secondary datum MMC, tertiary datum RFS

[mel12]

I am not formally trained in areas of design, drawing, GD&T, etc. What I know about GD&T, I learned mostly from reading Y14.5-1994 and looking at drawings. So my understanding is incomplete and probably often incorrect.

I recently came across an old customer print that I don't know how to interpret. The following is a stripped down version of the part. Imagine a hockey puck with a thru-hole parallel to the puck axis and a short cylindrical feature that is parallel to the puck axis and projects from one face of the puck. Neither feature is ON the puck axis. The 'A' datum is derived from one face of the puck. 'B' datum is derived from puck OD. 'C' datum is derived from cylindrical projection's OD. The cylindrical projection is constrained by the true position tolerance |dia .010 |A|B|. The thru hole then has the following true position tolerance applied to it: |dia .010(M) | A | B(M) | C |.
If I understand correctly, as the B datum feature's size deviates from MMC, the location of the axis of the B datum feature will be allowed to deviate from the location of the B datum. But how is it allowed to deviate? Can it only swing around a fixed C datum axis? Is the distance between the B and C datums fixed by the basic dimensions (or perhaps by the distance between the actual mating envelopes of the B & C datum features?), or is the C datum allowed to slide along a line passing through the B datum as the axis of the B datum feature deviates from the B datum axis?

Thanks in advance for any clarification.

 

[PeterStock]

Think of a gauge to check for this tolerance. It would consist of a block with a flat bottom hole in it. The diameter of this hole would be the smallest size that would always allow the A surface to contact the bottom when the puck is placed in it. These represent the A and B datums. The constraint for the C datum is a bit harder as you are using RFS. This would need to allow the puck to move in the hole for datum B (with the A surface always in contact) while centering the cylindrical surface for C. The hole tolerance would be checked with a pin with a diameter equal to the smallest dia of the hole minus .010 for the MMC allowance, no centering needed.

Peter Stockhausen
Senior Design Analyst (Checker)
Infotech Aerospace Services

http://www.infotechpr.net

 

[MartinSr00]

This question could benefit from a drawing, but here's an attempt based on my interpretation.

I believe that the datum feature simulator for B has a size equal to MMC. The datum feature simulator for C is located and oriented with respect to both A and B, and is equal to the largest inscribed diameter in the part minus .010 -- the extent of it's allowable mis-spacing. The C datum cannot translate.

In the 2009 standard, they have a new symbol which will allow translation of the C axis so that it only "clocks" the part. In the 2009 Standard, they have an example like your situation but even more complicated. In the 2009 standard, you can even prescribe the size of C's simulator in your example, by brute force in complex situations. But none of that is present in the 1994 standard. I think they added this for situations just like yours.

Now, I think you're assuming that the datum simulator size for C is equal to the largest inscribed diameter (without deducting the .010). If that were the case, the part would swing perfectly on C's axis as B departed from MMC.

I could be wrong. I'd like to hear what others say.

 

[Belanger]

Peter, to simulate A and B, I think the gauge should be a flat block with a large hole, made at the largest diameter of the puck's OD, since B is reference at MMC. Maybe this is what you are saying, but I just wanted to clarify.

If the actual puck is made at MMC, it will (barely) fit. If the puck is made less than its max OD, then there will now be some looseness. This looseness is the extra amount that will be felt in the position tolerance of the small hole.

To the OP's specific question, the extra looseness around datum feature B could potentially be felt in any direction, but it will depend on where datum feature C is made.

Imagine if you take a puck with the OD at MMC and lay it on the gage. It will be snug (and recall that C is hard to simulate because it's referenced at RFS, but that's a different discussion). But now remove the part, and grind down the OD to its LMC without changing the center axis. When this part is placed back on the gage, datum feature C will not have moved from the puck's axis. So now, the looseness created by a smaller OD is really only going to act in an arc direction that pivots around C. (That's why we have to be careful and say that the MMC modifier on datum feature B does not enlarge the position tolerance, but it does allow it to shift.)

Datum C itself is tied back to A and B with no MMC modifiers. So it is anchored by its basic dimension (and the .010 position tol). Hmmm -- now I'm starting to wonder if C should be secondary...

 

[mel12]

here is a quick sketch. it's not drawn by the book, but i hope it will be easily understood nevertheless.

 

[PaulJackson]

[Q] But now remove the part, and grind down the OD to its LMC without changing the center axis. When this part is placed back on the gage, datum feature C will not have moved from the puck's axis. So now, the looseness created by a smaller OD is really only going to act in an arc direction that pivots around C. (That's why we have to be careful and say that the MMC modifier on datum feature B does not enlarge the position tolerance, but it does allow it to shift.)[/Q]

Belanger, I disagree, is not going to act in an arc direction that pivots around [C]. has mobility in a circular area not an arc and [C] still only constrains the one remaining degree-of-freedom RFS... rotation about the axis .

Paul

 

[Belanger]

Well, I guess that brings up an inherent problem in this type of datum system. Since datum C is tertiary, in a gage it will only be grasped by a simulator consisting of two parallel planes that are centered around datum B. So when we "pinch" the sides of the pin/boss called C, the puck will swivel in an arc (presuming the OD is less than MMC) and also be able to move along an imaginary line connecting B and C.

In this case it just seems weird to allow the secondary datum to shift, but then keep the tertiary datum as RFS.

 

[dingy2]

"In this case it just seems weird to allow the secondary datum to shift, but then keep the tertiary datum as RFS."

Yes, Belanger, it is weird and I, for one, think that the Designer here really doesn't know the reason that the tertiary datum is in RFS while the secondary is in MMC. Although the standard does not specify, it certainly makes more sense that both datums (if they are features of size) should be in MMC or in RFS.

If the secondary and tertiary datums are holes where there are mating features, having both datums in MMC simulates the assembled condition and is a real life situation.

Dave D.

http://www.qmsi.ca

 

[PaulJackson]

Since datum C is tertiary, in a gage it will only be grasped by a simulator consisting of two parallel planes that are centered around datum B. So when we "pinch" the sides of the pin/boss called C, the puck will swivel in an arc (presuming the OD is less than MMC) and also be able to move along an imaginary line connecting B and C..

I agree... [C]'s axis would be permitted to translate nearer and farther from 's MMC center while the datum feature is permitted to translate within its boundary according to its permissible departure from MMC so that features that reference this "odd" datum structure may be verified.

In this case it just seems weird to allow the secondary datum to shift, but then keep the tertiary datum as RFS.

If this "odd" datum structure is exactly how the mating pieces are designed to function then it is correct... if not... then it is not.

Capture the function and detail it commensurately and it has a better chance of working as predicted.

Paul

 

[PeterStock]

Belanger, your first paragraph is a correct rephrasing of my post. However, where I may be in error is in insuring that the primary datum A is in full contact. This would require that the puck be placed, datum A down on a flat plate. Then a part with a hole with a dia equal to the minimum OD of the puck (plus any perpendicularity of the dia to A) would be placed over the part. This part would contain a slot to accept the C datum. With this part over the puck, a pin is pushed up thru the flat plate to pick up and confirm the hole.

Peter Stockhausen
Senior Design Analyst (Checker)
Infotech Aerospace Services

http://www.infotechpr.net

 

[PeterStock]

It might be easier to think of the hole as the datum C and the cylindrical feature on top as the item to be inspected.

Peter Stockhausen
Senior Design Analyst (Checker)
Infotech Aerospace Services

http://www.infotechpr.net

 

[MartinSr00]

Paul

Applied to the facts of your case, I used to think that C only constrains a final degree of rotational freedom, i.e. that it could theoretically float nearer and farther from B radially. I think the 1994 Standard may be a bit vague on this issue.

But in the 2009 Standard, this is explained in some detail in an example. Applied to this case, I think you would reckon the RFS size of C's datum simulator by first considering it's size as if the B datum were not there (oriented only with respect to A), and thereafter subtracting the allowed positional variation between B and C.

Once this was done, you would construct, as Belanger said, a large block with a hole at B's MMC diameter that is perfectly oriented with respect to datum simulator A. Then you would move by the defined basic dimension spacing, and locate and orient your C simulator (a pin) with respect to Datums A and B at the bottom of the hole.

Then you would first make sure that the unrelated mating envelopes for B and C were within their limits of size, and then fit your part on the gauge to see if it fits.

In the 2009 Standard, if you wish to set C free to move back and forth in B's radial direction, then you have to add a funny symbol (a large triangle) in the feature control frame.

In the example in the 2009 Standard, they gave an instance where it could become very ambiguous. In such instances, there's a way to specify the size of the simulator in the feature control frame (in square brackets).

If you didn't do this, imagine if B and C had no modifiers and were hence at RFS. That means the part would be over-consrained in the gauge, because B and C simulators would both have to be at their RFS size and spaced apart by the basic dimension between their centers.

In reading the 2009 Standard, it became clear to me that they wanted to clear up some things in the 1994 standard that were left a bit vague. I think this may be one of them.

 

[dingy2]

PaulJackson:

You said "If this "odd" datum structure is exactly how the mating pieces are designed to function then it is correct... if not... then it is not."

I certainly agree with you on this one but I wonder how many Designers look at datums in this manner.

If a secondary datum is a hole and referenced in MMC, it should reflect that a fastener (or cylindrical feature) must be the mating feature. I just can't think of a situation where a tertiary datum hole should be reflected at RFS when the secondary is in MMC. If the hole in not vital in any respects such as one to lighten the product, should it be used as a tertiary datum anyway? Should this hole be referenced in RFS? Maybe MMC?

Maybe you could help with a positional example where the secondary datum hole is in MMC while the tertiary hole is in RFS. What do you think?

Dave D.

http://www.qmsi.ca

 

[PaulJackson]

Sorry for the long delay I am doing grandpa stuff with the little ones.

In the 2009 Standard, if you wish to set C free to move back and forth in B's radial direction, then you have to add a funny symbol (a large triangle) in the feature control frame.

MartinSr00,
I do not have a copy of the new standard yet but I believe that the triangle symbol "Datum (Feature Simulator) Translation Modifier" permits a surface simulator to translate toward the secondary or tertiary datum feature when it controls only rotation about a primary or secondary axis. That however is not the problem in mel12's example. One cannot fabricate a hard attribute gage for the 0.75 +/-?.?? diameter hole [P|Dia .01(M)|A|B(M)|C] because the tertiary is referenced RFS.

I erred when I said

[C]'s axis would be permitted to translate nearer and farther from 's MMC center

I was thinking that if [C] was permitted to translate in a channel RFS that it would appear as if it was translating nearer and farther from 's MMC center while it actually remained static relative to 's RFS center. The more I thought about the problem the more I am convinced that the horizontal orientation of the axes and [C] needs to be maintained regardless of where is permitted to translate within its MMC boundary.


David, I cannot imagine a functional scenario for this datum feature structure but I will admit that I have been surprised before... if the function and assembly parameters are proprietary then mel12 will have to do his own analysis and critique otherwise he could post them here and let all y'all have at it.

Paul

 

[MartinSr00]

Paul

I used to believe as you -- that the C datum was a radial channel emmanating from B. When you get a copy of the 2009 Standard, I direct you to Figure 4-16 where they calculates maximum material boundaries. (In the 2009 Standard, features of size have maximum material conditions, datums [even some not features of size] have maximum material boundaries). It seems to me that a virtual boundary would be the size of a constant sized simulator. I used to believe that hole C would function very much like the centerplane of a keyslot. I've made more than a few drawings that assumed that this was so.

If you want C to perform as a channel, I believe you have to mobilize it with one of those funny triangles.

When I read the 1994 Standard through (I've done this twice now), I found several areas that were vague. This was one of them. Most of this has been clarified in the 2009 Standard. Now whether this was a "clarification" or a "change" is debatable -- but in reading the 2009 standard there is little functionally that is new. Most of what is new is new ways to add detail or "customize" or add detail to feature control frames for your situation to be more specific. After reading the 2009 Standard through, it occurred to me that the Standards Committee recognized these problems and responded by better organizing and re-forming the standard (in minor ways) to fix the problem).

In 4.16.9, "the translation modifier allows the ... datum feature simulator to translate while maintaining it's orientation to higher precedence datums."

In essence it mobilizes translation leaving orientation alone.

For simulator purposes, the size of C is, I believe the size at maximum material boundary (This seems to be the basic dimension less the locational tolerance from the superior datum).

 

[PaulJackson]

I used to believe as you -- that the C datum was a radial channel emanating from B.

MartinSr00,

I don't know what a "a radial channel emanating from B" is.
[C] is a cylindrical surface that has an axis derived from its orientation constrained actual mating envelope. A plane intersecting that axis... as-well-as the orientation constrained actual mating envelope of establishes the "clocking" rotation about the axis .

The coordinate system is established where the axis of 's orientation constrained actual mating envelope intersects surface [A] and it is clocked as detailed above.

Since is referenced MMC the coordinate system is permitted to translate (without any rotation) in any direction parallel to surface [A] a distance no greater than one half the difference between 's orientation constrained actual mating envelope diameter and 's orientation constrained maximum material boundary.
furthermore any candidate translation must apply simultaneously to all features identically controlled to [A|B(M)|C].

In a CMM program one would orient to [A], move reference to , stop rotation via [C] and turn on the MMC switch for . The integrity of that program would be commensurate with how well it accurately reflects the explanation above in discerning an orientation constrained actual mating envelope for its full depth as opposed to a least squares trend circle about the mid depth waist of each diameter (which when extracted from a thru diameter is typically not so far off).

Paul

 

[MartinSr00]

Yes, you're quite right. When I said "channel" I was thinking of C conceptually like a key slot. But even there, it's the center plane of the key-slot walls oriented to superior datums, that becomes the datum.

With translation, what you're saying is correct.

If I'm reading the 2009 Standard right, without a translation modifier, you would stop neither translation nor rotation allowing C's center to float inside a cylindrical tolerance zone .010 in diameter whose center is correctly translated and oriented with respect to datum B (The oriented axis of B). So, at MMC, B would be constrained tightly, while C would allow some play.

As for a CMM finding the datums correctly, you're also right. A datum feature simulator for datum A is a plane that touches the high surfaces of the planar feature -- not a best-fit midplane.

 

[axym]

This is an interesting question and I'm glad that Mel12 asked it. I've learned a little bit from reading the discussion and thinking it through. Here's my take on it:

Simulator B is a cylindrical cavity oriented at exactly 90 degees to Datum A, sized at the virtual condition of datum feature B. 5.00 plus whatever the perpendicularity tolerance is (which we weren't given). Simulator B is supposed to constrain all of the DOF's that it can, that have not already been constrained by A. These would be two translational DOF's, because A has already taken the two "leveling" rotational DOF's. Simulator B provides a datum axis that acts as the XY origin. There will generally be clearance between datum feature B and its simulator, so we have many possible locations for part relative the XY origin. The last rotational DOF is still open so the part can also "pivot" around datum axis B.

Simulator C is another cylindrical cavity oriented at exactly 90 degrees to Datum A and located at the exact basic distance from datum axis B. Because datum feature C is referenced at RFS (RMB in 2009), the simulator must be variable in size. It must be able to go from datum feature C's virtual condition size of 2.01 to its LMC size of 1.99. Simulator C is supposed to constrain all of the DOF's that it can, that have not already been constrained by A or B. The only one left is the last rotational DOF (clocking around datum axis B). Simulator C needs to contract until it contacts datum feature C and clocks the part. This is where it gets interesting. The part has clearance with Simulator B, but needs stable contact with Simulator C. How does datum feature precedence work in cases like this, where a lower precedence simulator has the ability to completely constrain DOF's that the higher precedence simulator only partially constrained? Does Simulator C have to contract as fully as it can (i.e. 3-point contact), often becoming the pivot axis and relegating B to clocking?

After much thought, I would say that Simulator C does not have to achieve 3-point contact but it is allowed to. I agree with Paul that the "candidate translations" around B can exist in any direction. Clocking on C must not eliminate any of them. For any candidate translation (and resulting XY origin) allowed by B, Simulator C must contract until it fully constrains clocking while keeping the origin constant. This will generally result in 2-point contact between datum feature C and Simulator C. For some candidate translations, Simulator C will be able to contract fully and achieve 3-point contact.

This is all very counter-intuitive, mainly because of the practical difficulties that would be involved in constraining the part in this way. So far none of us have been able to dream up a functional situation that would require the secondary datum feature at MMC and the tertiary at RFS. It seems to be something that is technically legal per Y14.5, but would be a very poor design practice.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[MartinSr00]

Evan

I used to think much the same way you did. I'm not sure the 1994 standard would unequivocally settle the issue. I think the 2009 standard, which is most clarifications with a few additions, settles it.

I used to think that the higher datums "gobbled up" degrees of freedom in order -- in essence taking them out of service. It seems instead that it may claim them, but not prevent the claims of lower precedent datums that don't conflict with it.

With the "gobble up" scenario, C has only a rotational degree of freedom, and it's simulator then is free to translate in a radial direction from B.

In the "only claim them" scenario, the simulator for C is hard-spaced by it's basic dimension from the center of simulator B as modified by its positional tolerance with respect to B and is additionally perfectly perpendicular to A, and perfectly parallel to the axis of B.

This is how I read the 2009 Standard. Taken as a whole, the 2009 standard is largely a clarification of the 1994 standard. I suspect the discussion we're having is largely because the 1994 standard is a bit vague on this issue.

In Evan's post, I think that in checking the part for the "only claim them" scenario, you would move over the correct basic distance from B with it's simulator set for RFS, and then "grow" your simulator to a diameter that would contact the hole such that it could grow no more, allowing the part to rotate a bit as the simulator grows. Then you would "grind down" simulator B to its MMC size.

To invoke the "gobble-up" scenario in the 2009 Standard, I believe you would simply put a translation modifier on datum C.

This post is halfway a question -- because I'm not cocksure that this "otherwise doesn't conflict" is the correct way in either the 1994 or 2009 standard. But it seems that it probably is.