Profile and position callout 1

[qcgoofyfoot]

I have a simple rectangular part, 6"x4"x.125" thick with a hole in the center.

The flat surface is called out at "A".

The 6" edge is "B".

The 4" edge is "C".

Profile of the perimeter is called out to be .01 with respect to datum A.

The position of the hole is called out with respect to ABC.

Is this a valid GDT callout?

Specifically, is it acceptable to callout the perimeter with a profile, and then use the edges as datums to find the hole?

The perimeter is not being called with respect to ABC, so I think it is ok.

 

[Belanger]

OK -- if the true profile goes wherever the datum plane goes, then what's the point of referencing the datums?

We already knew the geometry of the part from the true profile, and the datums contribute nothing in terms of orientation or location.

John-Paul Belanger
Certified Sr. GD&T Professional

http://www.gdtseminars.com

 

[axym]

Dave,

The datum reference frame is defined in terms of perfectly flat planes that the part is put into contact with. Physically, this can be accomplished using three perpendicular surface plates. This is what you've seen in functional gages. The DRF is in the plates, and they contact the part at 3 high points on the primary datum feature, 2 on the secondary, and 1 on the tertiary.

If a CMM method is used instead, things are more difficult. For each datum feature, you need to somehow approximate the plane that the physical plate would make contact on. The random 3-2-1 method doesn't work. On the primary datum feature, you need to probe a lot of points and then use the Tangent Plane algorithm. On the secondary datum feature, you need to probe a lot of points and then find the "orientation constrained tangent plane". Same thing for the tertiary. Some CMM software packages have a built-in algorithm for orientation constrained tangent planes, but a lot don't. So you often need to use other tricks to create them. This is difficult, and I can understand why a lot of CMM operators don't go to these lengths. But the reality is that the simple 3-2-1 methods commonly used on CMM's don't establish the right DRF according to the Y14.5 definitions. It's one of the biggest hidden pitfalls of using CMM's. Y14.5 was written in terms of physical simulators and high point contact and tangent planes, and approximating those using CMM's is not easy.

John-Paul,

If you don't feel that the true profile should go wherever the datum plane goes, then where should the true profile go instead? The point of referencing the datum features, as it always is, is to orient and locate the tolerances zones.

I'm not sure what you mean by the statement "we already knew the geometry of the part from the true profile". Can you clarify?

I don't agree that datums B and C contribute nothing in terms of orientation or location. In my first diagram, it is clear that the tolerance zone sits differently on the part when B and C are referenced. When only A is referenced, the zone floats. When B and C are referenced, the zone is oriented and located to the high points of B and located to the high point of C.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[dingy2]

Evan:

I do not disagree with finding the high points using checking fixtures but on a CMM, it is not done and that is reality.

Even if one was lucking thinking they found the high point, how does one create the plane since you need at 3 points on a rigid surface. I do realize the one could take numerous points using a CMM and create a best fit plane but does that include the high point or create a plane from the average of all the readings? It is the average of all the reading. How does one know that they have actually found the high point using a CMM?

Please explain the "tricks" that you mentioned in your last post which should be used by CMM Operators to find the high point and what should they do with it?

Per 4.10.1 it says "This primary datum feature contacts the datum feature simulator on a minimum of three points." This method is used if the feature if it is "rocking or unstable". I think finding a high point would render the feature unstable and susceptible to rocking.





Dave D.

http://www.qmsi.ca

 

[axym]

Dave,

We're starting to get way off topic here. CMM verification of GD&T is a whole other ball of wax. Perhaps we should start a separate thread.

I will comment on your reference from 4.10.1 though. It sounds like you're bringing some unrelated datum target concepts in and applying them to full-surface datum features. When it says "this primary datum feature contacts the datum feature simulator on a minimum of three points", it's not referring to the number of points you need to pick on the datum feature to define a plane. It means that when the primary datum feature (an imperfect surface) and its datum feature simulator (a perfect plane) are brought together to constrain 3 degrees of freedom, then the contact will be at 3 points. When the secondary datum feature (an imperfect surface) and its datum feature simulator (a perfect plane that is exactly perpendicular to the primary simulator) are brought into contact to constrain 2 DOF's, the contact will be at 2 points. Similar logic applies for the 1-point contact on the tertiary.

The reference to rocking or unstable datum features is to address possible convex conditions in which 3-point contact could be made in more than one way. One way to prevent rocking and instability is to specify datum target points. But that is a different thing entirely. With datum targets, the locations of the simulators (and thus the contact points) are well defined relative to each other. With full planar simulators, the contact is on the high points and will be different on every part.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Belanger]

Hi again Evan,
Per my last post, I meant that since the tolerance zone is all around, it follows the true profile. We know the true profile because the basic dimensions for length/width would be given. And the "implied 90" rule tells us that the four corners are perpendicular (basic angle).

Now superimpose the tolerance zone -- equal bilateral -- over that true profile. The size and form are controlled by this tolerance zone. But so are location and orientation. Location meaning the "position" of each edge from the opposite edge; and orientation because the perpendicularity of the four corners is constrained within our tolerance zone.

Referencing datums B and C adds no value to what I've described above. I sorta see what you mean about getting half the tolerance zone, but I'm just saying that we shouldn't go down that road in the first place.

John-Paul Belanger
Certified Sr. GD&T Professional

http://www.gdtseminars.com

 

[axym]

John-Paul,

The tolerance zone straddles the true profile, which is defined by basic dimensions and implied basic 90's and parallel relationships. No argument there.

The all-around profile tolerance zone controls the overall form and size of the pattern of surfaces (as long as we're clear that we're using the term "size" in the generic sense and not bringing in the Y14.5 idea of size). Orientation and location are controlled, but only in a relative sense.

Referencing datums B and C may not add any "value", but it does have an effect. For example, the perpendicularity of the left and right sides of the part relative to datum feature B is controlled more tightly.

I agree with you that we (as designers) shouldn't go down that road in the first place. But sometimes we (as QA) have to go down that road because the designer did. I have seen this exact thing on prints before. The designer had specified a general profile tolerance to ABC, that covered all of the surfaces on the part. The datum features were included in the profile tolerance, so the self-referencing effect was there. So we had to figure out whether or not this was legal and, if it was, how to orient and locate the tolerance zone to inspect the part.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[MechNorth]

OK, I really like this one, and I have a tangent to go to in a moment.

Evan, all-around is not the same as all-over, so the flatness of datum feature A and the opposite face are not controlled by the all-around surface profile. I do, however, agree with you that the tolerance zones for datum featues B and C are effectively cut in half when referenced in the all-around datum reference frame. And also agree that the shop is unlikely to ever recognize that reality. I would definitely just reference datum A in the all-around profile control. Tks for posting the graphic.

Dave, we've talked on the CMM simulation of datums before, and I still say that datum simulator fixturing should be used on CMMs. You have experience in certain sectors, and I have it in others. In (too few) cases, datum simulator fixturing is indeed used; it speeds setup and improves repeatability. No, it isn't practical in all situations, but I've established that it is often cost effective even on very large items such as 20'x5'x8' welded frameworks.

Now, my tangent scenario. Picture two coaxial cylindrical features used to establish a common datum axis, with a total runout control applied to each individual datum feature surface and referencing back to the common datum axis. This concept tends to blow the minds of designers and inspectors because they forget that datums and datum features are not the same thing. Granted it's unpleasant to inspect because you have challenging fixturing, but it is often the functional requirement, such as when bearing surfaces on a shaft are used as datum features. Just something for y'all to munch on.




Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[axym]

Jim,

Regarding your tangent scenario, you're talking about something like features C and D in Fig. 6-51 in '94 or Fig. 9-6 in '09. I've munched on this before, and it always leaves a bad taste.

Referencing the total runout controls for C and D back to C-D is a pain, and I don't think it's functional anyway. I would prefer a single total runout control that applied to both C and D and didn't reference any datum features. If you can find an axis that both C and D have total runout of .02 relative to, then the part passes the tolerance. But this isn't currently allowed in Y14.5. So we're stuck with the "self-inflicted" runout.

The thing I have a problem with is referencing a pattern of features of size as a datum feature RFS. The exact behavior of the datum feature simulators, and thus the definition of the "common datum axis", goes into the twilight zone.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[MechNorth]

Evan, I don't get the bad taste, but it's a buggar to fixture. You are controlling the location of each surface to the common axis; not sure how that's not functional. Following your preference, you'd do a total runout on the two surfaces to find a common axis to which both the individual features would be controlled, then what?... you'd control all the other features back to that common axis? What I read in your preferred scenario is essentially what referencing back to C-D accomplishes in the first place, or am I missing something not communicated yet?

I don't have a problem with patterns at RFS; I've worked with a few precision situations where it was absolutely necessary, and so a corporate addendum to the '94 standard was issued.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[Belanger]

I too wonder about a datum being a pattern of holes at RFS. Perhaps I'm thinking of it too much from a hard-fixturing point of view, but what happens if one of those holes is out of position slightly?

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

 

[axym]

Jim,

Controlling the location of each surface to a common axis is functional, I don't have a problem with that. The aspect that leaves the bad taste is inspecting the total runout on features C and D. If we were to set up datum axis C-D using physical equipment, features C and D would be inserted into two precisely coaxial chucks. But once C and D are covered up by the chucks, how do we inspect the total runout on them? The inspection could be done using a CMM, but Y14.5 wasn't written with CMM's in mind so I wonder what the real intent was. That's why I think that having the tolerances on C and D referencing the common axis C and D is a pain and probably not functional.

