Holding multiple surfaces parallel within 1 common tolerance zone 2

[SpaciouS]

How do you call-out Parallelism over multiple surfaces that are controlled within 1 common tolerance zone.

Can you simply call-out "x surfaces" under a Parallelism FCF and it's understood that all indicating surfaces are controlled simultaneously (similar to Flatness of multiple surfaces defined by a Profile FCF with such "x surface" notation?

Or would that indicate separate tolerance zones per individual surface?

Or, since I have never seen that example in any ASME standards, is that notation not appropriate when using Parallelism?

Is so, can I simply use Profile instead of Parallelism while referencing a datum and adding "x surfaces" beneath the FCF?

Or is it something else?

Thanks,
Sean

 

[DeanD3W]

Post should work and a couple of mistakes in the file corrected.

Dean

 

[Belanger]

Dean,
Your slides raise some good points, but they are points that exist for those examples. That is not evidence to discount the entire notion of using profile together with toleranced location dimensions!

For slides 1, 2, and 5, yes there is an issue with the sideways location. But I could draw an example where the surface could be located in horizontal and vertical directions with adequate ± tolerances, and there would be no issue. So I think your complaint is really with a dimensioning method which can cause problems for profile. It isn't with profile as an orientation control.

I agree completely with you on slides 3 and 4. But these situations shouldn't happen because they violate paragraph 8.2, which requires profile to be applied to a true profile (basic dimensions for the shape).

So again, I'm not discounting your points. But they are for situations external to the nature of profile, not to profile itself. IOW, we we can't say that profile must always have basic dimensions to tie it back to the datums just because of a few ambiguous scenarios, which seemed to be your underlying thesis.



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

http://www.gdtseminars.com

 

[DeanD3W]

John-Paul,
A true profile is more than just basic dimensions to define the shape of a feature. To complete a true profile basic dimensions to orient and locate the tolerance zone based on any constraint the datum features provide are also needed. You wouldn't say that true position is complete if one basic dimension is deleted from those needed to orient and locate a position tolerance zone would you? Why should true profile be any different?

Y14.5 implying that the absence of a basic dimension for location makes profile a form and orientation control is only done when combined with a size tolerance... Now you're bringing directly toleranced location dimensions into the picture. Directly toleranced location dimensions are ambiguous, since the direction of measurement is not defined and no standard describes a tolerance zone for the practice, so all you really have are two point measurements from one place to another. A combination of profile and an ambiguous location dimension might work on some simple parts in some environments where things are done by known people, but it's not a practice I would want to advise anyone to use.

I'm confused by what you say about the issues pointed out being valid, but not convincing since for other examples you wouldn't see such a problem... Shouldn't the methods of a good tolerancing system work equally well for all eligible applications?

I did expect some disagreement regarding the issues I am pointing out. I understand that the notion I am working to dispel is a firmly entrenched one for many people. We will probably end this cordially agreeing to disagree.

The good tools we thankfully now have are the CF modifier and customized datum reference frames. These two items in our "tool box" enable us to provide desired controls without engaging notions like leaving basic dimensions out of a true profile. If we could just have a modifier for profile to make it truly a general form control, instead of having a fixed geometry (shape) tolerance zone, then we would really be in a better place. Then we can have a form refinement of a non-round or non-flat feature with a profile tolerance zone that really does behave just like a form tolerance (as long as we're careful about simultaneous requirements). I think that modifier would be a good addition to a future version of Y14.5.

Best Regards,
Dean

http://www.d3w-engineering.com

 

[Belanger]

Ah, now I see the issue. I would say it's going beyond the standard to state that a true profile requires "basic dimensions to orient and locate the tolerance zone based on any constraint the datum features provide."

Why? Because of the first sentence of paragraph 8.2: "A profile is an outline of a surface, a shape made up of one or more features, or a two-dimensional element of one or more features." Notice that a profile is at minimum just a shape; there is no mention of any relationship to other things.
And of course, simply adding the word "true" means that we must use profile on a perfect, theoretical shape. (But not necessarily a perfect, theoretical location.)

If we desire to constrain it to datums for orientation or location, then that goes beyond profile's intrinsic purpose (which is fine), and the third sentence says that profile may or may not be related to datums.

I do agree that the all around/all over stuff as well as CF and customized DOF help improve the language for certain situations.

Also, my statement about a point being valid for a specific example but not necessarily for an entire concept troubles you
Well, let's go back to Fig. 8-27. I agreed that your examples had some issues with ambiguity on the location of a surface. But 8-27 does not. That's proof that a statement about one drawing doesn't equate to a blanket statement for the entire concept.

