement that already tells if it conforms and buries that to produce a normalized value that removes the underlying sense of the measurement.

Total runout is a one-sided limit. That reported value represents the full range of the controlled variation.
Profile is a two-sided limit. The reported value can over-report the full range of the controlled variation and doesn't report with respect to each limit.

Ok - I'll fix it - that won't change the point of the comparison which is that a two limit size gets two reported measurements that are representative but the two limit profile has only one that isn't representative. Is that less funny? Both have a single value that was applied to create two limits.

From the draft - "Method C surface deviations for profile of a surface tolerances are gathered for the measured feature. The surface deviation values are used to determine the measured profile value for the feature..."

The explanation seems to leave no room for what Method C is for. "the measured profile value" is a singular value, not one of the two critical values. Nor does 9.3 which clearly states that 9.3.1 and 9.3.2 define Method B and Method C reporting.

The "unless otherwise specified" applies to whether surface deviation is included or not, not whether the calculation is performed - that calculation appears not to be optional. Neither does the requirement to report X,Y,Z as a comment which seems really unhelpful on rotationally enveloping surfaces.

So, a calculation that is meaningless is required; reporting the actual deviation isn't, and pass/fail is allowed, but not depicted - all in a manner that promotes using a CMM.

My only care in this is that engineering has to make a special plan to get the only information that is relevant to the performance of the part. It appears that an engineering design only group to create a separate standard is required to set those requirements.

 

[pmarc]

From the draft - "Method C surface deviations for profile of a surface tolerances are gathered for the measured feature. The surface deviation values are used to determine the measured profile value for the feature..."

The explanation seems to leave no room for what Method C is for. "the measured profile value" is a singular value, not one of the two critical values. Nor does 9.3 which clearly states that 9.3.1 and 9.3.2 define Method B and Method C reporting.

The "unless otherwise specified" applies to whether surface deviation is included or not, not whether the calculation is performed - that calculation appears not to be optional. Neither does the requirement to report X,Y,Z as a comment which seems really unhelpful on rotationally enveloping surfaces.

The "unless otherwise specified" indicates exactly that some other methods than the default surface deviation method for reporting profile measurement C data can be used as needed. In Appendix A, para. A-1(d)(2) even gives an example of method C data for profile of a surface applied to a planar surface, that is not a surface deviation.

Total runout is a one-sided limit. That reported value represents the full range of the controlled variation.
Profile is a two-sided limit. The reported value can over-report the full range of the controlled variation and doesn't report with respect to each limit.

While I agree the reported method B value can over-report the actual variation, the idea behind the method B data is explicitly explained in para. 4.5.2:
"[...] The method B reported value is the particular measured value, or a value calculated based on a measured value, that is used to determine conformance to the tolerance specification."
That's its main purpose and from that perspective the over-reporting makes no harm, because it shall never show a non-conformance where there isn't any.

My only care in this is that engineering has to make a special plan to get the only information that is relevant to the performance of the part. It appears that an engineering design only group to create a separate standard is required to set those requirements.

This, in fact, may be seen as a convenience, because I am quite sure (based on my experience) that engineering teams in different companies will have differing opinions on what "the only information that is relevant to the performance of the part" actually means.

 

[3DDave]

Still unanswered - and the core of the question - what engineering decision is driven by that fictitiously derived value? If you have experience with engineering teams in different companies, surely it has come up. It would definitely have come up for inclusion as the core method in the math standard and that evidence should also exist for Y14.45.

Calculating "g" allows knowing if the value is acceptable. Acceptance and rejection is simply evaluating "<=" or ">", no T+2g math required.

All I need to shut up an accept this is an engineering calculation where a one-sided evaluation of a two-sided tolerance zone makes sense.

I looked at that nonmandatory appendix entry - if engineering depends on the flatness it would be on the drawing as a requirement. However, again, no need for Y14.45 to get that information - if manufacturing is concerned they would add it to the manufacturing plan and give it it's own characteristic identifier. Inspection would not know the intent of asking for flatness on that surface and should not report it as Method C for a different tolerance requirement.

But then I see Mandatory appendix I (isn't that letter typically avoided in engineering? Was it repaired to match nonmandatory appendix B?) about why one doesn't inspect basic dimensions and report them. If that is a necessary appendix the industrial knowledge of Y14.5 must really be bad.

 

[pmarc]

Still unanswered - and the core of the question - what engineering decision is driven by that fictitiously derived value? If you have experience with engineering teams in different companies, surely it has come up. It would definitely have come up for inclusion as the core method in the math standard and that evidence should also exist for Y14.45.

From my experience, the most frequent engineering decisions are the following:
-- If the feature conforms, engineering doesn't care. It is absolutely sufficient for them to see a reported profile value that is less than the value specified on the drawing.
-- If the feature doesn't conform, engineering obviously starts to ask for more data, such as: is the non-conformance on one side or on both sides of the tolerance zone, etc. But even then, it is usually not enough to make a decision. Because, for example, a surface that falls outside the tolerance zone boundary by X in just a single small spot vs. the same surface falling outside the tolerance zone boundary by X almost everywhere are two different scenarios that often require different treatments.

All these additional details simply cannot be presented as a single line in the inspection report. Also, it is physically impossible for Y14.45 to imagine all additional-data-request scenarios that the engineering of the world might have, therefore the room for method C data intentionally left to allow companies to specifically define the data they need instead of forcing them to use a method that a group of around only 20 individuals (even if they all were ultra-qualified design engineers) in the committee arbitrarily chose.

Calculating "g" allows knowing if the value is acceptable. Acceptance and rejection is simply evaluating "<=" or ">", no T+2g math required.

True, however this would not allow for the direct quantitative comparison to the drawing specification, which, again, in many cases is the first and only information really needed.

All I need to shut up an accept this is an engineering calculation where a one-sided evaluation of a two-sided tolerance zone makes sense.

This is not a direct answer to your question (assuming I even understand it correctly), but let me try anyway... What if the actual surface controlled with a regular equal bilateral profile tolerance with respect to a fully constrained DRF fully lies on one side of the true contour? How would you report this in terms of two-sided evaluation?

I looked at that nonmandatory appendix entry - if engineering depends on the flatness it would be on the drawing as a requirement. However, again, no need for Y14.45 to get that information - if manufacturing is concerned they would add it to the manufacturing plan and give it it's own characteristic identifier. Inspection would not know the intent of asking for flatness on that surface and should not report it as Method C for a different tolerance requirement.

The flatness wouldn't have to be on the drawing - engineering might simply want to know how form error contributes to the profile error. Unless you are saying that to achieve this it would be OK to specify |PROF|t|A|B|C| & |FLT|t| on the drawing (where t is the same number). Manufacturing and inspection should play no or little role in defining what engineering needs.

And if by any chance you are picking on the word "nonmandatory", the appendix is nonmandatory exacly becasue of the reason I described at the beginning of this reply.

But then I see Mandatory appendix I (isn't that letter typically avoided in engineering? Was it repaired to match nonmandatory appendix B?) about why one doesn't inspect basic dimensions and report them. If that is a necessary appendix the industrial knowledge of Y14.5 must really be bad.

The letter 'I' follows the same convention as was used in Y14.5-2018 (and probably in other recent Y14 standards, where applicable, as well), so no blame should be put on Y14.45.

If you measure the industrial knowledge by the action of reporting or not reporting basic dimensions, then in the eyes of Y14.45 it is bad. My experience shows this is a frequent issue and I am very happy that it's been finally captured somewhere. However, if the appendix was nonmandatory, I am pretty sure that those who think that basic dimensions should be reported would easily take advantage of this fact.

 

[3DDave]

What if the actual surface controlled with a regular equal bilateral profile tolerance with respect to a fully constrained DRF fully lies on one side of the true contour? How would you report this in terms of two-sided evaluation?

The actual limits to that deviation. What is the maximum material and what is the minimum material - if they are both material minus then, for example on an equally disposed 0.06 tolerance one case might be:

-.01 for material+ compared to the defined material+ 0.03 (0.06/2) limit and
-.02 for material- compared to the defined material- -0.03 (0.06/2) limit

engineering might simply want to know

They might want to know a lot of things. They might want to know how orientation affects it. But every thing I want to know has to do with performance of the feature and will have a tolerance for it. If flatness is a a requirement then it will be a refinement; anything else is passing curiosity and not engineering.

By way of example - a rectangular prism has 12 edges. Each edge is represented by a nominal 90 degree angle. Does inspection ever just go off and add 12 lines for those face pairs to the report because engineering might want to know the angularity between all face pairs?

