Re: ASME Y14.5M-1994 - positional tolerance with no datum/reference

[JNieman]

I have a "best practice" question that I couldn't find specifically addressed using the 'search' function:

We have a couple customers who will sometimes/often call out hole positions with a positional tolerance but not provide any datum or basic dimension.

We -usually- are supplied a 3D model in many cases, absent any annotation.

So what we have done before, in order to justify the parts being good, and explaining it to the buyer why we say it's good, is that we will make an assumption of what it should be relative too. Often times we will know the usage of the part, or can assume what it is based upon the geometry and features. Sometimes it's obvious, sometimes the part description indicates it, sometimes we have to take a shot in the dark.

OTHER THAN calling the customer and requiring amended documents, which would be the ideal, perfect-world scenario, that rarely happens with these customer, what would you do?

 

[MadMango]

I assume you are supplying a COC (certificate of conformance) with the parts, I'd just document that the datums were not provided. Let their Quality department do the heavy lifting.

"Art without engineering is dreaming; Engineering without art is calculating."

Have you read faq731-376 to make the best use of these Forums?

 

[CheckerHater]

It makes sense to assume that dimensions derived from 3D model are Basic. If customer doesn't specify it on their documents, then you should, as MadMango said it.
Naturally the questions of flatness, parallelism, perpendicularity, etc., etc., etc. still open.

 

[JNieman]

Most of the geometric feature conditions are covered by general notes as a bare minimum for acceptance. They're good about having documentation to cover most things. There is rarely a topic unaddressed.

It's merely the less-than-ideal annotation that can get us in a bit of a bind.

I understand measuring the model can be assumed to be 'basic dimensions' but that gets us nowhere further than a paper/pdf print does, does it? Which measurement do we assume is the primary, secondary, tertiary? It's still just assumptions and guesses.

We're making good parts that function completely, and have good communication with the customer which makes me confident they'd tell us if there was rework required on their part.

We do a lot of very-hot turnaround jobs for some customers, and I am very willing to go above the minimum requirements to make sure the part is good for their usage, and not just "legal" to the print. No one is happy if the customer is disappointed, even if it is their fault. My work can still put a bad taste in their mouth even if it was no 'fault' of my own.

We do, depending on the appropriate quality-level assigned to the order, supply a CoC, FAI, or varying level/thoroughness of inspection report. However, for these "ambiguous" dimensions, we'll often note "assumptions made" and provide the dimension. I guess it's just "value added" on our part (hopefully).

I appreciate the input. It's nice to know what other people tend to do, so as to modify, if necessary, our own standards and practices.

Thank you

 

[3DDave]

They have a common datum reference frame - each other, just like a regular pattern. Tell them they are all located to each other properly but there is no location tolerance to tell where on the part they should be.

In an ideal world every supplier given this documentation would either refuse to bid, or having accepted it, pick some orientation that is obviously bad and commence drilling. Unfortunately, if even one shop does produce a useable part with incomplete documentation, the issuer will believe they did it all OK.

The problem for the customer is that each place they go will tool up in a different way and they are likely to get 'compliant' parts that sometimes work right and sometimes don't. If there is no feedback to them, they will continue to send out drawings and get back liabilities.

 

[JNieman]

Well, they do and they don't. Functionally, each of the holes in the CURRENT part in question are independent and have no logical or practical need to be related to each other, but I'm wishing to discuss it generally, rather than to a specific example, which may be introducing ambiguity into a topic about ambiguity, ironically enough.

We do report any time we have to "assume" on inspection, however. For some customers they will simply change their drawing to whatever we tell them to, or ask us what we think it should be and revise. Sometimes they'll just get our parts in, measure it, and update their drawing to a sort of "as-built" drawing, and from then on, our part is their standard. Other customers, however, are giant beasts that are sometimes hard to get changes made, and other times we're making parts from drawings that are considered legacy, have been in service for a long time, and "unchangable". We do what we can, yes.

I think the variety of comments is interesting.

 

[CheckerHater]

Any CAD model has an origin, as well as XYZ axis.
In the models you receive, are they normally associated with any features at all?

 

[Belanger]

Any time two or more features are controlled by the same feature control frame using position, then they are AUTOMATICALLY toleranced for position to each other. That's true even if no datum references are given, so the callout is still valid (though maybe not wise).

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

 

[JNieman]

CheckerHater:

While you are correct that an intelligent designer or engineer would place their model accordingly with Datum Axes/Planes, or at least have those features defined in the model, sometimes we get a model whose origin is referenced to something far-off and imaginary, as far as we're concerned.

For example, we may be doing a simple jig used to guide drills to their appropriate position, to mount equipment on an airplane frame. This part is often extracted from a model of the entire frame, whose origin may be 30 feet away and askew.

 

[ewh]

Yes, it is fairly common to model aircraft parts in their assembled positions, then constrain when the design has matured. In such situations, the model absolute origin is of little use downstream when it comes to fabrication of parts. Rather than move the customers model to suit the machinist, it is much preferred to orient the WCS to what is good for the machinist and go from there. This can save countless hours trying to get the two files to align (especially with parts containing mostly non-uniform surfaces) if/when problems come up that need investigation as to why the two models don't match.

“Know the rules well, so you can break them effectively.”
-Dalai Lama XIV

 

[CheckerHater]

JNieman,
I’ve seen my share of garbage files, no doubt.
From the OP I understood you are dealing with documents that not always comply with formal rules, so we are discussing the ways of extracting as much as possible useful information.
All I was trying to say, that if saw model like this, I would make some reasonable assumptions about it:

 

[cjccmc]

On the project I'm currently working I repeatedly see an error kind on the flip side of this one. Holes etc. have full TP tol but datum planes are just drawn thru the part without identifying a datum feature. Many parts were done this way and delivered so not sure what the suppliers are actually using to establish the datums. As long as the parts fit things will likely continue is this less than ideal manner.

 

[KENAT]

cjccmc what you're describing certainly sounds wrong.

Look at it this way. If parts come in functionally wrong, but not wrong to the drawing how can you reasonably reject them?

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[cjccmc]

Kenat I hear you. I think that so far we have avoided any impacts because machining is accurate and fits loose enough to compensate for the fuzzy definition.

 

[JNieman]

@CheckerHater

I believe we're on the same page. Those datum planes you placed would be the same assumptions I would make if no other information was supplied.