2 hole pattern as datum 2

[cjccmc]

In attached sketch, I want to have the two .752 holes located +/- .004 dia to each other and perpendicular to datum A .001 dia. Then these two holes become datum B to locate the .252 dia hole. (We use model based definition where CAD file is queried to get dimensions that's why no dim are shown in attachment). Does what I show give me what I need in legal ASME 1994 terms? I was wondering if I need the ref to datum A in the TP frame. It doesn't seem to mean anything but I'ver never seen TP without ref to at least one datum. Thanks for the help.

 

[axym]

pmarc,

Yes, I am saying that datum feature simulator A shall stop at 22.98. Even if it could contract further. The reason that I say this is that referencing A and B as a multiple datum feature A-B treats them as one combined entity, and they lose their "individuality" so to speak. So we don't simulate A regardless of material boundary, and we don't simulate B regardless of material boundary. We simulate the combined A-B entity regardless of material boundary.

I feel that it must be this way, in order to maintain consistency between different datum feature configurations:
-If the datum feature was one continuous cylinder, the simulator would be one continuous expanding/contracting cylindrical sleeve. If the as-produced datum feature was tapered or stepped, the simulator might only contact the datum feature at one end (i.e. instability).
-If the datum feature was two discontinuous cylinders of the same nominal size and tolerance referenced as a multiple datum feature, I would say that the situation must be the same. The simulators should be two coaxial expanding/contracting cylindrical sleeves that expand/contract together. If one of the two datum features was bigger than the other, then only one simulator would contact but they would both stop contracting. If we allow the other simulator to keep contracting independently to achieve stability, then we are not treating the datum features properly as a multiple datum feature. They must be treated as one entity, not as two individual entities.
-If the datum feature was two discontinuous cylinders of different nominal sizes referenced as a multiple datum feature (as in Figure 4-25), I would say that the situation must still be the same. The simulators should be two coaxial expanding/contracting cylindrical sleeves. The A simulator starts at 23.02 and the B simulator starts at 28.02, and they must expand/contract together. Again, if B was produced at the large end of its tolerance (at 27.98) and A was produced at the small end of its tolerance (at 22.92), then only the B simulator would contact.

If the above cases do not all result in the same type of stability (or lack thereof), then I think we have a major problem.

John-Paul,

I hadn't considered the translation modifier in this context at all. This would allow the two datum feature simulators to shift relative to each other, so that their axes are no longer directly in line. I'm not sure that the translation modifier is workable with datum features that control translational degrees of freedom. I'll have to think about this, but my gut feeling is that it only works for lower precedence datum features constraining rotational DOF's.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[pmarc]

Evan,
I looked to Y14.43-2011 hoping to find an aswer to the problem through gage design rules. Unfortunately I did not find any clear statement, but found fig. C-5 (pages 139-140), showing pattern of 4 holes as secondary datum feature at RMB. The picture presenting datum feature simulator pins only says that 4X gage pins expand simultaneously from MMB of datum features until part is immobilized. And this "until part is immobilized" makes me think.

Imagine that actual holes are perfectly positioned, but 3 of them are bigger than the fourth one. Using your approach, 4 gage pins would expand only to the point where maximum contact with the smaller hole was achieved, correct?. But this would in fact mean that the part was free to rotate around that gage pin within certain range, so was not immobilized. IMO the immobilization is possible only if at least two gage pins achieved contact with datum feature holes. But in that case it would mean differences in their size.

Another aspect -- if the gage pins stopped expanding when one of them achieved maximum contact with corresponding actual hole, wouldn't that imply a possibility of some kind of rotational datum feature shift at RMB? Wouldn't that be against one of the fundamental GD&T principles?

 

[dingy2]

pmarc:

Interesting topic. Love to see, in real life, a RMB expanding gauge for 4 holes deemed as secondary datum. I think I know how Evan would set it up on a CMM and I know how I would set it up and it would not include all 4 holes. 2 holes as a secondary datum are OK in RMB but 4 holes - MMB.

Dave D.

http://www.qmsi.ca

 

[axym]

pmarc,

Correct. The 4 gage pins would expand only to the point where maximum contact with the smaller hole was achieved. The part would be free to rotate around the one gage pin and would not be fully immobilized. I would also agree that this rotation could be described as a type of datum feature shift. As far as this being against a fundamental GD&T principle, I'm not so sure. It definitely goes against a common perception that RMB results in stability and full immobilization. But I would say that the geometry does not always give us this result.

Further, we already have rotational datum feature shift at RMB in certain cases. If a single cylindrical datum feature is produced with taper, it can shift rotationally within the fully contracted perfectly cylindrical RMB simulator. So I don't see the problem with allowing the same thing to happen with datum feature patterns (in fact, I see a problem with not allowing this to happen).

