Effects of tolerances applied to datum features

[pylfrm]

When a datum feature is referenced without use of LMB or MMB modifiers, in some scenarios the tolerances applied to that datum feature can affect how the datum reference frame is derived according to ASME Y14.5-2009. For examples, see Fig. 4-29 (a), Fig. 4-30 (a), and Fig. 4-31 (a).

This seems like a bad thing to me, but I am interested to hear other opinions on the matter. Any insights on possible advantages or disadvantages would be appreciated.


pylfrm

 

[powerhound]

What seems bad about it?

John Acosta, GDTP Senior Level
Manufacturing Engineering Tech

 

[greenimi]

Just to get some traction on this discussion I am re-posting pylfrm's statement from the previous thread:

https://www.eng-tips.com/viewthread.cfm?qid=444710

Consider ASME Y14.5-2009 Fig. 4-29 (a). According to the "Means this" section of the figure and the text of para. 4.16.1, the datum feature simulator progresses from the MMB toward the LMB until it makes maximum contact with the datum feature. Unfortunately "maximum contact" is not well-defined, so in this case I interpret it to mean the datum feature progresses until it can't progress any further.* Let's imagine that occurs after progressing 25% of the way through the tolerance zone, resulting in a radius of 14.95 for the datum feature simulator. Now imagine we change the profile tolerance to reference B as the secondary datum feature. The boundaries are still R14.9 and R15.1, so 25% of the tolerance zone will be on the wrong side of the datum feature. This leaves 75% usable. With different as-produced geometry, the usable portion could be anything from 0% to 100%.

 

[axym]

pylfrm,

The secondary datum feature B is referenced with the implied RMB condition in these examples, and we are told that the simulator progresses from the LMB to the MMB to make maximum contact. I take it that this is what you're referring to.

It is true that the tolerance on the datum feature affects where the simulator starts and stops. The larger the profile tolerance on the datum feature, the larger the range that the simulator progresses through.

Actually, I don't really like this either. The underlying assumption that Y14.5 uses is that datum feature simulators conform to their tolerances. The simulator behavior is defined only in the range of conformance of the datum feature. What happens if we get a part whose secondary datum feature B doesn't conform to its profile tolerance? Or whose primary datum feature A doesn't conform to its size tolerance?

This issue is even more significant for MMB (and LMB) references. The simulator is defined as fixed at the MMB condition of the datum feature. What happens if the datum feature is nonconforming, and violates the MMB condition? Do we just stop, and say that the tolerance is nonconforming because the datum feature is nonconforming? If so, this is extremely limiting.

I believe that this assumption came from the context of hard gages. If the datum feature goes past its MMB, the part won't fit on the gage and so the part fails. If the datum feature goes past its LMB, it's too loose and so the part should fail. But I would say that Y14.5 needs to update this, and allow the scenario of a datum feature being nonconforming.


Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[greenimi]

But I would say that Y14.5 needs to update this, and allow the scenario of a datum feature being nonconforming.

Evan,
I know that the new draft updated a bit the verbiage for their equivalent figures. I do not have the draft in front of me, but I remember seeing some “new wording”/ “new verbiage”.
Do you know if the issue described by you above still exist / still applicable or has been taken care of?

Thank you for your contribution on the forum.

 

[3DDave]

Evan,

If the feature is non-conforming why would inspection continue? The part should be rejected at that point and not continue on the bad foundation.

 

[AndrewTT]

4.11.3 states "MMB, LMB, and RMB conditions may be applied/implied to any datum feature referenced in a FCF". Once they (the authors) make that statement they should include at least one example of MMB/LMB being applied to datum features that are not RFOS. I always assumed that the figures of the Pac-Man shaped parts were in the standard to do just that. If you apply MMB to a surface datum feature then this is what it means. I do not think that they were proposing or advocating a preferred design practice.

 

[drawoh]

pylfrm,

A crude rule of thumb about fixtures is that they should be ten times as accurate as the tolerances they are used to inspect. If my datum FOS has a tolerance of [±]0.02mm, and I am locating holes and profiles to 0.5mm, I don't care about the datum feature's MMC/B. If my datum[ ]FOS is [±]0.8[ ]RMB, we have a nasty fixture problem.

Can you show me your fixture for locating the above sloppily specified and fabricated FOS?

