Phantom line gages as datum "features"?

[Sem_D220]

What would be your reaction to a drawing showing a prismatic feature on a part, comparable to those that are found on V-blocks, a phantom line circle tangent to the 2 angled surfaces of the prism, the circle specified with a basic diameter, and a datum feature symbol associated with the phantom line cylinder?

The datum axis derived from the cylinder adjacent to the prism will be used as one of the datums to established a datum reference frame for the part.

The problem I see is that the datum "feature" to which the datum feature symbol needs to be attached and from which the datum axis needs to be derived is not a feature of the part at all, but an external auxiliary component.

Would you be concerned that the scheme is not supported by ASME Y14.5? If it is supported - where and how? If not, what is the closest supported alternative?

 

[pmarc]

I think the whole concept of datums and datum reference frames as defined by ASME Y14.5-2009 is an unnecessary distraction, so I generally ignore it. All that really matters is the relationship between the actual part and the theoretical geometry, and the constraint of that relationship by contact between datum features and datum feature simulators.

I fully agree with this.

 

[Sem_D220]

greenimi, chez311, thank you for bringing reference to the thread about fig. 4-48.
I think that the idea of the arbitrary origin of the DRF in that figure as described by Belanger and axym in that thread is pretty close to what I described in my post today at 05:18 (with the difference that there are actually 2 "default" options for the origin, as described by Belanger - and I managed to count 3 somehow - don't ask).

I've been aware for a while that there is often more than one valid option for the location of the origin of the datum reference frame, but I thought that the orientation of the DRF planes relative to the part is fairly meaningful for dimensioning because if the inspection department follows the default specification per Y14.5, then it is probably much more convenient when the basic dimensions are given normal/parallel to the planes of the DRF (at the directions of the X/Y/Z axes). But, perhaps someone may debunk this? It would be interesting to learn from others' knowledge and experience.

One thing that might be confusing (or even ambiguous?) about multiple options for the location of a datum reference frame is -which datum feature constrains which rotational degree of freedom. Recently in thread1103-451384 I posted the following image:

The blue is the as-produced part with the bottom right hole touching the MMB boundary and the orange are the datum feature simulators. Datum feature A is a face normal to the holes (either the shown or the hidden side in the view). If one establishes the datum reference frame per the standard specification (or should I say recommendation?), one of the axes of the DRF must coincide with datum axis B. In that case, as pmarc noted in that thread, it is clear that datum feature B only locates the part, and only datum feature C orients it. If some other arbitrary axis is chosen as the origin of the datum reference frame, such as one of the axes of the pattern of the 3 smaller holes produced more or less accurately, then it might appear as if for this specific produced geometry the rotation of the part relative to the DRF is constrained by B-C (together) rather than just by C tertiary. If the axis of the bottom right hole is chosen, than it might appear as datum feature B orients the part! The possibility to choose the DRF origin out of multiple axes led me to a misconception that the "w" rotation DOF is rotation about any axis parallel to any of the axes of the holes (designated Z for that matter). But only when the axis of the DRF is coincident with datum axis B it is clear that it is datum feature C that orients. I am thankful to pmarc for pointing it out for me, and that discussion led me to believe that the location of the DRF origin matters, and should be established per the specification in the standard. Here are some screenshots of that discussion:

The opinions expressed in this thread that the exact location and orientation of a DRF doesn't really matter and the referenced thread about fig. 4-48 leave me somewhat confused, at least for now. I would be thankful to anyone who can address this matter.

 

[pmarc]

In regards to the topic of how the axes of the coordinate system shown on Sem's drawings should be oriented, a figure that could help to some extent is fig. 4-43 in 2009. In that figure, axis X of the coordinate system [A,B,D] doesn't go through the datum feature D - it's parallel to it. If datum target point, and not the entire surface, was used as D, that wouldn't have to change the orientation of the coordinate system [A,B,D]. Of course, it wouldn't be a mistake if the X axis was passing through the datum target point and the entire coordinate system was shown rotated accordingly.

Going along the same lines, I would say that the coordinate system on Sem's drawing could even be shown with X axis horizontal and Y axis vertical in the front view. As a matter of fact, as long as there is a basic relationship between datum targets and the chosen coordinate system, any orientation of the CSYS is acceptable. The winning option is often the one that corresponds the most with how the part is going to be oriented in a higher level assembly.

