Co-Datums 2

[snowboard1]

Sorry to re-open a previous thread thread1103-364740 but I wasn't clear if AMontembeault's great question was fully answered and I've run into the same issue.

Copy and paste from that thread:

"Suppose I had a piece of plate with 2 holes in it. Each hole is specified as a datum (in this case, datums B and C, where datum A is the large flat face of the plate).

I have another feature with a positional tolerance with respect to A and a common datum defined by B and C (B-C).

The question we're debating is if B-C defines 1 datum wherein the axis of both B and C must lie (call it a horizontal datum), 1 datum between the axes of B and C at an implied center(call it vertical), or actually 2 datums orthagonal to eachother (controls both the horizontal and vertical datums)."

So, does having a co-datum B-C create controls for horizontal and vertical directions?

Thank you

 

[drawoh]

snowboard1,

Your datums, specified as A, B and C, are the primary, secondary and tertiary datums respectively. Your datum[ ]A provides a flat locating face. Your datum[ ]B hole locates in X and[ ]Y. Your datum[ ]C hole controls rotation. That is how the standard defines it.

--
JHG

 

[snowboard1]

Thanks for the response, JHG. I completely understand the trivial case of setting up three datums you mentioned above as primary, secondary, and tertiary. However, my question is different... I would like to set up the datums with Datum A being the large flat face and Datums B-C working together as a co-datum (without using B and C separately). If B-C are used as a co-datum, does that define a center-line datum between the holes in the horizontal direction AND provide for basis for the vertical direction as well (see attached pic). Thanks

Having some problem with attachment link so including it here as well.

http://files.engineering.com/getfil...3646-4d57-8da9-8bd865eb80df&file=IMG_4176.JPG

 

[drawoh]

snowboard1,

I forget the terminology. It is not co-datums. Your box goes

??? | 0.020 | A | B-C | D

This says that B and C work together as a sing datum. I cannot visualize how two holes would do this. You can do this with a U[ ]shaped part where you want to use the two ends as a datum.

Perhaps datums[ ]B and[ ]C can be fixtured as diamond pins. Definitely, you need a datum[ ]D to control sliding. You don't want people interpreting your datums like this. Perhaps someone can come up with an alternative interpretation!

--
JHG

 

[snowboard1]

JHG,

I suspect you do not need Datum D (since the combination of Datum A and the two pins B, C completely constrain the part) but this is the crux of the question from the thread.

 

[axym]

snowboard1,

Short answer, yes. The B-C reference in your attached drawing will control translation in both the X and Y directions, as well as rotation in the XY plane. In other words, the A|B-C| reference controls all 6 degrees of freedom and no further datum feature references would be necessary (or permitted).

Here are some further comments:
-I find it easier to understand (and explain) degree of freedom constraint in terms of the datum feature simulators (plates, pins, etc.) and their behavior, and not in terms of datums (planes, axes, etc.).
-For the A|B-C| reference, the datum feature simulators are a flat plate for A and two cylindrical pins for B and C. The pins are perfectly perpendicular to the A plate and fixed at the basic spacing X1. Because B and C are referenced "regardless of material boundary" with no (M) or (L) modifier, the pins must expand to establish full contact. Because the reference is B-C and not B|C, neither B nor C takes precedence over the other and the pins must expand at equal rates until full contact is established. When the part is mounted on a fixture with the plate and pins, the part gets fully constrained relative to the fixture. Hopefully this makes intuitive sense.
-Deriving the same result in terms of datums is more challenging (for me, anyway).
-In Y14.5M-1994, the datum from a pattern of holes was described as a datum axis at the "center of the pattern". This caused a lot of confusion, because a single axis would not be able to constrain the third rotational degree of freedom. This led to the common (but incorrect) belief that a secondary pattern only controls X and Y translations.
-In Y14.5-2009, things were clarified somewhat, with datums defined for different types of datum feature. The secondary B-C multiple datum feature (this is the Y14.5 term) would fall under the class of Linear Extruded Shape, with the datum being an "axis and center plane". So the datum for B-C would be a combination of a datum plane passing through the axes of the two pins, with an axis exactly halfway in between.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[TheTick]

I wouldn't do it just because I wouldn't want to spend the rest of my tenure explaining it. So much pain for so little gain.

 

[snowboard1]

axym,

Thanks for the clarification. Exactly what I was looking for.

 

[drawoh]

snowboard1,

You need datum[ ]D. The combined datums[ ]B and[ ]C constrain in X[ ]only. The only interpretation I can come with of your specification is that A and B are both diamond pins that do not constrain in Y. If you want one of the pins to constrain in Y, make this your secondary datum as I noted above. Pin[ ]B is round. Pin[ ]C is diamond. This is a standard and well understood way to use holes as datums. Carefully re-read TheTick's reply, above.

--
JHG

 

[Belanger]

Drawoh -- you don't need datum D. Re-read Evan's reply.
There is no diamond pin involved; that would be the case for B secondary and C tertiary. But this is regarding B-C as a hyphenated single datum. Thus both holes are expanded upon, and everything is controlled north/south/east/west as well as rotationally.

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

 

[3DDave]

I wonder what mechanism actually mates using pins that expand at an identical rate. What happens when one pin can no longer expand? How is the other mechanically decoupled to allow continued to expansion.

Depending on which hole it was that was smaller or less perpendicular, the part would be indexed off either hole. It seems like datum shift due to that shift would be difficult to factor in to tolerance analysis, basically doubling the work required downstream of establishing that reference frame.

 

[jassco]

Hi, snowboard1:

Datum "D" is not needed. I agree with Evan's explanation. B-C defines a plane with an axis which control 3 degrees of freedom. Material boundary applies to B-C, not to B or C. The pins must expand at equal rates until they come to full contacts with their respective mating envelops. By expanding at equal rates, the axis remains fixed. If one specifies material boundaries, they will look like the followings:

1. MMC
B-C Circle M
2. LMC
B-C Circle L

I don't think there are such things as follows:

1. MMC
B circle M - C, or B - C circle M.
2. LMC
B circle L - C, or B - C circle L.

That is why the pins must expand at equal rates until full contact is established as Evan described.

Best regards,

Alex

 

[snowboard1]

Thanks to everyone for the feedback. I was also able to speak with a CMM inspection house today and confirmed that using B-C datum in combination with a plane A, fully defined all the degrees of freedom and that no other datums are required. I gave them a choice of using a structure of Datums A|B|C or Datums A|B-C for inspection of a parts, and, to my surprise, they said that the latter was preferred.

The discussion on the expanding pins was quite interesting. I am used to obtaining parts that are CMM inspected at the vendor or take parts to a CMM house myself to confirm parts meet the print and have rarely had an instance where a complete gauge was required (exception being some simple go / no-go gauges). Makes me wonder if use of a CMM with Regardless of Feature call-outs on the B-C datums is 'no big deal' or am I missing something?

Thanks again for sharing your thoughts.

 

[3DDave]

I'm interested in why they think B-C is better. Is the part assembled with expanding fasteners?

 

[jassco]

Hi,

If one uses A | B | C, one needs to use datum translation (side way triangle) to the tertiary datum. So, A | B | C side way triangle is same as A | B - C.

snowboard1:

If those two holes are same size, then you don't need B-C. You classify them as a pattern (2x) and use a single B for the two holes as a datum.

Best regards,

Alex

 

[axym]

3DDave,

I would agree that referencing B-C at RMB (or a 2X pattern at RMB, which is equivalent) would not represent very many functional situations. Few mechanisms would have mating features that simultaneously expand in this way. How the pins behave if one hole is smaller than the other is not clear and is still being debated. Y14.5M-1994 did not even deal with this issue - multiple-feature patterns had to be referenced (M) or (L) and the RMB reference was not allowed. Y14.5-2009 allows patterns to be referenced RMB, stating that "the datum feature simulators shall expand or contract simultaneously from their MMB to their LMB until the datum feature simulators make maximum possible contact with the extremities of the datum features". Unfortunately, the term "maximum possible contact" is not defined and has led to different interpretations. Some experts think that both pins would stop expanding and there would be residual datum feature shift, while others think that the second pin would keep expanding in order to achieve stability.

snowboard1,

If CMM inspection is in the mix, then even more subtle issues come up. It's true that RMB references on multiple features are no big deal with the CMM. This is because the CMM's calculations usually don't follow the Y14.5 theory and thus do not fully duplicate the result that one would get with physical pins. By default, almost all CMM software establishes a reference frame from the two holes using a "least squares" algorithm that finds the hole centers and puts a best-fit coordinate system through them. This gets a kind of "averaged" result that will generally be close to, but not identical to, the result from a physical fixture. Without all the ambiguity that we had with the behavior of the pins! In most cases, this is "close enough" because the RMB reference doesn't represent the mating condition of the part anyway.



Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[snowboard1]

3DDave: The explanation the machining house / CMM house provided for preferred A|B-C datum instead of A|B|C is that the mis-location tolerance between datums B and C would help to 'split the difference' on locating the center of the part (see attachment

http://files.engineering.com/getfil...ca6e&file=ScreenHunter_45_Sep._01_12.35.jpg).

jassco: part of the motivation for using the A|B-C datum structure was to structure a datum set to be used for locating the rest of the part features and I figured if I can get that with horizontal and vertical datum structure from one interface, combination of datum B and C, it seemed elegant.

axym: I spoke with the CMM house and they confirmed that their software defaults to the "least squares" algorithm but said it was easy within their software to switch to maximum inscribed circle (see attached:

http://files.engineering.com/getfil...e40f1&file=ScreenHunter_44_Sep._01_11.25.jpg)