Localized Duplicated Datums

[MayCad]

Hello,

I am wondering if someone could shed some light on something. I have a drawing that is a rectangle. At the longitudinal CL of the rectangle is Datum "B". There are features at either end of the rectangle which are being dimensioned. It has been suggested that I create a Datum "C" and "D" in the same CL location as "B", that references "localized" features for either end.

I do not understand this. In a perfect world all features for the rectangle would be perfectly in line with "B", and need to be if possible. Why then, or what purpose would it serve to have a "localized" Datum to re-establish a CL that is already established when I would prefer that the CL for all longitudinal features not vary?

I'm unable to share my DWG but I can sketch a basic rectangle with features on each end that all share the same theoretical CL while calling out separate Datums for each end if this doesnt make sense.

Please, someone make sense of this.

 

[3DDave]

The datum feature can be the center plane of the width of the rectangle - it cannot be the centerline.

It isn't clear what you mean by "localized." I don't know what purpose someone else has in mind - perhaps ask the ones making the suggestion.

 

[MayCad]

Essentially they want me to reference features on each end to their "localized" datums although all features need to align to the Primary Datum.

By centerline, I mean the center of the width of the rectangle.

 

[pmarc]

MayCad,

This is how I visualize your description:

If my understanding is correct, then depending on the design requirement (which is often called a "function" of features) different options are possible. To give you just two of them:
1. Features C, D, E, F may all be controlled relative to datum B.
2. Features C and D may be controlled relative to B, but then E may be controlled relative to C and F may be controlled relative to D.

It is worth to remember that even though all 5 features of the part I sketched are shown centered in nominal/perfect condition, in reality they will never be perfectly centered, so depending on which feature is used/selected as datum feature, the result of position measurement of the other features may be different.

As a thought experiment, imagine that the part from my sketch has been produced such that feature B is offset from the common center plane, but the other 4 features still share the same center plane. For such an actual part, if the drawing definition was as described in the option 1 above, the inspection report would show non-zero position errors of features C, D, E, F relative to B. However, in the option 2 the report would show non-zero position errors for features C and D only. Features E and F would be reported as perfectly located relative to their "localized" datums C and D respectively.

 

[Burunduk]

MayCad,
Although a datum center plane is theoretical, it is derived from actual datum features.
"In a perfect world all features for the rectangle would be perfectly in line with 'B'" is true, but in a real world they are not.
So depending on which feature you derive the datum from, you end up with a different datum for the given as produced part.
Therefore, the choice between one datum feature to center everything to or local datum features, depends on functional requirements. Different datum features at each end will produce more tolerance accumulation between the end features (and by the way, there should also be some controls of the datum features relative to each other). But, in the case that the function of the part requires that each feature is located accurately to the local longitudinal feature near it, the suggestion you got makes sense.