Regarding location of the origin of the coordinate system, again all different options are possible and allowed. In such cases I like to say that the origin of the CSYS could even be located on the Moon provided that there is a basic location relationship between the Moon and the used datum feature simulators. But will this add any value to a part like a flat plate with 5 thru holes in it? Most likely not, and that's why the most natural and straightforward choice is to place the origin of the CSYS at the center of the feature that per the design intent is responsible for location of the part in an assembly. If someone decides to show the origin at a different place (such as one of the centers of the 4 smaller holes in the considered plate), it is his/her choice, but the risk may be that this will cause a confusion to the readers of the drawing. And for sure this will not change the way the part gets constrained relative to the datum feature simulators, because this is something that is encoded in the datum portion of the feature control frame and not in how a CSYS is shown on the drawing.

Long story short, I think that everything I just said (maybe except for the last sentence) is exactly one of the forms of distraction mentioned by pylfrm in this thread and Evan in the other thread.

 

[Sem_D220]

Thank you pmarc for the additional clarifications.
I have to note that this thread started from a pretty much technical question, and evolved into a very interesting and comprehensive discussion that made me review my whole perception of all the datum related concepts.

I am used to seeing the concepts of datums and datum reference frame as purely theoretical but necessary links that are used to make the transition from the physical datum features + the physical or theoretical datum feature simulators as the "input", to the "output" of the theoretical tolerance zone within which the feature controlled with an FCF referencing the above-mentioned datum features must lie. Along the way, there are rules regarding the establishment of datums and datum reference frames that must be followed as imposed by the standard.

I think the whole concept of datums and datum reference frames as defined by ASME Y14.5-2009 is an unnecessary distraction, so I generally ignore it. All that really matters is the relationship between the actual part and the theoretical geometry, and the constraint of that relationship by contact between datum features and datum feature simulators.

A question to pylfrm and pmarc:
If you were to advise to the ASME Y14.5 committee members how to eliminate the unnecessary distraction related to datums and datum reference frames, would your suggestion be based on:

a. Eliminating datums and datum reference frames from the standard whatsoever, and tying the tolerance zones directly to the datum feature simulators?

b. Keeping the concepts of datums and datum reference frames in the standard, but removing the rules that prescribe specific ways of establishment of datum reference frames, and instead instructing designers and inspectors to set an arbitrary datum reference frame for each part, located and oriented however it is preferred?

If none of the above, then what would be the advice?

 

[aniiben]

Pmarc, Sem_D220, all

Let me ask, what is your understanding of the following simple case:
Does a second or third datum plane (which intersect at the datum axis -fig 4-8 for reference) have to be coincident to the tertiary datum center plane when a tertiary datum feature is added for last DOF (rotation) constrain?
How to make these datum planes coincident?

 

[Sem_D220]

aniiben,
One of the datum reference frame planes intersecting at the axis will be either coincident with the tertiary datum center plane as in fig. 4-15, or offset parallel to it as in fig. 4-32.

Edit: another figure that displays the coincidence case in most detail is 4-6 (explaining this principle for a part initially shown in 4-5).

 

[aniiben]

That's exactly my question: how to make those datums coincident? Not sure I do understand the process.
Is it even possible to make those two datums 100% coincident?

 

[Sem_D220]

aniiben, see fig. 4-32. illustration a. and replace basic 5 with basic 0 (implied). Then at the "means this" portion, the text associated with the center plane of the slot datum feature simulator of the slot would say "Fixed at 0 basic".
Let me know if this answers your question or not.

 

[aniiben]

Well not really. The question is how to align -read make coincident --a datum plane which must stay perfectly perpendicular to the first datum plane to the datum plane driven from the tertiary one?

If the secondary axis must stay perpendicular to the first datum plane so its applicable datum planes created by this axis, am I correct?

Then how to align "perfectly coincident" a datum plane which is fixed in orientation (not location) to something (another plane) which is allowed to be out of orientation (perpendicularity) to the secondary datum axis?

 

[Sem_D220]

aniiben,
In your described scenario of three datum features as in fig. 4-15, the tertiary datum plane must stay perpendicular to the primary datum plane, and parallel to the secondary datum axis. Remember that when we say datum axis/center plane we are talking about the centers of datum feature simulators for which Y14.5-2009 imposes some rules:

"...(b) basic orientation relative to one another for all the datum references in a feature control frame.
(c) basic location relative to other datum feature simulators for all the datum references in a feature control frame, unless a translation modifier or movable datum target symbol is specified..." (Para. 4.5.2 - Requirements)

Be it the secondary or tertiary - both must be basically oriented and located to each other and to the primary.

