ISO GD&T vs. ASME GD&T 8

[pmarc]

What do you think about discussing more in details about differences between ISO and ASME GD&T? What are the most important discrepancies between the two in your opinion?

 

[KENAT]

I possess the 2768 standard thanks CheckerHater, both parts 1 & 2 and spent plenty of time looking at them in the past, just not going to bother now.

Next you'll be saying the bit ninja paraphrases is only in there to allow deviations/waivers etc.

Plus 2768 gets misused the same way block tols and the like do - though arguably that's not the fault of the standard itself.

Almost nothing is being said above that hasn't been said before in earlier threads on the topic.

Posting guidelines faq731-376

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

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

 

[dgallup]

The current revision of the ISO surface finish standard is totally FUBAR. I don't have it in front of me so I may not have this exactly right, but it says something like "only 86% of the surface has to meet the requirement"! That is so stupid. If I'm trying to specify a surface for an o-ring to seal on and only 86% of the surface meats the requirement it is going to leak like a sieve. 100% of the surface has to meet the requirement.

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.

 

[MechNorth]

So, JP, are you saying that you can have a surface roughness callout (i.e. tolerance) that is greater than the flatness tolerance on the surface? In otherwords, 0.1mm on flatness and 0.5mm on roughness? I don't think so. I understand that the two controls are generally orders of magnitude different, but the extreme is possible in some cases (cast surface for example). I'm not saying that it can't pass roughness and fail flatness; it's not uncommon for a lower-order control to pass when a higher-order control fails. Specifically what I mean is that the control tolerance must be a refinement of any higher-order control tolerance.

PMarc, the citation from 2768 is a good example of ISO double-speak. ISO (1101?) indicates that size and form must be controlled separately (as a default method), yet here they are implying that size also controls orientation? Orientation wrt what? ISO 2768's not valid for orientation control because it doesn't require the definition of a datum reference frame. I have seen people work around this by putting a generic note such as "GENERAL TOLERANCES WRT DRF /A/B/C" In 2768, does ISO treat parallelism as a localized thickness control? I've never been able to have an ISO-trained person answer that question and support a datum reference requirement per elsewhere in ISO GD&T standards.

I retract my earlier statement about ISO controls being cumulative. Per ISO 1101:2004(E), Section 17 Interrelationship of geometric tolerances; "For functional reasons, one or more characteristics can be toleranced to define the geometrical deviations of a feature. Certain types of tolerances, which limit the geometrical deviations of a feature, can also limit other types of deviations for the same feature. ... Orientation tolerances of a feature control orientation and form deviations of this feature and not vice-versa. Form tolerances of a feature control only form deviations of this feature." Looking at the '83 version, this was covered in less detail in 13.2. Interestingly in '83, figures 36, 37, 44 & 45 show parallelism to a surface without a datum reference, but rather with a datum triangle anchored to the "reference' surface.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[MechNorth]

Fully agree that engineers/designers/draftspeoples shouldn't abdicate their responsibility for clear documentation by invoking ISO 2768 on the drawing. It increases the risk of improper interpretation when the machinist, inspector read the drawing.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[Belanger]

Gotcha, Jim. I might say that it's pretty much a moot point to even compare surface finish and flatness, because their intents are totally different (thus the order-of-magnitude difference).

But I now agree that if someone were to create a roughness callout of 0.5 microns (with the cutoff width being the same as the full length of the surface) and a flatness tolerance of 0.0000006 mm, then we have a problem (both in terms of mathematics and practicality).

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

 

[fsincox]

I agree the no-rule #1 is a bigger difference than people realize at a casual glance. Remember, you can still invoke it, though. As most here know I am not a big believer in the problems created by the definition and the associated restrictions created to define a "FOS".

The ASME standard seems to be backing farther and farther away from the certainty it seemed to provide in 1982. I believe this is good as it was a myth anyway. Many tolerances no longer cover a FOS anyway but an AME, and we talk about them defining by abstract concepts. We do not use a lot of gauges here at my company all CMMs.

As far as ISO 2768-1's paragraph I believe that is a bunch of fuss about nothing, we accept parts out of spec every day, all this says is we can and will if it occurs on features covered ONLY by the general tolerances, it does not use "fubar" that is an incorrect quote added to put something in a bad light.

I really wish some of these ISO people would come on here and defend it themselves I supose we would need to be able to read German.

As far as the thickness of the ASME book well I have seen discussions here wanting to through out examples that have been in the book since 1973. After everyone here throws out the parts they don't like, the ASME standard might not be as thick and may even look like the ISO standard just showing general concepts and letting it go at that.
Frank

 

[CheckerHater]

Sorry Jim,

“ISO 2768's not valid for orientation control because it doesn't require the definition of a datum reference frame”
Wrong once again. Parallelism requires one and only one datum – have you ever heard of being “parallel to A, B, and C”? And for this datum 2768 picks the largest of two features – smart and simple.
So, yes indeed, THERE IS datum reference frame – just enough of it.

Note that 2768 is carefully avoiding controls that require more than 1 datum – position, profile, etc. One should agree that 2768 is too smart for what essentially is glorified tolerance sticker.
And “double-speak” is a harsh word. One day your company may have more ISO clients than ANSI/ASME clients. Just a thought.

 

[fsincox]

Jim,
As I recall it does attempt to define how to apply the form type tolerances. if it is not perfect well niether is ASME I reference the disagreements here on Fig 6-18 or 6-52 and as far the comments on taloring a tolerance to an inspection method, who can look at runout, Yes I know it doesnot have to be checked that way.

Checker Hater,
I like you more and more. I have a feeling we have had a similar experience somewhere, First, I want to say, I learned a lot from checkers and consider some were my friends. However, when I was trying to convert our system to GD&T (1982 standard) they were the ones most against change, as they liked the old way, it made them powerful. "my" GD&T threatened all that. The entrenched buracracy won that battle, company went out of buisness soon after I left.

Frank

 

[fsincox]

Sorry, I may have technically misspoke, they filed for bankruptcy, closed and are now reopened as a division of a Chinese Company (new owners). I believe that is more correct.
Frank

 

[MechNorth]

Interesting points on 2768. Quite true about parallelism also, but what of perpendicularity? 2768 covers it wrt one surface only. That's adequate for most uses. There is no datum reference frame, but there is a reference surface invoked by the leaders of the callout in those cases. And angularity? So 2768 covers the comparatively easy items. I don't have an issue, in reality, with 2768; I've used it as a design tool many times and it is excellent for general purpose design.
Overall, what I've seen is a cleaning-up of many of the outstanding issues in ISO's GD&T standards collection. That's a sentiment shared by many in the ISO user community as well. And, I've used stronger language in comments about ASME's GD&T standards.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[KENAT]

Paragraph 6/7 & Annex A4 is pretty clear in its wording.

The out come being that the drawing no longer defines what is an acceptable part.

Where on this am I mistaken?

Posting guidelines faq731-376

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

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

 

[MechNorth]

No, I read it the same way. Double-speak?

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[KENAT]

Maybe,

ASME put's the drawing originator in the driving seat. If a vendor or whoever wants me to accept a part that doesn't explicitly meet drawing then I get to decide if I want to spend the time and effort ($) looking into the defects and how they affect function (above and beyond that already defined on the drawing) and if they can be recovered... They can't force me to do so.

However, the ISO 2768 way, the drawing originator has to justify why it can't be recovered. I have to justify why I won't accept parts that don't meet my stated requirements. Whether I want to or not I have to spend the time to justify not taking them. Am I meant to have some other document besides the drawing that captures this? In the end who decides if my justification is adequate?

The ISO 2768 sure makes it easier to ask forgiveness than ask permission - forgiveness is the default state and not forgiving has to be justified. Though I suppose in a way that almost makes it Christ like.

Posting guidelines faq731-376

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

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

 

[MechNorth]

My suspicion is that what Kenat brought up from 2768 was developed as a catch-all for a poorly-documented design. I.e. If it works, use it unless you have a really good reason not to. ASME does tend to favor the designer over the manufacturer in this way.
On the plus side for ISO, I think that they're way ahead of some of the metrology and math-definition aspects compared with ASME.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[fsincox]

Ken,
Sorry,

"Workpieces may be whatever shape the manufacturer decides, the designer has no say in the matter unless it's totally fubar"

I believe you, and maybe some others around, may feel that this statement captures, in your opinion, the essince of the referenced statement, but, it is NOT a "TECHNICALLY" corect ISO statement. Any representation as such is incorrect. I would expect more persnicityness (is that the word I want?) from a real technical discussion, particularly from a checker (sorry, I had to say it).

I feel it my duty to any of the "uninformed" here to call it out as such. Understand, I believe the ISO standard is not perfect either, but I will say the more I learn the more I like it.
Frank

 

[fsincox]

The ISO standard is not stopping you from specifying any tolerance you wish and if you have a valid reason for the tolerance it supports that it is still not an acceptable part, is that really too much to ask. The truth be told we should have a reason for a tolerance or it just should not be, it is the way it is, folks, isn't that our job?
The ISO standard just recognizes that too many GENERAL tolerances are not based on practical absolute requirements. The basic supposition that they spell out is that it is not practical to control every dimension to the absolute extremes when process will GENERALLY hold them much tighter, did you read that part too? If you make millions of the same thing it may be worth the time to check, but in many cases it is not justified. Why don’t your inspectors check EVERY point of EVERY feature? I suppose there are some that actually think they do?
So if I am not going to check every single dimension to it's absolute extremes, it is only a rational buisness decision to say don't throw the part away unless it will not work. I believe this is a rational position to take give the premise they have laid out
There are some here that will say "they" have checked all of theirs, well good for you, yours are fine then. I am here to tell you most do not. I have been around too long and worked at too many places to believe it. I have even seen some admit it here.
As we have stated here, before, in this area the ISO is covering areas not covered by ASME and when they have attempted they just basically adopted the ISO's work (ref: ANSI B4.3).
Frank

 

[MechNorth]

I am guessing, but 2768 is likely based on an old DIN standard; the Germans did amazing work documenting "norms" for manufacturing processes. Other ISO-contributory countries would have verified the applicability of the general tolerances. From my understanding, the information is based on typical machine shop capabilities at the time of publication; 1989 being the release date of the copy I have, which supersedes the '73 release. Given the different tolerance classes, the selection thereof by the designer SHOULD be based on their own shop's or suppliers' capabilities. The problem is, I haven't seen too many companies that have the slightest idea of what those actual capabilities are (Europe, N.America, Japanese), and most of the 2768 users (designers/engineers) that I've encountered just select a tolerance class based on "desired" outcomes and go from there. In addition to that, even fewer shops understand the 2768 process (at least in N.America) and just "do their best". To me, that makes 2768 an inappropriate choice for those users. I have seen it used correctly, though surprisingly it was in N.America rather than in Europe. Note that this is not a criticism if the standard, it's a commentary on the state of engineering in business today.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[pmarc]

As the initiator of the thread, I'd like to say that although I was rather thinking the discussion will focus more on differences between ISO and ASME GD&T, I'm really glad it went into ISO 2768 direction - after all "general tolerances" concept is nicely showing how ASME's and ISO's approaches to dimensioning and tolerancing differ in general.

I believe the approach difference is a key to everything here. And I wouldn't even try to judge which one is correct and which is not. In fact, in my opinion, there are strong and weak points in each of them. And what we shouldn't do for sure is to arbitrarily assume that ISO standards (in this case 2768) are bunch of useless junk without any logic behind - they have been created by people with tremendous amount of experience and practical knowledge.

Here are some of my other thoughts:
I still do believe that ISO standards in general are oriented for 'functional design' even if some paragraphs may sound otherwise. The point is that ISO additionally pays a lot of attention to how the part is manufactured and even more to how it should be inspected. But is it wrong? I wouldn't say so. For me the functionality is not only a question whether something will assemble with something else. It is a matter of manufacturing abilities and economy too. One can have perfectly 'functional' assembly but the effort put on manufacturing and on the inspection will generate such a high price that at the end nobody will think of this product well. As usual, the compromise has to be found, and this is one of the purposes of 2768 standard - it is trying to tell that with this kind of tolerances a design can be economically optimized, adjusted to manufacturing and inspection abilities of a workshop.

But unfortunately this is a noble theory, nicely described in the annex A of both parts of 2768. The reality shows that 'general tolerances' concept is so general and broad that it had to be boiled down to some relatively simple cases and even with that it gets quite unclear. Example: take the general symmetry tolerance for instance - ISO 2768-2 is saying that: " The longer of the two features shall be taken as the datum; if the features are of equal nominal length, either may be taken as the datum." That is clear, but what if there were 3 features of different length shown nominally symmetrical? Which one should be chosen as the datum? Figure (b) in annex B5 is showing such a situation but not explaining how the third feature should be controlled. Should the two shorter features be controlled wrt the longest one or should there be two different datums for two positional callouts?

This was an example only - all I am just trying to say is that the standard is not perfect and leaves plenty room for assumption, but that is life. For some of us it can be useful, for some can't. I saw drawings of really complex parts with almost no geometrical tolerances shown but stating that general tolerances are according to ISO 2768. You would not even imagine how much time a product engineer spent on explaining to inspector which geometrical characteristics should be measured relative to which features. They finally found an agreement but only because they were working in the same plant and were speaking the same language, so they sat together with their bosses and clarified all the issues. Can you imagine what would happen if one was speaking American English, the second one Chinese English and they were located thousands miles apart?

And just for the end, despite of all shortcomings of ISO standards (not only 2768), I agree that they have one big advantage over Y14.5 and associated documents - they are way ahead in mathematical definition of inspection routines for geometrical tolerances. I fully respect and admire Y14.5 for its versatility in describing so different dimensioning and tolerancing schemes within one document, but the fact that those concepts are not fully supported by mathematical definitions is one of its biggest weaknesses.

 

[MechNorth]

Pmarc, check out Y14.5.1 for mathematical definitions of Y14.5. I believe that the sub-committee is working on the next revision now.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[pmarc]

Jim, I am well aware of Y14.5.1, but in my opinion 1994 revision is not doing a proper job. Very small percentage of concepts presented in Y14.5 is thoroughly covered by Y14.5.1. I hope the newest edition will be much more specific.