Why Place MMC in Reference Datum when tolerance is in RFS? 3

[dingy2]

I see this so often and I wonder the purpose of placing a reference datum at MMC when the tolerance is shown in RFS.

Let's say, as an example, we have a profile of a surface tolerance of 0.2 mm referencing primary datum A, secondary datum B (RMB) and tertiary datum C (in MMB). What is the purpose of having a datum in MMB when one must actually measure the profile tolerance on a CMM? I know that it is legal (nothing in the standard stating otherwise) but what is the value or purpose?

I can see when one has a profile of a surface tolerance of 3 mm and the secondary and tertiary datums are at MMB, one can utilize a checking fixture with the outer and inner profile tolerance boundaries shown on a checking fixture but a tolerance of 0.2 mm??

In another situation, one has a positional tolerance of a diametrical tolerance zone of 0.3 in RFS while the secondary and tertiary datums are at MMB. Shouldn't the positional tolerance also be reflected at MMC or should both the positional and reference datums be reflected at RFS/RMB?

I have always been blown away with the application of MMB on the reference datums while the tolerance is in RFS but this may have something to do with my background in measuring rather than designing.

Maybe some of the Designers here could help me out on this one.


Dave D.

http://www.qmsi.ca

 

[drstrole]

fsincox - I agree, if that’s how it works put it on the drawing that way. When we try to second guess or make things easier, I have found that usually the opposite occurs.

Drstrole
GDTP - Senior Level

 

[fsincox]

I am not saying this type of discussion is not valuble for those of us who don't work with inspection. To provide education to us from more experianced members, but we are not re-writing the standard here, in this forum. Opinions are fine, and we all have them, but this thread in itself shows the danger of how the stated intent of the standard not to define methods and processes can be corrupted to the point of negating the standard, itself. This is one of the many problems I see with the current status of the standard (ideal) in the workplace (real).
Frank

 

[fsincox]

People rave about how much better the ASME standard is than ISO, because it uses gaging methods to describe the concepts, but here we have an example of how this has actually corrupted the discussion. Now we are being asked to agree we can't use MMB unless we are gaging parts.
Most of us seemed to agree, in other discussions here, that the use of actual "gages" is limited (I know the auto companies use them, they make millions of the same part!). I have the same problem with those who believe I should not be able to use concentricity, if you want to check it as runout OK but if it does not pass don't bother me until you check what I actually asked for, ISO coaxiallity.
Frank

 

[dingy2]

Hi Frank:

Good thoughts from a design perspective but placing a requirement that is beyond the scope of our present state of the art in industry doesn't help anyone. We can always say that we did our job and it does meet the standard and I agree that it does, but it still is "pie in the sky".

I guess that we could also have a positional tolerance of 0 in RFS referencing datums at MMB. The datum shift could give one the actual calculated tolerance if it could be performed. That, I believe, also meets that standard but is it practical??

Thanks for your comments and, obviously, other people agree with you on this one.

Dave D.

http://www.qmsi.ca

 

[fsincox]

Dave,
I mean noting against you personally. You are right I am looking from a design POV, I understand that. I also have tried for years to understand why the standards are not followed and I think this is an example of how it happens, people say what is the point if they can't use it anyway. I feel there is a greater purpose gained solely in being able to express what is needed and you saying: "I know what he wants, the best I can do is this" and we acccept it at that. That is the best we can do, currently. I also feel things change and to limit by what we can do now is narrow sighted.
Frank

 

[dingy2]

Hi Frank:

Thanks for your comment.

I have always stated that having MMB on datums meets the standard. It is certainly not wrong applying it just as it is not wrong utilizing concentricity. Is it desirable? practical in all applications?

I really hope your company encourages Design Review since a lot of my thoughts would come out there. Sometimes there are more than one way of achieving the same result but design intent and integrity is of the utmost importance.

I remember quoting to do training in one company that produced one off parts - atomic generators. The Engineering Manager said that he didn't want anyone to learn anything about MMC. Everything here is RFS. I think the Manager would have a heart attack if he ever saw MMC on a drawing. That was their culture as I am sure that your company has a design culture too.

The reason I started this thread was that I was at a ASQ meeting (American Society for Quality) recently and the topic was GD&T. I just wanted to see how the trainer presented the subject although I am certified in GD&T but never took a course. He used a canned computer files and I noticed some of the drawings had a Profile of a Surface with the primary, secondary datums referenced in RMB but the tertiary datum at MMB. I knew feature wasn't measurable as shown in the FCF.

I wanted to stimulate a discussion on applications of GD&T that are within the scope of the standard but could cause concern on the shop floor.

Thanks, again, for your input.

Dave D.

http://www.qmsi.ca

 

[MechNorth]

In the Tec-Ease material, we emphasize the legality of MMB in the profile DRF, but counsel not to do it because it isn't measurable. I may over-simplify it, but when pushed beyond the obvious, beyond what's documented in the standard, and beyond the understanding of the limitations of metrology, I'll point out that profile establishes a boundary in space in which you want your feature to reside ... if you add "datum shift", then you're allowing that surface to be someplace else ... is that your intent?

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

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[axym]

Hi All,

Interesting discussion. But I feel the need to comment on the statements that profile tolerances with MMB datum feature references aren't measurable and that their use should be questioned or discouraged. This, combined with other misleading statements like "we cannot measure the center of a hole that is not restrained", is sending the wrong message. The addition of an MMB reference does not instantly make gaging the only option - these things can all be measured. It may be difficult and time consuming to make the measurement, as is true for many geometric tolerances. But there is nothing that is fundamentally unmeasurable.

If it can measured with RMB references, it can be measured with MMB references. The difference is that the relationship between the datum features (and thus the toleranced feature) and the simulators (and thus the datum reference frame and tolerance zone) will not be fully constrained. This does not mean that the part is not restrained or that there is some sort of moving target or instability. It means that different "candidate" relationships are possible, and therefore that different candidate measured values are possible. Finding the best (i.e. smallest) measured value involves optimizing the relationship between the feature and the tolerance zone, within the applicable constraints. This is true for simpler characteristics like Flatness as well, it's just an easier optimization with the simpler geometry and lack of constraints.

Specifying a datum feature at MMB brings in the possibility of datum feature shift, in which the datum feature can shift (translate and/or rotate) on its simulator. Datum shift is easy to work with in a pass/fail hard gaging context, but it presents special challenges when numerical measurements are desired (especially if you want repeatability). Simply put, any degrees of freedom that are available to "jockey" the part on a pass/fail gage must be optimized to get the best possible measured value. This optimization can be done on a physical holding fixture by trial and error - jockeying the part around on the fixture elements to try to get the best measurement. Or it can be done mathematically (virtual jockeying ?), by collecting a cloud of points and fitting them to tolerance zones superimposed on the CAD model. For flat thin parts, transparent overlays can be used.

In any case, how far we go with the optimization determines how close we will get to the "ideal" smallest actual value. If the optimization is not complete, we will choose a candidate actual value that is larger than the optimal actual value. On the plus side, this is conservative. On the minus side, repeatability (and especially reproducibility) can go out the window.

This is one of the double-edged swords of geometric tolerancing, especially in Y14.5. Certain functionally-derived things (like datum feature shift) are very easy to make use of in a gaging context but can be very difficult (but not impossible) to deal with in a measurement context. Why do you think most GD&T textbooks emphasize the benefits of gaging? ;^)

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[dingy2]

I can measure anything!! Many years ago, I was a CMM Operator for GM but, at that time, we we called Precision Inspectors. I believe that I was the first person in Canada to open up an independent CMM service in the early 70s. Evan is correct, everything can be measured.

Can it be effectively measured? What confidence limit would one have in the results?

I remember having an Inspector, at one time, who would take out his pocket scale, look at the part, place the scale on it and determine that the feature was out of specification .010. Now as far as confidence in the result, that would be another topic.

Have any of the people in the forum ever taken a part to the CMM room for some sort of measurement. When you went back, the CMM Operator said the part was OK and then quickly scurried away? Why was that? Did you not brush or teeth this morning? It had nothing to do with you but the fact is that the CMM Operator did not want you to talk about his measuring method. He did not have 100% confidence in the result. As a matter of fact, maybe the confidence level was down to 50%. It was measured though.

Datums referenced at MMB are conducive to hard gauging. CMMs are, at this time (state of the art), are not that good at datum shifts. Yes, we could make a hard gauge (pin) inserted in the one hole and then start jockeying the part around to get the best fit using a CMM. Is it practical? No. Does it happen? No. Any time there is a datum referenced at MMC coupled with a Profile of a Surface, the CMM Operator sets up the part in RMB.

I will rephrase an earlier statement. I said "I knew feature wasn't measurable as shown in the FCF. " Instead, this statement should have read "I knew feature wasn't measurable from a practical or confidence perspective as shown in the FCF."

I agree with Jim Sykes on this one. While the application of MMB on datums where the feature is controlled by a profile of a surface is legal, I would also discourage it.

Dave D.

http://www.qmsi.ca

 

[pmarc]

Dave,
I do not buy it. No offence, but you sound like the only measurement equipment available for you is CMM.

MMB concept is to open the tolerances due to functional reasons and that is why it helps by nature - I believe we all agree on that.

But if somebody thinks or knows for sure that the profits of having it specified on the print do not outweigh the issues related to inspection (measurement method, availability of proper equipment, time needed for measurements etc.), it is probably better to leave them far away from the drawing. I think this is the point of Tec-Ease tip.

But one has to always remember that if he decide not to use MMB only due to inspection issues, some functional parts may be rejected - just because he wanted to make inspector's life easier.

 

[axym]

Dave,

You've been involved with metrology and CMM's for a lot of years, and I have as well. We both know what is possible, and what typically goes on. But I don't think it sends the right message to designers for us to point out certain Y14.5 characteristics and say "don't even think about specifying that even if it's functional, it's not going to be measured correctly anyway". If I were to list everything in Y14.5 that I thought was typically not measured in an optimal way in industry, it would be something like this:

-Size (Rule #1 is often ignored)
-Straightness (both Surface Line and Derived Median Line)
-Position of Slot/Slab Features
-Anything involving Simultaneous Requirements
-Anything involving partially constrained DRF's
-Anything involving datum features at MMB
-Anything involving LMC or LMB
-Anything involving threads
-Projected Tolerance Zone
-Concentricity and Symmetry (not that I'm suggesting that these be specified)

Does this mean that I should discourage designers from specifying any of the above? Stick to Flatness, Perpendicularity, Position with a fully constrained DRF, and the runout tolerances?

I agree with the main intent of what you are saying, maybe it's just the blanket statements that I object to. There are a lot of things in Y14.5 that lend themselves to gaging, and are very difficult to get reproducible and correct measurements for. Even when CMM's are used, the measurement results are often (usually?) questionable. No argument there. But there are places that have the personnel and equipment to do these measurements correctly, they're just the exception.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[dingy2]

Thanks Guys:

Yes, I was involved in metrology and CMMs for a lot of years. I have also used chucks or divider heads on small cylindrical features to measure circular (total runout is tough) runout. I have used height gauges and templates too. That was so many years ago but I have never forgotten.

Datums at MMB are vital when we are utilizing positional tolerances at MMC and other applications. It does simulate the actual assembly operations and can be verified with a hard gauge. As a matter of fact, all applications of MMC or MMB are applicable to hard gauging which has a high confidence level and is more appropriate than a CMM. I agree with Evan that some of the result from a CMM are suspect and a hard gauge supersedes the CMM if the results differ (as long as the gauge was made correctly).

Evan - In your list of other geometrical application that are difficult to understand, there are only two that I would agree are not utilized or understood on the shop floor but that is another thread. I don't think we are discussing difficulty in understanding GD&T here. It is the questionable application of MMB on datums when the tolerance is in RFS.

Having MMB on a datum when the feature is in RFS is legal. I think we can all agree on that.



Dave D.

http://www.qmsi.ca

 

[fsincox]

First, let me say thanks to Evan for his addition, here. It seemed like the standard was rewritten and I was not been informed of it.

Sorry, I feel you guys are being very convenient with your objections, when you do not also object when anything is not checked with a gage then. I receive CMM reports all the time with maybe 5 or 6 points checked for the flatness of a surface or 2 point size checks, no thought at all to an envelope, without anyone batting an eye. Then you expect me to believe a concentricity check requires two directly opposed points everywhere, I am sorry I really have to call that for hypocrisy I believe it to be.
Frank

 

[MechNorth]

Good discourse, guys. Nobody got (terribly) upset and the tangents were minimal. Particularly good that we got some good, realistic input from both design and inspection.

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

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[fsincox]

I am really not trying to be hard on anyone, either, particularly the person starting this discussion, but, I do want to take the opportunity to use this to show how the principals we are all taught to believe often get "tempered" shall we say. In the end sends mixed messages, at best, to those who would rather not be bothered with the whole GD&T issue to begin with, providing them a tool to use with management to circumvent the use of the same standard we all claim to support.
Frank

 

[axym]

Frank,

It was an interesting discussion, wasn't it? The original premise of the thread was to question and discourage the use of certain functional, legal and explicitly illustrated Y14.5 applications because measurement was perceived to be impractical. This is definitely an unusual issue to raise on a forum of this type. We usually bash bad drawings, nonfunctional specs, etc.

I agree that this type of argument can easily (and often is) extended to apply to all of GD&T. I have heard it before, from companies that want to avoid the hassle of GD&T and stick with plus/minus tolerancing. "The vendors can't measure it, so what good is it to put it on our drawings?".

If we choose to specify characteristics that can be meaningfully applied to real-life geometry to control functional relationships, these characteristics are going to get complicated. This is because real-life geometry and functional relationships are complicated! If actual values are required, measuring these (and even defining them in the first place) is going to be complicated. This is part of the deal with GD&T.

I suppose that the question comes what should be done if the functional characteristic is deemed to be impractical to measure. If the characteristic needs to be simplified in order to make measurement practical, at what stage should that simplification occur? On the drawing or in the quality plan? Do we specify functional characteristics or do we specify simpler (and necessarily tighter) characteristics for ease of inspection? I suspect that there will be differing opinions on this.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[fsincox]

I agree, and I feel designers should be state the assembly condition and comprmises can be made as they must be, but the actual condition is preserved. If we all understood GD&T then it, I believe,it would not be as bad it is the fact that many do not that makes everyone want to simplifiy (idealistic POV).
Frank

 

[dingy2]

Evan:

I agree that this discussion was interesting but for other reasons. I asked from a design perspective why one would place a MMB on a datum when the feature is in RFS. I stated that it could not be checked but after this discussion, It can confirmed although quite difficult and in most cases, time consuming.

The answer that I received was that it was legal and it does simulate the actual assembly process. One can performed this requirement on a CMM with a combination of a checking fixture for the datum structure and multiple measurements on the feature using the CMM probe. It can be done. Of course, a checking fixture is best and quickest but it only is a 2 dimensional approach which profile of a surface is, usually, 3 dimensional.

I certainly have never seen an example of this in Y14.5-2009 or earlier edition where one has the tolerance in RFS and the datums in RMB. If there is an example, please let me know where it is in the standard.

I never did suggest that we get into +/- tolerances and I don't know where that thought came from. People seem offended that I asked this question. I don't know why. I think it is valid to ask about the applications of GD&T and the rational behind the application. Any of my queries would have surfaced in any design review if your company promotes that structure.

You were, at one time, programming measurements or layout on parts using a CMM. I wonder how you approached this situation? Could share your experience with us? Did you set it up with the hard gauge on the datums at MMB and then take multiple measurements and report the best condition after jockeying the part and the datums? Did you disregard the MMB on the secondary and tertiary datums and apply RMB?

The ASME standard is quite explicit at times and in other situations, conflicting. We may have visual examples that don't match the verbiage or in some cases left overs from previous editions that possibly should have been deleted but that is another thread. I never suggested that we go off standard and always promoted that the design integrity.

I think I will approach datums at MMB when the feature is toleranced in RFS that same way that I have approached concentricity in the past.

Thanks for all the input.


Dave D.

http://www.qmsi.ca

 

[dingy2]

J-P:

I gave you a star for your previous comment about using 4 corner holes as a datum at MMB. In your example, we now have a Profile of a Surface relative to the datum structure at MMB.

To confirm this requirement, one would need a checking fixture for the 4 datum holes. We would end up with plate with 4 pins on true position at MMB size. Secure the fixture on a CMM and then place the part on the fixture over the pins into the pattern of 4 holes. Secure the part on the fixture and then, using your CMM probe, zero off (obtain the center) of 2 of the pins on the fixture. Sweep the surface from the datum structure with the probe which mean taking over 20 hits. Record the both extremities of the readings.

Move the part on the fixture, secure it, sweep the surface again and record your high and low results. We can do this numerous times and the actual result reported would be the best result obtained from all the measurements and all the locations with the datum shift.

We would have great difficulty setting up in RMB on the 4 holes and would, probably, end up using only 2 of the holes. This is wrong. We need to use all 4 holes since the full pattern is the datum rather than only 2 holes in the pattern.

Using MMB in this situation would simulate assembly, best method on confirming the requirement and it is not off standard. We even accrue bonus tolerances from the datum shift. It may be costly in the case of additional checking fixture and time to confirm but I would recommend this approach in your example.

Thanks for your comment.

Dave D.

http://www.qmsi.ca

 

[axym]

Dave,

I don't know if I would say that people were offended that you asked this question. We're all friends here, I think, and we can have our reasoning strongly questioned without taking offense. You and I have locked horns on quite a few issues in the past, and we often agree to disagree. Perhaps your original question didn't come across as just an inquiry from a design perspective. It came across more as "Why would you specify that if we can't measure it?". The designers reacted with "We would specify that because it's what the function requires" and I reacted with "We can measure that, it's just difficult". I don't know of any examples in the standard that have the considered feature RFS and the datum features RMB - there may not be any.

When I was first starting out with CMM programming years ago, I was lucky enough to be using software that handled MMB datum features properly and did the required optimization. This wasn't the software that came with the CMM, it was a third-party off-line package (Valisys). It was great - the software would graphically display the tolerance zone and deviations and report a transformation matrix when it did the datum shift optimization. I learned a ton about GD&T by using this software. In later years, I worked at other companies that had simpler CMM software and it seemed like a step backward. I didn't do much jockeying on hard gages - I would usually just program as if the datum features were at RMB and deal with the occasional rejection of a good part. The datum features typically had fairly tight size tolerances, so the magnitude of the datum feature shift was generally not very significant. But I didn't suggest to the designers that they change to RMB references because I couldn't measure MMB ;^).

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca