Projection Error When Evaluating True Position of Threaded Hole Far away from Primary Datum

[ktnorberg]

On a Primary datum with a small surface, there is bound to be some amount of error in approximating this plane on CMM. My quality team is telling me there is projection error when evaluating the .030" true position on the opposite end of the part by keeping datum A as primary. The true position measures in tolerance when evaluated using the |B|A|C| datum structure instead. The customer wants to keep the datum structure the same to control perpendicularity. Is there any allowance in ASMEY14.5 or a measurement method that would let me measure this part accurately on CMM or is the only way to push for a drawing change?

 

[3DDave]

Which surfaces are allowed to have a gap between them and the mating part in the next assembly? Example, is it OK for the part to lean and leave up to a .5 gap for the datum feature B surface with only a couple of contact points?

 

[ktnorberg]

@3DDave
Yes, the datum B surface may only have a couple contact points. Being that datum B is the secondary datum, currently the surface is scanned by the CMM, then the model is constrained by 2 outer tangential contact points on the datum B surface.

This has been causing the model to tilt giving what I'm told is projection error when evaluating the true position of the hole. I am hoping to understand the best way to eliminate that projection error while still conforming to Y-14.5.

 

[3DDave]

I'm not talking about the model - when you hand a part to the assembler is that what you want to happen?

 

[ktnorberg]

Oh, yes there is allowed to be a gap.

 

[3DDave]

Like a half inch gap, maybe an inch?

Anyway - just noticing this "projection error." There is no projected tolerance; the position tolerance applies only over the length of the feature.

 

[Burunduk]

ASME Y14.5 specifies that datum features need to be of sufficient size to permit its use. This is specified in para. 7.8 of the 2018 edition and para. 4.8 in the 2009 edition. Not sure I know what you mean by "projection error" but I suppose it has nothing to do with a projected tolerance zone Y14.5 concept as there is no circled P modifier in the position callout. But if the results you get inspecting the position requirement are not repeatable (i.e you get different results on multiple measurement attempts in the same part) you have a sufficient basis to require a drawing revision.
The small face doesn't look like an appropriate primary datum feature that's able to stabilize (constrain) the part in the required degrees of freedom both for inspection and at assembly.

 

[TheTick]

I don't even know where to begin. If it was food, I would send it back and order something else (probably somewhere else).

 

[Jacob Cheverie]

By projection error, ktnorberg is referring to how skewed the measurement becomes with any small variation on the establishment of datum A. Being the primary datum, A sets the overall orientation of the coordinate system from which measurements are taken. If there is any issue with the establishment of datum A, the error will be "projected" over the length of the part.

For example: If datum feature A had a slight tilt (out of perpendicular to datum C), then using it to establish the YZ-plane would cause any positional error to become more dramatic (in X and Y) the farther the positioned feature is from datum feature A.

As far as I'm aware, this is simply a poor design and there is no real way to circumvent it from Y14.5. Typically you would request a change to the datum structure. The designer can possibly add a perpendicularity refinement if that is a control that they wish to use.

 

[ktnorberg]

Jacob Cheverie is correct. Thank you for clarifying. Unfortunately drawing changes take a long time to get approved from my customer. Usually I provide a redline print with an alternate datum reference frame making datum B the primary in similar cases.

Is there no allowance or anything to move the coordinate system? I found an article that provides an example of what can be done to logically solve this measurement error(

http://www.geomet-cmm-software.com/KB/App/2/Chapter4/KB10229.htm;

example #2), but I do not understand if it is permitted per Y14.5.

 

[jassco]

Hi, ktnorberg:

Apparently, the designer does not understand what PRIMARY, SECONDARY AND TERTIALY datum features mean.

Best regards,

Alex

 

[Jacob Cheverie]

ktnorberg,

The example that you provided a link to isn't quite the same situation. There are no datum references. In any event, you can translate your coordinate system along the direction normal to your datum feature A (as that direction coincides with the axis of the considered feature and position isn't being measured in that direction). Unfortunately, this won't help you. Being a primary datum, datum feature A sets the plane for measurement. Whether you look at it from a different location or not, the considered feature will suffer from this "projection error".

 

[drawoh]

ktnorberg,

Forget the CMM. Let's stand this on a granite reference block using the datum[ ]A feature as the primary reference. We simulate datum[ ]B by pushing a perpendicular fixture up against the face, contacting at two (not three) points. Datum[ ]C is another perpendicular fixture that contacts at one point, but I see no problems.

A good working assumption and good practise, is that your primary datum feature is your primary mounting feature. Really?

It sounds like you are the fabricator. You are supposed to use the primary datum feature for your primary fixturing. Is this even possible? On this part, if I called up feature[ ]B as my primary datum, I would specify that the datum applies when the face is clamped to a flat reference surface.

I suggest you call your customer. Tell them that their datum scheme is not feasible, and that you want to use feature[ ]B as your primary datum. Either they send you an official letter acknowledging this, or you tell them you will make a "best[ ]effort", but cannot guarantee compliance .

--
JHG

 

[ktnorberg]

@Jacob Cheverie, That all makes sense. What about translating the coordinate system along the direction normal to datum feature A until it reaches the surface the hole is cut into, then rotating the model so datum A plane is parallel to the surface with the hole? I know this seems like it's going off the rails, but is this explicitly forbidden by Y14.5? Or would I run into other problems?

@drawoh, I think you're absolutely right. Using a datum simulator would be the best method to evaluate this part, but the image I posted is a simplified version of the actual part. The real part has many holes around the perimeter and have true positions called out similarly. So using a datum simulator setup would take a long time to evaluate all the holes compared to the CMM. The designer's intent was to control perpendicularity between the datum A surface and the hole. So my redline has datum B as the primary and constraints put on perpendicularity of the hole, but I was hoping to learn about a CMM measurement method that would eliminate the back-and-forth and circumvent pride of the designer when this inevitably comes up again on other parts.

 

[Jacob Cheverie]

ktnorberg,

As soon as you rotate your coordinate system, you violate the datum reference frame and are thus no longer reporting what you are being asked to report. There is a nominal location and orientation for the hole. It is nominally perpendicular to A, .xxx from B, and .yyy from C. Any violation of either of those three requirements will change your position measurement.

The "projection error" isn't an unintended consequence of using the CMM or even your choice of software. It is simulating what would happen in the actual mating situation. Imagine that your datum feature A surface was not parallel to the surface that the hole is drilled into. If you truly interpret per Y14.5, you would first place A on it's simulator. This means that you would put the part down (A first) on a granite slab. Now you can imagine how more error in parallelism (and/or more length) between the two faces will actually "shift" the position of the hole if you were trying to put a bolt in. As soon as you start rotating anything, you are essentially taking A off of the plate. Therefore you are no longer simulating your primary datum, which implies that you are not reporting per the requirements of Y14.5.