 

[chez311]

Then how to align "perfectly coincident" a datum plane which is fixed in orientation (not location) to something (another plane) which is allowed to be out of orientation (perpendicularity) to the secondary datum axis?

aniiben,

The gist, I believe, of the above conversation is that location/orientation of the datum planes and origin is really quite arbitrary - what really matters is relationship/constraint of the part to the datum feature simulators. There are some "common sense" or "default" configurations such as that shown on 4-15 (2x datum planes through the axis of B and aligned coincident with centerplane C) or 2x planes coincident with the axis at the center of a symmetric circular bolt pattern, however this is not a requirement set in stone. If as pmarc says the origin was located on the moon, this would not change the relationship of the part/datum features to the datum feature simulators - it would just change the origin of where measurements were taken, which as long this is held consistent between measurements, does not truly matter. I believe this is why pylfrm referred to this whole concept as "a distraction".

If, for whatever reason, you desire a particular location/orientation for the origin/DRF then you can specify it in the manner shown by Y14.5-2009 4-28 or 4-43. This would be the only way to guarantee "100% coincidence" as you stated, which I do not think is nearly as important as you think it is.

 

[pylfrm]

If you were to advise to the ASME Y14.5 committee members how to eliminate the unnecessary distraction related to datums and datum reference frames, would your suggestion be based on:

a. Eliminating datums and datum reference frames from the standard whatsoever, and tying the tolerance zones directly to the datum feature simulators?

Essentially option 'a', although I wouldn't say directly:
[ul]
[li]The tolerance zones are related to the basic geometry of the toleranced features according to definitions in the standard.[/li]
[li]A drawing or CAD model defines the basic geometry of the toleranced features and datum features, as well as the relationship between them.[/li]
[li]The datum feature simulators are related to the basic geometry of the datum features according to definitions in the standard.[/li]
[/ul]


Regarding thread1103-451384 and related issues:

If a degree of freedom is unconstrained until the tertiary datum feature is considered, it seems natural to say that the tertiary datum feature constrains that particular degree of freedom. This does not mean that higher-precedence datum features are not involved though.

Consider the position tolerance applied to the pattern of four holes in ASME Y14.5-2009 Fig. 4-19. If we assume datum feature A constrains [u,v,z], then datum features B and C actually play equal roles in constraining [w].

Although I admit it can be a useful way to think about things, I disagree with pmarc's assertion that [w] represents rotation about a specific axis. If rotation is constrained about any one axis, it is simultaneously constrained about all others parallel to that.


pylfrm

 

[Sem_D220]

pylfrm,
I think I can relate to your choice in option a. If the tolerance zones are defined relative to the basic geometry of the datum features (and therefore also to the datum feature simulators) then there seems to be no need for defining an axis system, other than for either theoretical discussion purposes (labeling the translation/rotation directions - but maybe that too can be avoidable) or for inspection-related purposes (setting an axis system for measurements). Regarding datums - I suppose you can't avoid the use of axes or centerplanes of datum feature simulators for features of size datum features, but instead of calling them "datum axis" and "datum center plane" for the sake of referring to them one could simply call them directly - "center axis/plane of the datum feature simulator". Perhaps less separate-but-related terms would generate less confusion and misconceptions. Do you see it the same way? The only issue could be with cases where a customized datum reference frame needs to be used. What would be the alternative?

So basically it almost(?) seems like a dimensioning and tolerancing standard could do fine without datums and datum reference frames, perhaps it is less so for the mathematical definition standard?

If a degree of freedom is unconstrained until the tertiary datum feature is considered, it seems natural to say that the tertiary datum feature constrains that particular degree of freedom. This does not mean that higher-precedence datum features are not involved though.

Makes sense. Thank you for this clarification.

 

[Sem_D220]

Consider the position tolerance applied to the pattern of four holes in ASME Y14.5-2009 Fig. 4-19. If we assume datum feature A constrains [u,v,z], then datum features B and C actually play equal roles in constraining [w].

Let's talk about fig. 4-9 in order not to be distracted by the translation modifier in 4-19 which I believe is not relevant to the subject of discussion. Considering your above statement, what would be the difference if the position control for the 4 holes referenced A|B-C| rather than |A|B|C|. One difference I can think of is that the fixture would need to be made to allow "sequential gaging" as explained by pmarc in the case of |A|B|C|. For |A|B-C| the datum feature simulators for B and C could engage the part simultaneously. But once all datum features meet the simulators - at the final stage when the inspection of the control begins - is there any difference in how the part is constrained and how the tolerance zones are defined?