If we're trying to put rules and math to this (which we are in Y14.5.1), then the rules need to describe the geometry and behavior of the datum feature simulators. The real-world results of the interactions of the simulators with as-produced geometry (such as datum feature shift and instability) are consequences. If we try to go the other way and write the rules with the real-world results as the starting point, then we run into trouble. This is why associating stability with perfect-form datum feature simulators has not been completely successful. The geometry doesn't always work this way in the real world.

I would say that if stability and full immobilization is the design intent, then this should be designed into the parts. Take the case of a single planar surface specified as a primary datum feature. The designer may not have rocking as part of the design intent, but rocking is a possibility with an as-produced surface interfacing with a perfectly flat simulator. So if this is a concern, the designer has the option of specifying datum targets. Three datum target points will result in a stable datum plane with no rocking. The same should be used with cylindrical datum features and datum feature patterns. If stability is the design intent for a single cylindrical surface specified as a primary datum feature, instability (from taper) is a possibility with an as-produced cylinder interfacing with a perfectly cylindrical simulator. So if this is a concern, the designer can specify datum targets! Two datum target lines or sets of equalizing targets, one near each end of the cylinder, will minimize or eliminate the instability. This, to me, is a better strategy than trying to artificially impose stability on a system that is fundamentally unstable with certain as-produced conditions.

For the four-hole pattern example, I would say that if full stability is required then the designer shouldn't design the part to locate/clock on a pattern of 4 features. Even if RMB is specified, there is still the possibility of instability with certain as-produced conditions such as the one you described. The designer should design the datum features to constrain the required degrees of freedom in a way that does not lead to instablity. Designing one hole as the secondary datum feature and another (slotted) hole as the tertiary datum feature is one possible method. RMB on the 4-hole pattern is not the way.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[axym]

I agree with Dave that the more holes in the RMB pattern, the more impractical a physical simulator setup would be. But I would also say that from a conceptual standpoint, 2 holes as a secondary datum feature pattern is just as bad as 4 holes. The same type of datum feature shift and instability is still possible with 2 holes.

Regarding how CMM setups would be done, another thing to keep in mind is that the rules of datum feature simulation must be method-independent. The different methods must converge on the same result. In other words, a perfect physical gage must get exactly the same result (i.e. relationship between the part and the datum reference frame) as a perfect CMM inspection.

When I see datum feature specifications that result in wildly impractical datum feature simulator configurations, this is a flag that the specification was not based on how the part really interfaces with its mating features. If the mating part does not have 4 simultaneously expanding/contracting pins, then the designer should not reference the 4-hole pattern at RMB.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[pmarc]

Evan,
I have been thinking about your response, and to be honest there may be only one conlusion out of this - since none of us can find any clear statement in any Y14 standard proving his point, we will not convince each other.

If I could just refer to one point of your last reply. In my opinion the situation with what I called rotational datum feature shift at RMB for pattern of holes is not comparable with what you called rotational shift on a single cylindrical datum feature produced with taper. In your example datum feature simulator could not contract further, so the only thing that would left in order to assure part's stability (if this was a noticable problem) would be to use stabilization procedure per the candidate datum set. In my example, 3 out of 4 gage pins could expand to make part really immobilized. This is different, don't you think?

 

[axym]

pmarc,

If this is different (which I agree is debatable), then I would say that it should not be different. The gage pins in your example should not be allowed to expand/contract individually, even if they physically can. My opinion is that because the 4 holes were referenced together as a datum feature, then the simulators must all expand/contract together.

If we allow the simulators for a multiple datum feature to expand/contract independently to achieve stability, then we get certain results that (to me) do not make sense. Here is an example:

Scenario 1: Say we have a 200 mm long cylindrical pin whose OD has a size tolerance of 20 +/- 1. The OD is specified as primary datum feature A at RMB. Its simulator is a continuous cylinder that starts at a size of 21 mm and contracts to envelop the as-produced pin on its high points. If the as-produced cylinder is tapered (larger at one end and smaller at the other), then there will be residual datum feature shift and instability. There would be candidate relationships between the datum feature and simulator, because the simulator must maintain perfect form. It rocks because all of the high points of the as-produced pin happen to be at one end.

Scenario 2: We change the design of the pin so that it has a 2 mm long groove (like an o-ring groove) halfway along the length of the pin. The pin's cylindrical surface is now split into two 99 mm long sections, so we must treat it as two features. We control their sizes with 2X 20 +/- 1 and control their coaxiality with a Position of zero at MMC. The pattern of two OD's is specified as primary datum feature A at RMB. The simulator set is two perfectly coaxial cylinders that start at a size of 21 mm and contract together to envelop the as-produced pin on its high points. If we get an as-produced pin with the same tapered condition as before, should one simulator be allowed to keep contracting to achieve stability? I would say no. This would not make sense to me.