All in good fun, right? What a way to spend a Saturday evening

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

http://www.gdtseminars.com

 

[dingy2]

I have been reading the comments on this and tend to agree with J-P here. Fig. 8-27 would blow a lot of minds since the profile appears to a be refinement of size but this is an anomaly rather than the current practice. True profile does require basic dimensions as J-P has stated.

I also wonder about the statement that Dean mentioned. He stated "Directly toleranced location dimensions are ambiguous, since the direction of measurement is not defined and no standard describes a tolerance zone for the practice, so all you really have are two point measurements from one place to another." I know other individuals with design background have also mentioned the 2 point measurement approach but Fig. 2-5 in 2009 does not support that concept. It does reflect a tolerance zone on a surface. Besides, love to see how someone could measure the surface using 2 points.

Dave D.

http://www.qmsi.ca

 

[DeanD3W]

Dave,
Fig. 2-5 is for Dimension Origin. I thought it would be clear enough that the directly toleranced location dimensions I was referring to as ambiguous were not those using Dimension Origin. Dimension Origin is essentially the same as a profile of a surface, so of course there is a tolerance zone when using dimension origin.

Combining directly toleranced dimensions (which are ambiguous) with profile is not a good practice.

Since there are ways to accomplish the control shown in Fig. 8-27 that don't have the issues associated with them that combining profile with directly toleranced dimensions does, why wouldn't we all just use the methods that are not problematic?

Dean

http://www.d3w-engineering.com

 

[Belanger]

And what again were the issues with Fig. 8-27? (Not another example, but specifically 8-27?)
Combining profile with toleranced dimensions, in that case, has no issues that I can see.

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

http://www.gdtseminars.com

 

[DeanD3W]

John-Paul,
The problem with Fig. 8-27 is that it uses a method that is contrary to a logical tolerancing system. To be logical, the types of geometric control that profile provides would be based upon the fact that it imposes a location constrainable tolerance zone combined with the constraint capabilities of the referenced datum features. If we want to release any of those degrees of freedom, as is desired in the case shown in Fig 8-27, then we can use the customized datum reference frame method described in sections 4.22 and 4.23.

To have one exception that allows a basic dimension to be replaced by even a valid size tolerance (one that is applied to a valid feature of size, as in Fig. 8-27) would be an odd and unnecessary thing to do, within the logical approach that I prefer to stick to.

I hope the slides I posted on April 28th show well enough why the presence or absence of a basic dimension affecting what profile controls would not always be clear.

A even more simple reply to the problem with Fig. 8-27 could be "Why isn't parallelism of line elements on the upper surface of the part specified instead?"... Wouldn't that be a lot simpler? For cases where that upper surface was more complex then profile with a customized datum reference frame could be used.

Of course Fig 8-27 is part of the standard, so the method shown there can be used... I just wouldn't recommend that approach since I think there are better ways to accomplish the same thing.

Best Regards,
Dean

http://www.d3w-engineering.com

 

[pmarc]

Dean, J-P,
Allow me to jump into discussion just for a moment and share a thought or two.

I am rather with J-P on the debate about fig. 8-27, meaning I do not see significant problems with it. Agree, the same goal could be achieved by at least two different methods, like Dean pointed out, but in my opinion that does not mean the third option is vague. As we all know and very often repeat, GD&T is a language, and like in any language there are multiple ways to express the same intent. Some ways fit better to certain situations (applications), while others work fine somewhere else. I think we face such situation here.

Dean, a question to you: as you already emphasized couple times in this thread the availability of customized datum reference frame tool, why couldn't we for example get rid of composite positional tolerancing and use two or more single segment positional callouts with customized datums in lower FCFs instead? Wouldn't it be the equivalent way of expressing the same design intent? Like in composite tolerancing, lower FCFs would only control spacing within the pattern and/or orientation of the pattern relative to referenced datums without having translational DOFs contrained. If you agree they are equal, wouldn't it be reasonable to forbid composite and really help those folks who have serious difficulties in grasping the difference between composite and multiple single segment posistional tolerancing? Wouldn't it simply make Y14.5's GD&T more consistent?

 

[fsincox]

Dean,
The real problem I see, and constantly stress here all the time, is the detachment of the philosophy and direction of the standard from the reality of the real world. Am I the only one who sees this? Are the companies I work with that far behind?
According to this thread:
thread1103-321252 there are still people who shun GD&T altogether.
I agree that the spirit of the standard according to its own “Foreword" is in line with where Dean is going. I just think it has left the real world way behind and becomes in danger of being irrelevant. People here say they avoid it at there own peril, but, actually for the most part they do pretty well and seem more than willing to take the chance. This is if I judge by the actual actions of the companies and not what people here say!
Frank

 

[Belanger]

Dean,
I am shocked, shocked that someone would disagree with me
It seems you're making this harder than it should be, what with customized DOF, etc.
But I challenge a broad assumption that you are making:

"To be logical, the types of geometric control that profile provides would be based upon the fact that it imposes a location constrainable tolerance zone combined with the constraint capabilities of the referenced datum features."

Why should profile be "location constrainable"? Profile is capable of controlling form, orientation, location, size, or a combination of those things. You seem to jump right to location, but form is actually the only one in that list that is a requirement (assuming that there is no T modifier).

Could we take Figure 6-2 and replace it with profile of a surface? I am hearing from you that this would not be logical, since it lacks the location aspect.

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

http://www.gdtseminars.com

 

[DeanD3W]

John-Paul,
I said location "constrainable", not location "constrained".

What you call a broad assumption includes no assumptions..? It is a statement that I believe if free of assumptions.

I would never apply profile of a line to accomplish the the same thing that parallelism of line elements could accomplish, so I think anyone arguing that Fig 8-27 shows a good practice is making things complicated. I would use profile with a customized datum reference frame only if the upper surface of the part were not a planar surface.

Fig. 6-2 shows a figure with the entire upper surface of the part having its orientation and form refined with a parallelism tolerance. That seems fine to me. The size tolerance also applied in the figure seems likely to be OK too. One could question whether the features included with the size tolerance constitute a valid feature of size, since the two surfaces are not fully opposed (one being smaller than the other), but I believe the design engineer needs to decide whether a feature such as this should be considered a feature of size. Somewhere between fully opposed and fully offset, the feature must transition from a feature of size to one that is not a feature of size... The line where that transition occurs needs to be determined by the design engineer, I believe.

Frank,
Your concern about the acceptance of GD&T and whether Y14.5 and other Y14 standards are becoming less relevant as they are developed further is valid for sure. We're all in an odd state where ambiguous specs are often tolerated by industry, the quality of measurement data is often low, many in the academic world view GD&T as "too easy" ("it lacks academic rigor", I've been told) and a drafting or manufacturing subject that isn't worthy of adequate coverage in their mechanical engineering curriculum, while industry views GD&T as "too hard" ("if I put that on the drawing no one will understand it"). There are things we could do to simplify the language a bit, such as reducing or eliminating the need to calculate where boundaries are. We could instead allow direct statement of the location and type of tolerance zone boundaries. We could clarify the orientation of line elements or cross-sections with more explicit notations.

We could also have some sort of certification for metrologists. Having an ASME GDTP certification may help a metrologist, but what about the need to iterate to establish a repeatable datum reference frame with a probing CMM, the need to verify repeatability of data in a standard way, and when and how to properly design fixtures to restrain parts for measurement.

Overall, I see "Mother nature" as the entity that requires the level of complexity in GD&T... Dealing with imperfect geometry in an explicit and complete way is not so easy (as most of those reading this know). I think especially upper level managers will always need convincing with regard to the need for GD&T in order to optimally design and produce physical products. Measuring a part with a pair of calipers seems so simple to them. I haven't encountered an electrical engineering subject which would be viewed as something that can be simplified to the point that someone stepping in off the street should be able to understand it. Our mechanical engineering subject of GD&T suffers from the apparent simplicity of "dimensions", I believe. Some academics and some managers in some industries "get it". When we can say that most "get it" then we will be past our current state of affairs & in a better place, I hope.

Pmarc,
Yes, I think two single segment feature control frames with position or profile could replace composite feature control frames. They're almost equal, except for how simultaneous requirements works (with "SIM REQT" being required to tie together patterns that are toleranced with separate composite feature control frames... no such need with a set of single segment feature control frames). Customized DRFs are better since they more explicitly state what they mean than composite FCFs do, and the method is also more flexible/powerful since even rotational degrees of freedom constraint can be selectively specified using the customized method... I will create a couple of examples and ask in a new topic what others think of setting aside composite in favor of customized. It makes sense to introduce customized drfs before composite would be set aside, but I don't know what is best with regard to the overlap time, when both are available in the standard.

Best Regards,
Dean

http://www.d3w-engineering.com

 

[fsincox]

Dean,
I thank you, and others here, for my contiuning education in this subject. Nice presentation, too.
Frank

 

[Belanger]

Got your point about the location constrainable terminology. Don't know if it affects my basic premise:
If we want a planar surface to be parallel and not constrained with size, sure, we can use parallelism. But if the surface is curved, we can simply swap in the profile of a surface symbol; this is the crux of my argument. There is no need to mess with customized DOF.
Profile controls form, orientation, location, and/or size. To understand which aspects are being controlled, we simply look at the given parameters:
Is the profile "all around"
Does it reference datums?
Is it tied to those datums with basic or toleranced dims?
Answering those and other questions will easily settle the question of which aspect(s) a given profile tolerance is controlling.

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

http://www.gdtseminars.com

 

[fsincox]

J-P,
I know I will have many drawing errors out there if it doesn't break your way. The dilemma I see is drawings must get done day to day and cannot wait 10 more years, for decisions on this stuff. It is shown in books as Dean has mentioned, the company book I wrote has this concept in it. I am not saying that things can't be changed, but, it really puts us, out here in the trenches trying to fight the good fight, in a tough spot.
Frank

 

[SpaciouS]

I think we strayed a bit from the original question.

Q: How to control 'multiple planar' surfaces co-planer to each other and parallel to a referenced datum while not constraining with size. (re-worded a bit)

Are we at least in agreement that the proper way to do this is with a profile FCF, using a datum reference and with notation stating "X SURFACES"...?

However this thread has been quite entertaining

Thanks,
Sean

 

[DeanD3W]

Frank (and J-P),
Frank, I completely agree with your concern. I too am in the trenches of GD&T application and then at least a close observer of subsequent measurement data gathering and reporting.

The issue I thought was worth bringing up here isn't so much a problem when a dealing with a feature of size, as in Fig. 8-27, but we all know that misconceptions regarding that figure are very likely, such as thinking the practice can be extended to mixing profile with directly toleranced dimensions that are applied to a set of surfaces that are not a feature of size... Those specs are ambiguous, so then we're in a mess of having a spec that can be interpreted in more than one way by inspection.

I am all for simplicity as long as we can have it without significant risk of ambiguity... If tolerances are loose enough then maybe some ambiguity, like in one of those instances that someone applies a directly toleranced "location" dimension (excepting dimension origin, so therefore an ambiguous location tolerance), can be OK (call it "bonus uncertainty" ), but for more critical things I think we will all agree that our methods should be sound, with only one "interpretation"/"decoding".

I've had very knowledgeable, well accepted GD&T trainers tell me that profile applied to a cylindrical feature with the basic diameter replaced with a directly toleranced diameter dimension will have a tolerance zone that expands or contracts in diameter, just like cylindricity... That is not correct, but it is an extension of the impression that people take from Fig. 8-27 and a couple of others.

We can all only work to the standard we have, and Y14.5, and especially its 2009 version, is the best GD&T standard, in my opinion. Even then I prefer to choose what I think are the best methods available within Y14.5 and avoid the ones that I have seen cause problems. As the standard advances I hope the problematic methods are reduced and deleted and robust,useful, and, as much as "Mother nature of mechanical parts" allows, simple methods are added.

Best Regards,
Dean

http://www.d3w-engineering.com

 

[DeanD3W]

SpaciousS,
Sorry for another off-track post... I didn't see yours as I was writing that one.

I think profile with a "nX", where n is the number of surfaces would be best.

You will notice that "x SURFACES" no longer appears in Y14.5 as of the 2009 version (unless I'm mistaken). See sections 1.9.5 and 1.9.5.1 in Y14.5-2009. The issue with having "nX" and also "n SURFACES" is that people would tend to believe that their meaning was different, or that each should be used for specific situations, but not in others. There never was a specific difference in their meaning or application, as far as I know.

The only place I recommend the use of the word "SURFACES" is for the "(n SURFACES)" notation I think should be added as reference information (so in parenthesis) with a <CF> modifier.

If you are running into opposition to profile, of if you think it is a more intuitive spec for less educated suppliers, then a single Flatness tolerance with <CF> (n SURFACES) is an alternative to control coplanarity and form.

Dean

http://www.d3w-engineering.com

 

[Belanger]

I too apologize to the OP for my share in derailing things

Dean, I must get in one final response, though...
We both agree that profile of a surface on a cylinder would require a basic diameter, rather than a toleranced diameter. But it's not for the reason that you give. It is because profile must be applied to a "true profile" -- a profile where the shape itself is basic. And Fig. 8-27 does have a "true profile" (a flat surface is implied as zero millimeters of curvature). You're focusing on the fact that it doesn't have a "true location" to the datums. But that is not a problem, and it doesn't correlate to the cylinder example.

OK, now I'll exit stage left...

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

http://www.gdtseminars.com

 

[DeanD3W]

J-P,
As I have already said, Fig 8-27 is part of the standard, so the method shown could be used. Should I have written that in all CAPS?

I've also said several times that it unnecessarily shows an application that leads to misunderstandings and ambiguous applications in other cases. It does not point out that the method shown is only valid if the directly toleranced dimension is a valid size tolerance, and that this method works only with the one simple case shown (the feature could be internal or external, but must be two opposed parallel planar surfaced). Again, as I've said several times, there are better ways to accomplish the control shown and those methods can be extended to other types of features.

Dean

http://www.d3w-engineering.com