Still - it would be a separate requirement from the profile to trace where the request for information came from. If it's not on the drawing inspection needs to justify the cost of performing the flatness check and reporting the flatness result so it would be a separate sequence number.

---

I was making clear that if there are people reporting basic dimension variation then Y14.5 is losing the fight and those won't be people paying to get a copy of Y14.45. For them Y14.45 is a checkbox on the CMM software and they will figure out some way to get the values from the CMM onto a page.

I have previously contrasted Y14.5 with programming languages, with the difficulty for Y14.5 being that there aren't compilers or interpreters widely available for Y14.5. Instead it ends up just as bad as if anyone were to try to learn to programming by talking to other students about what a computer might do and believing their program they wrote by hand would work as they expect it to without error. It doesn't matter how much individuals are expected to "study" Y14.5. Once they get it wrong they will tend to remain wrong. This makes the scattershot overused condition where training materials. particularly in the standards, showing easy special cases a real problem.

How tough is "Basic dimensions have no variation so no variation can be reported for basic dimensions. If you feel otherwise, tell your employer you have both made a terrible mistake in your hiring." Put that in the footer of every page.

---

I think the failure in the Y14.5 standard is from using the document as a cheering section, avoiding showing applications where the acceptable result is far from the naive expectation - I failed several times to convince coworkers that for mounting holes the face they were drilled into was sometimes a poor choice for the primary datum reference. The worst was a formed flange - if the flange was bent to an outside face angle that was slightly less than 90 (as springback was likely to do) then the resulting position of the hole with respect to the nominally parallel face would be unwanted.

where the flange had a +/- 1 degree tolerance (though angularity would be as relaxed similarly) with this variation exaggerated in the diagram. The horizontal portion was about 30 inches wide - 0.017 * 30 inches > 0.5 inches vertical displacement vs. the dia 0.030 position tolerance.

The argument would be that the inspectors would use a height gauge from datum feature B to check the basic location and that it was important to have datum feature A as primary to make sure the hole was perpendicular to that face. In the 0.030 sheet that would be (0.017*0.030 = 0.0005 inch lack of perpendicularity for datum feature B as primary.

Does any standard show this potential pathological case, a very common problem, and indicate why it's bad? Of course not - it's a perfectly valid callout to have A primary and B secondary. It's used all over the standards for position tolerance. The standard writers are cheerleaders who will not show their favorite tool being misused in any way.

Lacking some access to the equivalent of a compiler or interpreter that works only off the evaluation rules and can generate geometry that is acceptable to the requirements to show it will accept unwanted parts, it's just a free-for-all.

In my case - manufacturing told management that engineering was stopping production. Management was uninterested in understanding engineering - they got MBAs so they didn't have to - but did understand missing delivery dates to a customer who had penalties for late deliveries. I could go to the standard but that showed, when a datum feature surface was penetrated by a hole, that datum was always primary.

---

It's not a "convention" when the ASME Y14 committee is alone in doing so. "I" doesn't come before "A" and mixing Roman Numerals with Alphabetic characters as heading identifiers isn't conventional.

Per IEEE guidelines:

Appendix headings are a special case.
The primary heading(s) in the Appendix or Appendixes (note
spelling of plural) are set according to the usual style, except that there is flexibility in the enumeration of the
heading. The author may use Roman numerals as heading numbers (Appendix I) or letters (Appendix A).

https://www.ieee.org/content/dam/ieee-org/ieee/web/org/conferences/style_references_manual.pdf

Also - the page numbering was jacked up. Didn't notice until I went to search for page 98 per the table of contents and there isn't a page 98. It looks like the table of contents was frozen but the automatic numbering wasn't when the Foreword, et al, was suppressed.

It is an example though - MS Word is an interpreter for .doc and .docx files. Even if the issuer thought the table of contents would match the sections, the interpreter operates according to the set of interpreter rules and generated an acceptable document even when it's not what would be desired.

 

[pmarc]

3DDave,
I am afraid we will not be able to convince each other (which I take as my loss). I understand and agree with your point about "the difficulty for Y14.5 being that there aren't compilers or interpreters widely available for Y14.5", however for the Y14.45 part of the conversation, I feel like we are starting to run around in circles because our expectations of what the standard should be doing are different.

Perhaps you had already worked this problem out multiple times, but in case if not, I hope Y14.45 will at least serve you as a baseline (good or bad) for creation of the best possible measurement data reporting standard that satisifies all your needs.