Scenario 3: We keep the split pin design with the groove halfway along. Instead of using the 2X multiplier, we specify the 20 +/-1 size tolerance individually for each feature. We use the same coaxiality tolerance of zero at MMC, by using two leader lines. One OD is labeled A and the other labeled B, and the pattern of two OD features is specified as primary datum feature A-B at RMB. I would say that the configuration and behavior of the simulators is identical to Scenario 2. The A and B simulators must expand/contract together, and we would get instability on a tapered as-produced pin.

Scenario 4: We change the design of the pin to make the two sections slightly different nominal sizes. One section is 21 +/- 1 and the other is 19 +/- 1. We specify the same coaxiality tolerance of zero at MMC, using a Position FCF with the annotation 2 COAXIAL HOLES underneath. One OD is labeled A and the other labeled B, and the pattern of two OD features is specified as primary datum feature A-B at RMB. How should the simulators work now? I would say that the simulator set is two perfectly coaxial cylinders, one starting at a size of 22 mm and the other at 20 mm, that contract together to envelop the as-produced pin on its high points. If we get an as-produced pin in which the A section is produced at 21.99 mm and the B section at 18.01 mm, should the B simulator be allowed to keep contracting to achieve stability? I would say no. The simulator set would be tight on the A section and loose on the B section.

This is all based on the idea that when a pattern of features is referenced together as a multiple datum feature, the features are treated as a combined entity. Therefore the simulators act as a combined entity, and do not adjust to the condition of individual features in the pattern.

I realize that this has gotten very wordy. I'll try to make up figures to illustrate these scenarios. In the meantime, what do you think?

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[TrailMaker004]

I may be taking a left turn off the beaten track here but if MMC were added to the .004 TP requirement on the (2) .752 holes, how would that be interpreted? That is, if the TP .004 effectively tolerances the center to center distance between the holes to +/-.004 (as the OP suggests), then if one hole were at .753 and the second at .751 would we simply add .003 to both sides? (+/-.007??)

 

[pmarc]

Evan,
If drawing depicts splitted cylinder as two separate features, the features shall be treated independent of each other, unless otherwise specified. Consequently sizes of their datum feature simulators shall be independent, even if the features are part of a datum feature pattern.

That said, the key difference between scenario 1 on one hand and scenarios 2-3 on the other (I am intentionally not talking about #4 as it deals with cylinders of different size), is that in 2-3 we have to treat the portions of cylinder as separate entities, because this is what actually the print tells us.

The situation would be totally different if <CF> modifier was used instead of position of zero at MMC callout and "2X" prefix. In that case the datum feature simulator would behave as a single envelope, so we would have exactly the same situation as in scenario 1.

The sad thing is that just one small addition to Y14.5 could easily solve the problem once and for all. If fig. 4-25 in addition to explanatory note: "Datum feature simulator is the smallest pair of coaxial circumscribed cylinders" showed sizes of these cylinders, we would immediatelly know who is right. Now all that is left is to wait another 10-12 years until hopefully the committee dispels the doubt.

 

[axym]

pmarc,

I don't think the fact that we have two separate features necessarily means that the size of their datum feature simulators shall be independent. The orientation and location of their datum feature simulators are not independent - they must have perfect orientation and location relative to each other. The two simulators can only rotate and translate simultaneously as a pattern, so I would say it makes sense that they must also expand/contract simultaneously. The text also specifically mentions this, using the word "simultaneously" in the description of how the simulators contract from their MMB to their LMB. It makes no mention one of the simulators continuing to contract "non-simultaneously" independently of the other. The text also only mentions that maximum possible contact is reached, not that stability is reached. So if we have two different interpretations of what "maximum possible contact" means, then the one that preserves the simultaneous expansion/contraction would seem to be have more evidence to support it.

While I agree with you that the print tells us that the two portions of the cylinder are separate entities, I think there is more to it than that. When features are referenced together as a multiple datum feature, for the purpose of datum feature simulation they are treated as a single feature. This is really the crux of my argument, I think.

I do agree with you that Y14.5 still leaves things open to debate and I'm probably not going to convince you (or the many other people who share your viewpoint). The text leaves some room for interpretation and the figures don't clarify the issue. The configuration in Figure 4-25 shows both simulators in full contact with their datum features, which supports your interpretation but doesn't conflict with mine. On the other hand, the previous figure for the MMB case also shows the simulators in full contact with both datum features, which would not generally occur (there would usually be clearance in the MMB case). So we can't always look to the figures for clarification.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca