concentricity vs. the cmm (concentricity is evil) 2

[joebk]

I have been arguing with the QA folks about what the ASME Y14.5M-1994 standard calls concentricity and what they are checking.

As far as I am concerned if concenctricity is called out between two cylindrical features, the median points of all correspondingly located elements have to fall within the cylindrical tolerance zone (more or less taken straight from the ASME standard). So if concentricity is specified, the median points of the entire surface in question must be generated and compared to the datum axis to ensure they all fall within the tolerance zone (as I understand it).

The issue is that our QA folks seem to have a different definition for concentricity (note that company policy is ASME Y14.5M-1994 as is stated on our drawings). When they check it on the CMM, they only check it from circle to circle in a single cross section and not over the entire surface. To make matters worse, the manual for the CMM software actually states "Use only a circle or arc. Concentricity tolerances do not apply to other types of features" i.e. cylinders. The software gives an error when attempts are made to check concentricity using cylinders.

Now it has degenerated to "this is the way we have always done it and this is the way everyone in quality control checks concentricity". I hate statements like that unless there is something backing it up, but I also know how hard it is to change things like this in manufacturing.

The problem is that the part rotates at high speed - balance is an issue - and our customer is having problems (and is disputing our concentricity methodology).

Please don't get me wrong, I am not in love with concentricity - far from it. It is a royal pain and in my humble opinion is rarely (if ever) needed and is confusing and this particular case is no exception. The surfaces should most likely be controlled by either runout or total runout (maybe a combination of the two) and a balance spec should be added to the drawing.

So am I way off base or what? Has anyone ever had to deal with concentricity and QA procedures / CMM limitations? I am probably going to change the drawing as indicated above but any guidance or experience with similar situations would be appreciated. Sorry for the rant and thanks for your patience.

JBK PE

 

[TheTick]

Maybe use runout? Usually folks specify concentricity when runout will do.

Honesty may be the best policy, but insanity is a better defense.

http://www.EsoxRepublic.com

-SolidWorks API VB programming help

 

[sbozy25]

Yeah runout should suffice nicely.

The problem you are having with the CMM is a common one. The way they like to do it is create your two circles and generate an artificial cylindrical zone. There logic is that if the two ends are good, the whole part is good. Well as we all know that is not the case. If you change it to runout they will be able to use a dial indicator instead of a CMM. Everyone wins.

 

[dingy2]

You really didn't state the full definition of concentricity. Yes, it is median points of diametrically opposed elements and that is pretty hard to do on a CMM. I am talking about the diametrically opposed elements part. Having a pimple on one side and a flat on the other really shifts the median point. Roundness of the feature is now involved.

I bet that the guys in QA are taking many points on the datum to create a centre line. They are also taking many points on the feature in 1 slice perpendicular to the axis and then finding the centre of the feature and compare centre line.

Concentricity can be taken on any slice of the feature perpendicular to the axis. If QA people checked in a couple of areas (slices) on the feature, they should report the worst case.

That is NOT concentricity but positional in a RFS mode. They did not take median points from diametrical opposed elements and, frankly, they probably can't.

Change the concentricity to circular runout and it should be relatively easy to do but circular runout cannot be performed on a CMM. It should be confirmed with a divider head (chuck) on a granite table with a digital indicator and stand. Yup - no CMM here.

If it is required to check the full feature since balance is important, then use total runout.

What they are doing in QA is the best they can do. Here is something else one could do. If you see a concentricty call out of diametrical tolerance zone of 0.4 mm., it means that the median point(s) cannot be off centre over 0.2 mm. They could change this to a circular runout of 0.4 and it would be about a close as one will get but not perfect. We still didn't get diametrically opposed elements.

I wish that ALL Designers would replace concentricity with circular runout. You achieve what you want and it can be confirmed correctly.

Dave D.

http://www.qmsi.ca

 

[powerhound]

I've had numerous dealings with concentricity vs. position or runout, etc. and one thread on this page caused me to delve more into it and from what I've learned, there is no substitute for concentricity when balance at high RPM is a factor. I'll have to give more thought to what Dave said about circular runout being sufficient. I may have just learned something new.
Personally, I have never, in 18 years of being a machinist, CNC Programmer, and Draftsman (mostly all at the same time), seen a case where concentricity was called out appropriately. Every time I've seen it, it was because the designer wanted the features to be "concentric". Position or Runout has always been the more appropriate callout. The concept of Actual Mating Envelope is foreign to those who think that GD&T is simply knowing what the symbols mean.
To sum this up joebk, your customer is correct to dispute your QC's method.
Try this:
Concentricity is called out with respect to a datum axis, so use the CMM to create the datum axis through your datum feature. Now, touch off one wall of your feature being controlled and then again at another point 180 degrees revolved around the datum axis. The center of those two points should fall within your diametric tolerance zone. This procedure is repeated however many times your QC department personnel see fit. The points should not be taken along the same plane or Z level every time. Every median point between the 2 touch offs at 180 degrees separation should fall within the tolerance zone.
Your customer seems to be one of the few that actually knows what concentricity really means (probably because of their application) and your QC department is probably typical of most. You have the right idea regarding the true meaning of concentricity but I don't know how much I'd argue with QC. In my experience, they are the "know all to end all".

Powerhound
Production Supervisor
Inventor 11
Mastercam X
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[434343434342]

How about a CMM with a scanning probe? What is an acceptable point sampling density required for runout or circularity?

 

[dingy2]

43::

Scanning probes are great for profiles, cylindricity, flatness and wild curves. As far as circular or total runout ----mmmmmm. A divider head (chuck) and an indicator are best but still tricky to do.

Powerhound:

Looking at your experience as a Machinist, performing (circular or total) runout with a divider head and indicator would entail infinite number of contacts around the feature. While the method you used for concentricity is correct, we would not achieve infinite contact as we would with the runouts.


Dave D.

http://www.qmsi.ca

 

[powerhound]

Dave,
I understand that point completely and I believe it actually exposes an inherent flaw in the concentricity callout. Obviously, touching an infinite number of points is unfeasible and probably unnecessary. I didn't mean to imply that that was what needed to happen, but where the customer wants concentricity checked as correctly as possible, that's where it starts. The actual number of points acquired would be an issue between QC and the customer.
As I've pondered the circular runout vs. concentricity issue, I think the biggest difference between the two is that runout controls form while concentricity doesn't. That may be moot but there may be some use somewhere out there in fantasy land for controlling centerlines without controlling form.
I sure hope this isn't a hijacking...I don't mean for it to be.

Powerhound
Production Supervisor
Inventor 11
Mastercam X
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[joebk]

Thanks for all of the feedback everyone!

The situation is crazy because our customer calls out concentricity at TIR on the drawing!? So in my opinion they are actually asking for total runout. When they discuss this with us, they are quoting numbers for "full indicator movement". So much for concentricity. Sorry, I should have mentioned this from the get-go.

Our QA department checks concentricity by comparing the center of two circles (on the CMM) and nothing more. They have refused to even entertain any other method so I think this is going up the food chain. I think it is tough to be 2D in a 3D world.

I have to admit that I tend to be relatively pig-headed on stuff like this, so I think QA and engineering are not so different. The key in our case is that we define the ASME standard as the controlling standard for GD&T but our QA folks tend to ignore the standard and operate under historical practice (i.e. "this is the way we have always done it"). So as an engineer, I can't be certain that what I have put on a drawing is what is being produced. This is the real problem.

You did hit the nail on the head, when we have a customer who is knowledgeable on the ASME Y14.5 standard, this could really bite us.

Since I am fortunate enough to be able to make some decisions I have decided to outlaw concentricity (and symmetry for that matter) for any drawings produced by our engineering department.

If we have to use a customer drawing that includes concentricity, it is now our policy to discuss the specification with the customer so that we understand exactly what is meant on the drawing.

I agree that concentricity is most appropriate when balance is an issue but it is so misunderstood that I think it is better left off drawings. I have seen drawings of sheet metal parts with concentricity on them, I assume to maintain alignment between two holes. This is nonsense.

Since we have the means to dynamically balance these parts, the drawings are to include balance notes when balance is an issue. So balance is controlled by the balancing process and form/size/etc are controlled by dimensions/tolerances/notes/etc.

The drawing in question has been changed to replace concentricity with total runout and a balance note has been added. I think this has cleared up the concentricity issue (at least for a little while).

The customer seems happy with this, and that is really the most important thing.

Out of the strain of doing, and into the peace of the done!

JBK PE (and all around pain in the butt)

 

[PaulJackson]

Since so many are disenchanted with the current definition and proper measurement of ASME concentricity, I thought that I would throw my two cents in on substitute callout considerations.

