What is the ô"tertiary datum problem"?

The datum references should reflect the mating part.

If the mating part has a pin and a diamond pin, then one hole diameter will be identified as a datum and the width of the other hole will be identified as a datum.

If the mating part has two pins, then the holes should be identified as a compound datum or a single datum of two holes, not as one secondary and another tertiary.

It is likely the change was a correction.

One of the main failings in the presentation of Y14.5 has been the lack of acknowledgement that the reason for describing acceptable variations starts with parts mating with other parts. in particular, that datum features represent the condition of the idealized mating part surfaces.

The problem with Y14.43 is that is offers the chance for a divergent opinion from Y14.5, which is not helpful. It might have better been a treatise on a variety of ways to examine product acceptance to guide engineering, QA, QC, and Manufacturing to choose controls for product development, rather than a formal standard.

gabimot,
This may interest you (post from 24 Apr 08 8:08):

Thanks for the link.

A search for "tertiary datum problem" turned up the name Herb Voelcker. Talk about providing extensive solutions to problems people don't have and making more problems at the same time.
Check:

To reduce it to its core, the problem is what to do when five of the six degrees of freedom have been taken by the first two datum references and the third datum reference can constrain more than one degree of freedom.

In the case of a plate with a hole, a primary datum plane constrains three degrees of freedom, and the pin takes two. A second pin would also constrain two degrees of freedom, which is one too many. Making the two holes compound causes the mating pins to control only three degrees; using only the width of one hole as the third reference constrains only one degree of freedom.

Adding the translation modifier a la Voelcker, eliminates one of the degrees of freedom from the hole, but leaves me with the question - what actual part/assembly reasonably behaves in a way that is represented by translation?

Dave,
I agree, we can look at the problem from the perspective of DOFs overconstraining.

However, there is another aspect of the "tertiary datum problem" mentioned by Mr. Voelcker, and also described by James Meadows in his book "Geometric Dimensioning and Tolerancing – Applications, Analysis & Measurement [per ASME Y14.5-2009]":

As to "what actual part/assembly reasonably behaves in a way that is represented by translation?", imagine, for example, that the plate from figure 11-27 is assembled with another part with the use of 2 floating fasteners that go thru holes B and C, but the hole corresponding to the hole C is elongated (exactly because designer was aware about the overconstraining issue). This is the situation where the hole C and its counter-feature in the other plate are designed just to orient the parts together (or to be more specific, to constrain last remaining rotational degree of freedom). So in fact there is no point for datum feature simulators B and C to be fixed at basic 51 distance during inspection of plate from fig. 11-27 - this simply will not be a simulation of real behavior of both parts in the assembly. Thus the translation concept which allows to override the default requirement for basic distance between the simulators B and C.

I was hoping for a picture or link to actual hardware. Something that was used to convince the committee to add at this additional level of complexity. Something who demonstrated the critical need.

The example commits the two pins to be a certain distance apart and sets a maximum diameter for the pins so they can be inspected with a fixed pin gage. After making sure of this they become transformed so one moves?

I was hoping for a picture or link to actual hardware. Something that was used to convince the committee to add at this additional level of complexity. Something who demonstrated the critical need.

Click to expand...

Isn't the need to correctly reflect how a part is really contrained in an assembly critical enough to introduce this concept?

The example commits the two pins to be a certain distance apart and sets a maximum diameter for the pins so they can be inspected with a fixed pin gage. After making sure of this they become transformed so one moves?

Click to expand...

That is correct. First, the |pos.|dia.0(M)|A| requirement is verified with the use of two pins fixed in size, spaced at basic 51 and perpendicular to datum A, because there has to be something that controls mutual distance between both holes (otherwise the drawing is incomplete). Then the pattern of 21 holes is verified in a different setup that reflects how the part is mounted in the assembly. Thus the part is placed onto a gage with movable pin C which actually simulates one of the fasteners that does not need to be constrained in location relative to the other fastener.

There could have been a rectangular positional tolerance applied that showed the relation of Datum C to Datum B. This would have unabmiguously described the allowable location of datum C.

I haven't seen any assembly that required a part to be constrained the way described in the example. A standard should not be complicated by adding controls that don't have a frequent, demonstrable application.

It is not about making locational relationship between datum features B and C looser (sometimes this may be highly undesirable for different reasons).
It is about fixing a part for measurements of other features in order to reflect how the part is fixed in a real assembly.

If you have seen an assembly similar to what I tried to describe in my previous replies, then I can tell you that you have seen a situation where use of the translation modifier is a reasonable choice.
If you haven't had a chance to see it yet, I don't think it is a reason to say that this control does not have a frequent application, thus should not be present in the standard. Unless you have seen every application in the world.

"If you haven't had a chance to see it yet, I don't think it is a reason to say that this control does not have a frequent application, thus should not be present in the standard. Unless you have seen every application in the world."

What I implied was that the committee should not have added it unless it was a frequent application, and had seen at least one useful example, preferably many.

So far no support for this approach has included even one application. All examples have been artifically constructed for the purposes of explaining how the concept works, but not a single one explaining why anyone would use it or showing how it is used in a real assembly.