DesignBiz,
You had some questions on the A-B concept relating to datums, datum reference frames, and their definitions in the standard. I've been putting off answering them because what I'm going to say might be taken the wrong way if I'm not careful. Or it might portray me as someone worthy of ignoring in the future. Either of these is probably still likely even if I am careful!
Relating the "two skewed cylinders A-B concept" to datums is where the problems start for me. Here's the definition from 1.3.3:
Datum. A theoretically exact point, axis, or plane derived from the true geometric counterpart of a specified datum feature. A datum is the origin from which the location or geometric characteristics of features of a part are established.
Presumably, the purpose of datums is to provide geometry to orient and locate a coordinate system relative to. The idea is that the datums are extracted from the simulators and then the DRF is constrained to the datums in the specified order of precedence. This can work well for simple datum feature types:
Planar datum feature: datum plane
Cylindrical datum feature: datum axis
Spherical datum feature: datum point
Width datum feature: datum centerplane
Coplanar surface datum feature: datum plane
Coaxial diameter datum feature: datum axis
In certain combinations of these simple datum features, a DRF can be constructed on the datums in a fairly obvious way. Three orthogonal planar surfaces (figs 4-2 and 4-3) is one example. Primary planar surface - secondary cylinder - tertiary slot (figs 4-6 and 4-7) is another. Primary planar surface - secondary cylinder - tertiary cylinder (figs 4-8 and 4-9) is another. Y14.5 and a lot of GD&T textbooks tend to illustrate DRF construction using these simple cases. On the plus side, a good majority of part interfaces are covered by these cases. On the minus side, these are very special cases with unique simplifying properties that do not exist in more complex (but still very common) situations. Many important effects and complexities are masked.
The explanations in Y14.5 are also full of nasty pitfalls. There are several statements and depictions in Chapter 4 that are misleading, not generally true, or outright incorrect. The use of certain terminology is also imprecise and inconsistent and, IMHO, has led to a huge amount of confusion over the years.
One example, that I've mentioned before in another thread, is the diagram on the front cover. The lines labeled "datum axis" are not datum axes, they're coordinate axes of the DRF. The point labeled "datum point" isn't a datum point, it's the origin of the DRF. The planes labeled "datum planes" are datum planes, but they're also coordinate planes of the DRF in this special case of 3 orthogonal planar datum features. The datum planes are also labeled as origins of measurement, which is incorrect. The coordinate planes of the DRF are the origins of measurement, and again they happen to coincide with the datum planes in this special case. So misleading.
Another example is a statement made in section 4.4.2 regarding cylindrical datum features. It states that "a cylindrical datum feature is always associated with two theoretical planes intersecting at right angles on the datum axis". How much confusion has this caused for members of this forum alone? The problem is that the statement is not generally true - it's only true in certain special cases. One is a cylindrical primary datum feature. Another is a cylindrical secondary datum feature that is nominally perpendicular to a primary planar datum feature. A third is a cylindrical secondary datum feature that is nominally in line with a primary spherical datum feature. There may be a few other cases that I haven’t thought of. But that's it. In all other cases, the “two theoretical planes” on the datum axis conflict with DRF planes that have already been established. For me, trying to apply the idea of these two theoretical planes in a general sense has been a conceptual wild goose chase. I’ve tried to make sense of it but for the life of me I can’t. Based on posts we’ve seen on this forum, others have run into similar difficulties.
For datum feature configurations other than the aforementioned “special cases”, constructing a unique DRF from the datums just isn't possible, for a variety of reasons.
For simple datum features that are not orthogonal to each other, the datums are well-defined but the origin and/or clocking of the DRF is not obvious or unique. One example of this is the "two skewed cylinders" referenced A|B. The primary datum axis establishes the direction of the DRF's Z axis and the X and Y origins. Because the secondary datum axis is skewed relative to the primary, there is not an obvious and unique way to constrain the DRF's rotation about Z and or define its Z origin.
For non-simple datum features, the location of the datum is not well defined. As a consequence, the origin of the DRF is not well defined. One example is the hole pattern secondary/tertiary datum feature, where the location of the "datum axis" is arbitrary. For circular or rectangular patterns, the center is an obvious, but still arbitrary, choice. For other patterns, there is no obvious center or origin point. There have been many discussions of the “center of the pattern” issue on this and other GD&T and CMM forums.
Some GD&T books, as well as the recently released Y14.5M-2009, expand the definition of a datum to include combinations of a point, a line, and a plane. This allows datums to be identified for features like cones, extruded shapes, and complex surfaces. For example, the datum for a pattern of parallel holes would be a “line on a plane” and the datum for a complex surface would be a “point on a line in a plane”. This is somewhat useful for visualizing what degrees of freedom are constrained by a particular datum feature when it is referenced as primary, but that’s about it. There is still the problem of non-uniqueness. For the hole pattern, the locations of the plane and the line are still arbitrary. For our two skewed cylinders referenced as A-B, the datum is a point on a line in a plane. But the orientation and location of the plane, the orientation and location of the line in the plane, and the location of the point on the line are all completely arbitrary. So we’re not any further ahead as far as defining a unique DRF origin. I just don’t find it very useful. Don't even get me started on how these datum types would constrain DOF's when referenced as secondary or tertiary.
So where does that leave us? In most FCF applications, it doesn’t really matter where the DRF origin is because everything is relative anyway. If certain DRF axis directions and/or origin are desired by the designer for some reason, these can be annotated on the drawing or model in the form of a labeled set of coordinate axes related to the datum features by basic dimensions and angles. The model defines the relationship between the datum features and the DRF, and thus defines the relationship between the datum feature simulators and the DRF. When it comes time to build inspection fixtures or CMM programs, a unique DRF can then be constructed directly from the (physical or virtual) simulators. Without the need for extracting combinations of points, lines, and planes for each datum feature and trying to hang a DRF on those. I’m not saying that this “simulators to DRF” approach is easy in every case, but it is doable.
Comments?
Evan Janeshewski
Axymetrix Quality Engineering Inc.
![[stpatrick2]](/data/assets/smilies/stpatrick2.gif "[stpatrick2] [stpatrick2]")
