Assigning datum features to hole patterns 3

They can be dimensioned as is reasonable. The idea that multiple holes collectively define some axis is not very good. Depending on locations, orientations, and sizes of the holes, two of them will be the limiting factor for rotation in one direction and some other two will limit rotation in the other. Done in a certain manner the part, with clearance holes, won't be able to rotate at all.

I don't see the term 'centroid' mentioned in the 1994 or 2009 versions of Y14.5; it may have been in the 1982 or earlier version.

tunalover said:

can dimensions to features elsewhere on the part be taken off of any hole in the group

Click to expand...

I'd say yes, you can.

Yes, but this is a case where the MMB modifier would really help answer the question. With that "M" then you can dimension off any one of the features because the real measurements would be taken from the datum, which is simulated by a virtual condition pin. And this is something that is constant.

Paragraph 4.12.3 of Y14.5 is the source for this. The best sentence for your question: "The origin of the datum reference frame may be established at the center of the pattern of the datum feature simulator where it intersects plane A, as shown in Fig. 4-26, or at any other location defined with basic dimensions relative to the datum feature simulator as in Fig. 4-28" (emphasis added).

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

Insert here: long discussion of why RFS applied to hole patterns used a datum reference is a bad idea.

Belanger-
Does similar verbiage appear in the 1994 standard? Our company is still working to the 1994 standard.
TL

Tunalover

Star for 3DDave.

I'd go farther and assert that using a pattern of features as a datum is a bad idea, regardless of how their size or interrelationship is accounted for.

My first GD&T instructor, in 1967, asserted that GD&T came out of the gage-making trade, and that it helps to think of datums as being features of a gage or fixture, not features of the part.

Thinking of datums as gage features, consider the complexity necessary for a gage to pick up a group of holes of varying size and location, and calling some reasonable point on that gage a datum. What's reasonable? Centroid only makes sense if the pattern has symmetry. If one hole is intentionally offset for clocking purposes, as sometimes happens, that moves the centroid, but it might be more reasonable to omit the clocking hole from the centroid calculation, and it would be even better to omit any calculation entirely.

Even if you use a CMM instead of fixed gages, the complexity remains, and adds to the normal confusion.

Remember that you are expecting your drawings to be interpreted correctly, every time, by every viewer, by people who may not be GD&T experts, and who may not have much formal education of any kind.

Remember also that the only reason you have a job is because the machinists/ CNC operators/ artisans charged with actually making the parts do not have erasers or 'undo' buttons.

</soapbox>


Mike Halloran
Pembroke Pines, FL, USA

Not quite sure about that (discouraging the use of a pattern as a datum). Instead of focusing on gages, I would simply point to the function. If the function of a pattern is to be the main "holding" of a part (which is usually what a bolt circle is for!), then using the pattern as the datum feature(s) makes perfect sense. Of course, that also means that the MMB modifier makes sense. Yes, the gaging is a good way of talking about it, but only because a gage simulates the function, and function is what really matters.

Tunalover -- in 1994 check out paragraph 4.5.8. It doesn't say anything about having a choice for where the theoretical center is; it just says that "individual datum axes are established at the true position of each hole [virtual condition]." So the 2009 standard is more clear about it, but I would still say that even in 1994 it doesn't matter which hole the dimension is displayed from, or even from an imaginary center (if one can logically be found).

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

The ideas of "centroid" and "rotating part" were abandoned in 1994 version. 2009 version explained further that "The origin of the datum reference frame may be established at the center of the pattern of the datum feature simulator…, or at any other location defined with basic dimensions relative to the datum feature simulator…"
So there is no need to look for a centroid and it doesn't matter how pattern looks like.

"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

I must loudly support JP on this. The training and reading I do seems to indicate that the desire of the D&T standard (if you will) is to indicate functional requirements and not process. If a hole pattern is the driving feature what is wrong with saying so, the knowledgeable industrial engineer should translate it to a manufacturing set up that achieves that goal, if they want to use only one or two of those holes for their own purpose that is their decision.
This is the fundamental issue we seem to go back and forth on here, and in industry, that I feel shows how insecure/tentative the support for the whole D&T thing really is. To me the standard has already spoken on the issue, manufacturing people have never liked when we did this and they obviously still don't. Everywhere I have worked that had an internal shop fights it, looses the support and goes back to the “good old way”.
Frank

But I've heard that it's difficult to inspect hole patterns from a datum formed by a hole group. Should I stick with composite positional tolerancing to tolerance functional groups? Or can I go all out and make, say, a 36-hole pattern a datum. I've got some complex parts with lots of holes of different sizes to deal with and would like to know the best way to deal with them.

Tunalover

Understand, I am talking big picture stuff, I also understand compromise is stuff we have to live with day to day, it IS real life.
I am talking the ideal, the drawing should state the function and let qualified manufacturing-inspection people do their job, figure out how to make it and inspect it.
If I have learned anything in this forum, it is that I may be too idealistic.
Frank

fsincox-
In your next-to-the-last post, is the "good old way" with simple, yet confusing and ambiguous, bilateral tolerancing e.g. without geometric controls?
Thanks,
TL


Tunalover

Basically, or "manufacturing" datums on the extreme edge corners, just like shown in "the books".
Frank

Or, picking one hole in the pattern as secondary datum and another hole as tertiary.

I admit to having done this on occasion rather than picking the entire pattern as the datum. While it's still a compromise from true functionally driven datums it's arguably less of a deviation than some other options.

Ken, I agree
If the holes are referenced MMB it really does not "do violence" to the design due to "the datum virtual condition rule" (formerly, rule #5).
Frank

Picking two out of n features in a pattern as secondary and tertiary datum features referenced at MMB creates a risk that bad (non-functional) parts will pass inspection.

Just as an example, picture a situation where related actual mating envelopes (RAMEs) of these two datum features are at different-than-MMB size, and RAMEs of at least two other features within the pattern are at MMB size. With the entire pattern selected as datum feature there would be no datum feature shift available in such case. But with just two features chosen as datum features the inspection will be allowed to take advantage of a datum feature shift which actually will not exist in reality.

As for the discussion about a centroid, I agree that there is no need to look for it and that the origin of DRF can be anywhere as long as basic dimensions define location of the origin relative to datum feature simulator. I would just add that this applies regardless of whether the datum pattern is referenced RMB or MMB - in both cases the DRF origin is derived from datum feature simulator pins (or holes), and in both cases mutual location of those pins (or holes) is the same.