--
JHG

 

[pylfrm]

Although in some cases there is no need to continue inspection after finding a non-conformance, that's certainly not always true. It could be valuable to note the issue and continue on if the goal is to provide feedback for manufacturing.

There's also the question of what to do when the datum features don't have tolerances at all. Sticking with the same example figures, imagine datum features A and B are specified to be as-cast and are only toleranced on a different drawing and part number. How should the datum reference frame be derived in this case?

The obvious solution for Fig. 4-29 (a) and Fig. 4-30 (a) is to not place limits on the progression of the datum feature simulator. For these cases, no disadvantages of this change come to mind, so I wonder why the limits were included in the first place.

Fig. 4-31 (a) is substantially different. Although critical to the procedure described, the idea of "maximum contact" is never explained. Theoretically there could be either zero, one, or two points of contact, so does that mean we get to choose any option with two? I doubt that was the intention. Declaring RMB invalid (or defining it to mean [BSC]) in cases like this is the only robust solution I've thought of.


pylfrm

 

[3DDave]

If the size of a datum feature is important, the dimensions and tolerances governing it should appear on the drawing. Usually this is objected to on various grounds when the feature is created at a lower level (such as castings) or on another drawing (such as for modification drawings.)

The three objections I've been given:
1) It's double dimensioning -> even though the dimensions and tolerances apply only at the level they appear (1.4(o)).
2) It's impossible to be sure the same values appear on two different drawings -> never look at screw thread specs or mating hole patterns.
3) This means the feature will have to be inspected twice; what happens if it fails the second inspection -> that's a problem with execution, not part description.

 

[axym]

3DDave,

I agree with pylfrm that we may want to continue inspection even if a nonconformance is found. I would even go further than that, and say that we should be able to evaluate the conformance (and actual value) of each characteristic regardless of the conformance of the others. This means that, for example, a position tolerance could pass even if one of the datum features is nonconforming. In order to do this, we would have to do as pylfrm suggests and not put limits on the progression of the simulator.

pylfrm,

I agree with your concerns about "maximum contact". This term is used in several places in Y14.5 but not defined or explained. The situation in 4-31 (a) is similar to a rocking secondary datum feature, where maximum contact would generally be at 2 points but other scenarios are possible. I'm curious - what scenario did you envision as having 0 points of contact?

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Belanger]

Evan -- unless I'm missing something here, I think I agree with Dave. Sometimes an individual tolerance can be assessed without regard to others, but with the FCF that I think we're discussing, there is an interrelationship that cannot be ignored -- having to do with higher-order datums.

So if a datum feature doesn't meet its own tolerance, then the "downstream" FCF may (and I emphasize may) already be set up for failure.
Thus, is that downstream FCF really being assessed properly?

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

 

[greenimi]

The “improvement” I was talking about was regarding fig 4-31 with datum feature B (secondary) modified at MMB.

“Datum feature B must remain in contact at a minimum of one point” has been revised to “One or more maximum material extremity shall remain between MMB and LMB”.

 

[pylfrm]

If the size of a datum feature is important, the dimensions and tolerances governing it should appear on the drawing.

I generally agree with this, but we seem to be in the minority based on my experience.

I would even go further than that, and say that we should be able to evaluate the conformance (and actual value) of each characteristic regardless of the conformance of the others.

Do you mean to include MMB and LMB datum feature references as well? If so, how might that be made to work?

The situation in 4-31 (a) is similar to a rocking secondary datum feature, where maximum contact would generally be at 2 points but other scenarios are possible. I'm curious - what scenario did you envision as having 0 points of contact?

I hadn't considered applying the candidate datum set idea to this case, but it sounds somewhat reasonable. Is that how you'd interpret it?

I didn't mean to imply that zero points of contact might be a valid option to establish the datum reference frame, just that it's the only real possibility other than one or two points.

So if a datum feature doesn't meet its own tolerance, then the "downstream" FCF may (and I emphasize may) already be set up for failure.
Thus, is that downstream FCF really being assessed properly?

If the datum reference frame can't even be derived, then the answer is probably "no" according to the standard as written. I'm wondering why the standard would be written this way though. What benefit might the be in requiring tolerance-compliant (or even just toleranced) datum features to derive a datum reference frame?


pylfrm