 

[pylfrm]

Regarding datums - I suppose you can't avoid the use of axes or centerplanes of datum feature simulators for features of size datum features, but instead of calling them "datum axis" and "datum center plane" for the sake of referring to them one could simply call them directly - "center axis/plane of the datum feature simulator". Perhaps less separate-but-related terms would generate less confusion and misconceptions. Do you see it the same way?

Axes and center planes can be avoided by considering datum feature simulators to be boundaries offset from the corresponding true profile.

I think most of the potential benefit lies in skipping an unnecessary step in the definitions. Simplifying terminology probably wouldn't hurt though.

The only issue could be with cases where a customized datum reference frame needs to be used. What would be the alternative?

I think the customized datum reference frame concept can be interpreted as customization of the relationship between the datum feature simulators and the basic geometry. The notation requires that the orientation of a coordinate system be defined, but "three mutually perpendicular intersecting datum planes" are not really needed.

So basically it almost(?) seems like a dimensioning and tolerancing standard could do fine without datums and datum reference frames, perhaps it is less so for the mathematical definition standard?

I haven't looked recently, but I don't remember anything fundamentally different about ASME Y14.5.1M-1994 in this regard. Ignoring the intermediate step of establishing datums shouldn't change the meaning of the tolerances.

Let's talk about fig. 4-9 in order not to be distracted by the translation modifier in 4-19 which I believe is not relevant to the subject of discussion.

See response in thread1103-451384.


pylfrm

 

[Sem_D220]

pylfrm, that's an interesting vision.

A couple of questions about the following:

Axes and center planes can be avoided by considering datum feature simulators to be boundaries offset from the corresponding true profile.

1. Does a "zero offset" boundary from the actual produced datum feature surface as in the RMB case count as "offset from the corresponding true profile"?

2. Wouldn't the dimensioning still need to be to the axis or center plane of the datum feature simulators, and the tolerance zones of controlled features disposed about them?

Thanks for the explanation regarding A|B-C| vs. |A|B|C| on the other thread. I think I understand now how that may lead to different results.

 

[pylfrm]

Sem_D220,

1. The simulator boundary wound not have a constant zero offset from the actual datum feature due to imperfection of the surface, but the offset between true profile (if defined) and simulator boundary might happen to end up being zero.

2. I just meant datums, not dimensioning or tolerance zones.


pylfrm

 

[Sem_D220]

pylfrm,

1. So essentially boundaries "offset from true profile" serving instead of datum axes and center planes (which immediately made me think MMB/LMB) do not exclude the case of RMB datum feature simulation. Right?

2. I am used to thinking of basic dimensions defining the true profile of part features as either being given directly to datums derived from datum features, or to datum reference frames constructed from those datums (can't really pick one way out if of these two, I guess they are somehow interchangeable or mixed up in my mind). If we skip the step of deriving datums - there should be some alternative to the usually-described origin of basic dimensions which define the true profile. If the alternative to datums is "boundaries offset from true profile" (for example a cylindrical boundary, or a two parallel planes boundary) I'm not sure how to these can directly be the "origin of dimensions". Or do we even need such origin? Perhaps the answer should somehow be understood from the 3 principal points from your post from 23 Apr 19 03:57? Could you clarify?

I feel that even though we are talking here about some imaginary future simplification of the standard specifications, it helps me to understand the current concepts better. So, I appreciate this conversation a lot.

 

[pylfrm]

Sem_D220,

1. Right.

2. Definition of basic geometry can be considered separately from tolerances and everything that goes into their evaluation. Dimensions can be thought of as relating features to other features instead of features to datums. Ignoring the concept of datums does not mean dimensioning practice should change.


pylfrm

 

[Sem_D220]

Dimensions can be thought of as relating features to other features instead of features to datums.

Good point. I can put a check mark on my understanding of how datums can be avoided when dimensioning of a part is discussed.

I would still have a problem trying to avoid datums when describing how tolerance zones are constrained. Let's take for example fig. 4-44 in Y14.5-2009. When describing how the 0.2 tolerance zone cylinder for the hole is constrained relative to the datum feature simulator of the cone, I would normally say that the cylindrical tolerance zone should be coincident at it's center to the datum axis and at a fixed location of 73 from the datum point. If I avoid the term datums ( but not the concept) I would say that it should be coincident to the center axis of the conical datum feature simulator and at a fixed distance from the vertex of that simulator. As mentioned - I avoided the term but not the concept.

To fix that I try to utilize your first point from the post 23 Apr 03:57:

"The tolerance zones are related to the basic geometry of the toleranced features according to definitions in the standard."

But, I am yet to succeed. Is there a way of explaining the fixed location of the tolerance zone for the considered hole without dealing with a datum axis and a datum point (as concepts)?