It would be better to have a separate tolerance on the datum features, to qualify them for use as datum features. This limits the uncertainty in the common datum axis that will be extracted from them.

I have some further comments (questions really) on exactly how the two chucks would be adjusted to establish the C-D axis (the problem of the pattern-of-cylinders datum feature at RFS). As John-Paul brought up, ambiguities can arise if C and D are not perfectly coaxial (which they never are).

Perhaps in the next post.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Belanger]

Gosh -- When I posted my last comment I wasn't really paying attention. You guys are talking about two coaxial cylinders, a la Fig. 4-25 of Y14.5. I don't know if that's such a big deal; the holding device simply closes down until it gets a good grip on the two cylinders.

My comment was targeted at Jim's statement that there's no problem with patterns at RFS. I was thinking of something like Fig. 4-26, if it didn't have MMB on datum reference B. Then I think we'd be in a tight spot.


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

 

[fsincox]

axym,
I am sorry to have to disagree; I see 2 bearing journals establishing a common (spindle) axis as a VERY functional thing. I believe you are using the term functional, when you seem to mean “easy to inspect”. This is one of the things I have been trying to get a feel for from this group, and appear to be slightly be more radical on, so to speak. In my opinion design criteria requirements is what we want to inspect for, easy or not, what else is the point.
In my machine tool days, I have seen these spindles inspected on a machine with a micro-tilt table with adjustable indicators. They place a spindle, large end down on the table and have an indicator on each bearing journal. The table is tilted until the indicators are zeroed for a full rotation. Another indicator is on a movable slide to traverse the toleranced features. I did not do this myself, it is obviously a special dedicated machine and not necessarily easy.
I need to start another topic on this runout thing. I am not sure runout is a functional thing, I see it as as an easy to inspect thing?

 

[MechNorth]

Haven't seen it myself, but an inspector described a fixture he used to hold the bearing-surfaces C & D to establish the primary datum axis, then clamped up on other coaxial features, released the C & D clamps, and went on to check the C & D surfaces. Complicated, yes, but apparently doable.

My preference with the coaxial bearing surfaces is to specify the tooling centeres in the ends of the part as the datum features (C-D); it has little effect on the inspected outcome or the function, and is typically how ALL those features were produced in the first place, so little effect overall. Even when the part comes back for refurbishment, those tooling centres are typically in good enough shape to re-establish the datum axis again.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[fsincox]

Jim,
I, in no way, meant to imply it is the only way possible. Can't you inspect to centers and take the relative measurements and get close to there anyway for a simpler solution than special equipment? I really am a book engineer and not a hands-on engineer. This is why I was looking for a forum to discuss issues like this. The point really is, it is absolutlely a possible functional requirement.

 

[MechNorth]

No, no. I understood it to be one possible method, just as mine was another. I would be interested in seeing the setup you describe; I'm just not envisioning it right now, so I'm not sure how it establishes the datum axis or revolves around it. Really, it's all going to come down to the tolerance, and how much risk you can tolerate. I was dealing with surface profiles of 0.0025mm, and I don't know how much of that tolerance your suggestion would have eaten up; 10% is generally considered the accepted limit of measurement system error. The results would not be the same, but may be acceptable depending on the tolerancing and function. If you can accept more risk by way of possibly passing an unacceptable park, you may lower your production/inspection costs; on the otherhand, if a field failure of the component could mean signficant safety and/or cost issues, then you will be more risk averse and to a true rather than approximated inspection. If you have some pics of the setup, that would be great.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[fsincox]

Jim,
It has been 7 or 8 years since I worked there. I am thinking is was mimilar to the setup below, except, I seem to remember more indicators than one. I guess it is possible the zeroed indicators on the extremes of the bearing journals were just separate indicators, used only to zero out the table and as a control, that would work wouldn't it?

http://news.thomasnet.com/companystory/828297

I know this can not be the same one.
I believe using the relative indicators as a part is turning on lathe centers would lead us the same way would it not, except there is no adjustment it is either there or it is not?

 

[MechNorth]

Hmm. It seems more like a dynamic balance test to me, or like a manual means of trying to achieve what a CMM would do. There seems to be a "differential" reading needed between the two journal indicators to establish a common axis of minimum total runout between the two surfaces. I don't see that as exactly the same as using physical chucking, but may not be significantly different depending on the quality of the work (fabrication & inspection) and the tolerances specified. After tilting the workpiece to establish/approximate a common axis, you'd then have to lock it down and start probing the other features.

The double-chucking seems cumbersome for individual pieces, but on production runs, customized fixturing is often both cost-effective and practical.



Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com