Tertiary Datums and Bolt Circles 5

[Daekar]

Hello all,

I'm a CMM programmer, and at my current place of employment and my last job, I have struggled to find someone who knows the answer to this question. On a circular part with a bolthole circle that has a true position tolerance applied to it, how many datums are required? All the engineers I've spoken to seem to think that one (governing X/Y) is enough, two (one for X/Y and one for Z) is generous, and three (one for X/Y, one for Z, and one for rotation) is overkill to be avoided at all costs. My problem is this: Because of the nature of CMMs, I am forced (as far as I know) to arbitrarily choose one of these holes to set as angularly "perfect" - in essence, make it a tertiary datum - and dimension the rest of the holes off of that hole. Even worse, in parts with more than one bolt-circle, I am faced with the prospect of taking the position of holes from one bolt circle relative to a single hole in another bolt-circle, or doomed to specify several holes as band-aid rotational datums - and try to dimension the entire part this way. It just feels wrong, I'm almost certain that's not the way it's supposed to be, but every other programmer I've met does the exact same thing and never thinks twice about it 'til I bring it up - and the head of our drafting department continually beat around the bush trying not to give me an answer without me realizing he was doing it.

See the attached example drawing for clarification, please ignore missing dimensions, the fact that the diameters of the boltcircles aren't specified, etc... what EXACTLY does this mean? Can those bolt-circle rotate relative to each other? The head of drafting muttered something about "being on centerlines so they can't," but what reference do I use to determine if they have or not if the tabs on the sides aren't datums or aren't present?

 

[KENAT]

In your example there does not appear to be any control over the rotation of the bolt circles.

If rotary control - clocking - is required I believe it should be explicitly stated on the drawing.

See ASME Y14.5M-1994 2.7.3, 4.4.3 & figure 4-6.

Not sure where this "three (one for X/Y, one for Z, and one for rotation) is overkill to be avoided at all costs" mentality comes from. If angular orientation - clocking - is functionally important then it should be explicitly stated.

Showing features in line means nothing - think about it, how inline would they have to be, what is the tolerance on it?

However, I'm answering from a drawing point of view, not inspection so maybe I'm missing something in your question.

Posting guidelines faq731-376

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

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

 

[Daekar]

No, there's nothing different about inspection, only limitations placed on our capability by software. The standards are the same, thank goodness. Thank you for the references to the drawing standard, that told me exactly what I needed to know - that essentially, if there is no clocking datum, the bolt-circle can spin every-which way as long as the holes are in position relative to each other. Well, that's one problem solved - most of our drawings are wrong. And I think the problem created by the CMM software is just that... an unacceptable limitation build into software that was written far after the 1994 standard was adopted. Oh well, I guess I'll complain that we need the new version!

 

[Belanger]

Surprisingly, those two patterns are not allowed to rotate relative to each other (beyond the small position tolerance that each hole is given, of course).

In the 1994 ASME standard, this is spelled out in paragraph 5.3.6.1 (or 7.5.4.1 of the 2009 edition): "Where multiple patterns of features are located relative to common datum features not subject to size tolerances, or to common datum features of size specified on an RFS basis [that's you], they are considered to be a single pattern."

In other words, the 13 holes are to be gaged simultaneously. Well, on a CMM they might not literally be measured simultaneously, but we can't measure the 9 holes and then jiggle the part around differently to measure the 4 holes; I think the clocking of the patterns is established once you hit any 2 of the 13 holes; everything else is rotationally locked from that point on.

To avoid this situation, the designer would have to add some sort of note saying that they are to be gaged separately, although the standard makes it sound like that option is only available if datum A is specified with the "M" modifier (shrug).

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

 

[Daekar]

Yes, I had noticed that caveat about separate gaging only permissible with a datum at MMC when I was reading the standard. I do have a question for you though: you said, "I think the clocking of the patterns in established once you hit any 2 of the 13 holes; everything else is rotationally locked from that point on." May I ask why two are required instead of one? The reason I ask is, I was taught to take hits on one hole, construct a line between the center of the datum diameter and that hole, and rotate my X/Y axes to this line. Is there another way to do it, even if you have an odd number of holes?

 

[Belanger]

Yeah, I guess one hole is enough -- I had forgotten that the center of the part was already established. That's what happens when I try to think late at night...

But I'll stand by the rest of my post

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

 

[pmarc]

Daekar,

Answering to your original question:
"Can those bolt-circle rotate relative to each other?" - in the way as it is specified on your sketch they can't. Because they are referenced to the same datums in the same precedence, they have to be considered as a single pattern of holes. And as John-Paul said the allowable rotation is only within the small position tolerance that each hole is given. I also agree with Kenat that showing features in line does not mean anything and is irrelevant as long as tolerance betweeen these features is not specified.
If you want to have two patterns as separate ones you would have to type SEP REQT beneath both position feature control frames as it is shown for example on fig. 7-54 in Y14.5M-2009 standard.

In terms of constraining rotation of the pattern(s) of holes:
If we assume that outer tabs are not there or are not specified as datum features, picking one of the holes as tertiary datum seems to be the only solution.
The question is why shouldn't we consider one of outer tabs as tertiary datum feature? In my opinion this would be very reasonable choice.

John-Paul also said that if one wants to specify separate requirements:

the designer would have to add some sort of note saying that they are to be gaged separately, although the standard makes it sound like that option is only available if datum A is specified with the "M" modifier (shrug).

Please take a look at figs. 4-40 and 4-41 in 2009 standard. Datum feature A is specified on RMB basis there, so it looks like separate requirements concept doesn't require datum feature specified on MMB basis.

 

[fsincox]

I am curious how the outer (3) lugs are dimensioned and is there any functional requirement of orientation for them relative to what we have?
Frank

 

[Belanger]

Good points, pmarc. It illustrates a minor inconsistency in the new standard. SEP REQT can be used on RMB datum references per Fig 4-41, yet in paragraph 7.5.4.2 it says that when MMB is invoked, SEP REQT is an option. You'd think they would also mention that option in 7.5.4.1 (RMB datums)!


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

 

[HGMorgan]

We would run on CMM as follows:
Align using specified datums -A- & -B-, then clock via one of the outbd holes (@12:00)as -C-.
(This is not so much driven by CMM software. You have to have a complete axis in order to check anything to model/dwg. You have to start somewhere.)
Run all 13 holes per model/dwg.
Run Best fit with a rotation axle @ C/L of -A- & perpendicular to -B-.
This would meet dwg requirements and give the product all the tolerance.

 

[pmarc]

I agree, John-Paul.
Maybe I am missing something, but my understanding of para. 7.5.4.1 is that when all datums in FCF are specified on RMB, separate requirements are should not be considered at all. But fig. 4-41 shows exactly opposite situation.

 

[KENAT]

Oops thanks Belanger & pmarc, I missed the common datum reference which yes ties the two patterns to each other, sorry Daeker.

Posting guidelines faq731-376

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

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

 

[fsincox]

Daekar,
I work with CMM people quite a bit and they all have similar concerns as yours. This issue is covered in the standard because of "the simultaneous requirement" the ISO has chosen to take more of an understanding like you would, namely, if it is not specified it is not implied “the principle of independency”. As far as the number of datums issue you mention this is most likely a result of the evolving nature of the standard. It was acceptable, and still is to some here, to reference only one datum for patterns, most of the guys we all work with "prefer the old ways". The biggest issue with your drawing is how the other features are called out and is there a requirement left implied that should be stated. Plus and minus dimensions are not covered by the security blanket of the “simultaneous requirement”.
Frank

 

[axym]

Daekar,

The group has pretty much covered everything, but here are a couple of additional comments.

All 13 holes must be measured in the same coordinate system, and that coordinate system must be leveled to B and centered on A. There is no tertiary datum feature specified, so the clocking of the coordinate system is not constrained. But the CMM software requires you to constrain the clocking anyway - that's one of the difficulties with inspecting GD&T using CMM's. As HGMorgan described, pick something convenient as a clocking datum. Then have the CMM software float the rotational degree of freedom and "best fit" the pattern, to calculate the optimal clocking and smallest actual value.

Different CMM software has different best fitting and optimization functionality. A simple rotational optimization like the one in this example should be well within the reach of most software. Some are able to properly handle optimization for datum features referenced at MMC, but a lot aren't. Only specialized post-processing software will handle more complex types of datum reference frame optimization correctly.

This is familiar territory - I've been through it before with various different CMM softwares. Get in touch with me if you would like to discuss more details.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Daekar]

The question is why shouldn't we consider one of outer tabs as tertiary datum feature? In my opinion this would be very reasonable choice.

Yes, it would seem so - but the issue is that the drafters where I work have been creating drawings without tertiary datums even when they're technically needed for 40 years, and they're not about to go back and fix them just because I point out they're wrong.

I am curious how the outer (3) lugs are dimensioned and is there any functional requirement of orientation for them relative to what we have?

Well, actually, I did this drawing in about 5 minutes from my imagination - it is not a real part. What I'm gathering is, if there is, as you asked, a functional requirement that the lugs and bolt patterns align, then a tertiary datum and reference to it in the true position feature control frame is necessary. If there is no functional requirement that the pattern clocking correspond to the lugs, then no third datum is required.

We would run on CMM as follows:Align using specified datums -A- & -B-, then clock via one of the outbd holes (@12:00)as -C-.(This is not so much driven by CMM software. You have to have a complete axis in order to check anything to model/dwg. You have to start somewhere.)Run all 13 holes per model/dwg.Run Best fit with a rotation axle @ C/L of -A- & perpendicular to -B-.This would meet dwg requirements and give the product all the tolerance.

All 13 holes must be measured in the same coordinate system, and that coordinate system must be leveled to B and centered on A. There is no tertiary datum feature specified, so the clocking of the coordinate system is not constrained. But the CMM software requires you to constrain the clocking anyway - that's one of the difficulties with inspecting GD&T using CMM's. As HGMorgan described, pick something convenient as a clocking datum. Then have the CMM software float the rotational degree of freedom and "best fit" the pattern, to calculate the optimal clocking and smallest actual value.Different CMM software has different best fitting and optimization functionality. A simple rotational optimization like the one in this example should be well within the reach of most software. Some are able to properly handle optimization for datum features referenced at MMC, but a lot aren't. Only specialized post-processing software will handle more complex types of datum reference frame optimization correctly.This is familiar territory - I've been through it before with various different CMM softwares. Get in touch with me if you would like to discuss more details.

Well these two posts seem to be good confirmation that we have been doing things ALMOST right - the optimization you mention is the final step that we haven't taken. We're using PC-DMIS 3.7 CAD++ (which is a bit old, now) so I'm not sure what kind of options there are - I'll find out tonight. If I have trouble, Evan, I might take you up on your kind offer of assistance.

I can't thank you all enough for your help - hopefully this will be helpful for more people than myself. I'm so glad I found this forum...

 

[pmarc]

Daekar,

I would only like to add a comment to consider:
From theoretical point of view, if we imagine that not only 13 holes but also 3 outer lugs are all positionally toleranced in reference to the same datums, in the same precedence (e.g. like you have |B|A|), then this group of 16 features should be considered as single pattern unless otherwise stated. In my opinion tertiary datum for constraining rotation would not be needed in this case, because the configuration/relationship between the features would have to be exactly as it is shown on a drawing.

 

[Daekar]

pmarc, that is an excellent point! It didn't occur to me that features other than holes could be part of the pattern, but of course - as long as they had true position tolerances with the same datum references, that would make sense. Good heavens... I don't think our CMMs can do that. I found out that it is possible to create a "feature set" of circles or cylinders and apply true position to a hole pattern that way, but I don't think it will work with linear distances, boxes, etc. The unfortunate thing is, even with just holes, it will only give you values for the pattern with datums at RFS. I'm all about educating people, but I don't think I'll tell the drafters about that idea (they'd start using it out of spite), because right now they would never put a true position on something like those lugs. Not unless a customer like Lockheed-Martin required it, anyway.

 

[pmarc]

Daekar, here is a link where you can find a little bit more info about the topic. It also mentions that simultaneous requirements can be very difficult to inspect with CMM's.

http://www.tec-ease.com/gdt-tips-view.php?q=126

As I remember this was discussed on the forum some time ago.

 

[KENAT]

Daekar, the example in the standard of tertiary datum actually uses a notch or keyway or something like that if memory serves.

Posting guidelines faq731-376

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

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

 

[Daekar]

Well, I've been learning and refining my programming for the last few days, and I have run into a little snag, not unrelated to this thread topic, so I thought I'd post it - see the example image with this post. The issue is with the curved slots, for the most part: to which bolt-pattern are they dimensioned? Or are they dimensioned to any bolt pattern at all? Particularly in light of the fact that the two bolt circles are not required to clock together (different datum references), this seems to be unclear. The problem I'm struggling with is, I don't really know how this information could be conveyed. Ironically, this is dimensioned identically to a real part drawing we use, where the actual design function requires that the two bolt circles clock together, and that the slots be within the title-block tolerance (0.5 degrees) of the boltholes they are (supposedly? assumed to be?) dimensioned to.