Concentricity as defined in the 1982 standard and prior controlled the “central tendency” of a feature without constraining its form to a…circle, cylinder, or any other symmetrically circumferential surface! Its functional use could control the location of the multi-lobbed cam surface on the distributor shaft that fired the points on my 67 Camero RS and without skipping a beat it could tolerance the location of the odd-toothed driven gear at the opposite end. My point (pardon the puns) is that Concentricity WAS not EVIL for those that applied it to ODD shapes. Its shortcoming was in the definition of how that measurement of concentricity was accomplished.

The new definition of concentricity in ASME Y14.5M-1994 (codified in the mathematical definition of Y14.5, Y14.5.1) attempts to do the same thing but falls short in the ODD shapes department. By tolerancing median points of diametrically opposed elements of the surface can be any size and shape, so long as it is uniform, so long as it is central “having median points within the zone” and since 1994 so long as its uniform shape is divisible by an even number.

I’ll agree with others that recommend position RFS is probably the best substitute for most applications of concentricity but the choice should be driven primarily by the feature’s functional liabilities to variation. Where a feature’s form is circular or cylindrical without interruption and form is equally important as location I would say that runout or total runout should have been specified originally but if the shape is not continuously circular or its shape is not as important as the “central tendency” then runout is not a functional choice.

Position RFS carries with it baggage as well that doesn’t reflect the liberties that concentricity originally entitled. Figuring the axis or median planes of a true geometric counterpart reflects contact with the surface only at its extremities. True geometric counterparts don’t have to be round or cylindrical they can reflect any specified geometric shape but in order to construct that true geometric counterpart one would have to assume the feature’s specified form constraints independent of size. Form restrictions and extremity location are characteristics that neither the new nor old definition of concentricity includes.

So what is the best substitute for the “evil” concentricity? That depends, sometimes there is no better substitute, sometimes runout, and sometimes position RFS. If the standards guys would have left it alone last time then what the inspectors typically do to check both position RFS and concentricity with a CMM (a circular least squares regression) would still be the best estimate for both with that instrument.

Paul F. Jackson

 

[TheTick]

I went through this with a client a couple months ago. They put concentricity on a drawing. I changed it to runout and explained why. They changed it back to concentricity, then wrote a special work instruction explaining how concentricity was to be measured. The explanation was a near-textbook measurement of runout!

 

[PaulJackson]

Tick, What exactly was their special work instruction?

 

[joebk]

I have basically the same problem. Customer calls out concentricity at FIM. When customer inspects, our part is placed on a mandrel between centers and rotated. An indicator placed on the surface in question is monitored for FIM, this reading is used to qualify this feature. So it is circular runout.

We can deal with differences between the standard and what the customer wants but the problem is that on their drawing they invoke the ASME Y14.5-1994 standard (we do as well). In addition, in emails to us they refer to the standard and the definition for concentricity (they do state the definition for concentricity correctly to the standard).

The worst part - our customer is rejecting parts because they do not conform to the runout/concentricity specification and nobody is willing to change to runout to more accurately reflect what is expected.

Our QA department inspects position at RFS on the CMM and refuses to inspect runout - this is not an exaggeration. QA is also hotly debating the definition of concentricity stating "this is the way we have always done it, why change now?". They are currently in a pissing contest with our customer concerning this issue. It is nuts.

Our sales and manufacturing departments are big time against runout because they know we cannot produce parts that meet the customer's spec using current manufacturing methods (without increasing the cost of the part).

So I stuck between a rock and a hard place, it is a no-win situation. Mainly because of confusion in regards to concentricity and a lot of bad decisions and flawed logic.

I proposed changing the drawing to runout and increasing the price to compensate for increased costs but this is not acceptable. So we continue without resolution.

 

[TheTick]

PaulJackson:
per what joebk said

 

[dingy2]

The QA department is checking positional in RFS rather than runout. Runout cannot be checked on CMM but, as I stated before, in a divider head.

What you describe from the customer is definitely circular runout.

Good luck in your rock and a hard place. Been there, done that and don't like it.

Dave D.

http://www.qmsi.ca

 

[PaulJackson]

Tick, I understand and agree that your customer is describing runout when their "special work instruction" says to do “full indicator movement” as joebk replied. I was just curious, since the thread is titled "Concentricity vs. CMM...,” what their "special work instruction" for concentricity using the CMM actually was.

Where I do agree that doing the ASME …1994 estimation of concentricity on a CMM is difficult because the midpoints must be derived from diametrically opposed points on the surface from the DRF axis, both runout and total runout can be estimated fairly well on a CMM. It’s done all the time. So long as the software is written (or the programmer writes the routine) to examine the difference between the minimum and maximum radius values from the surface points in each perpendicular cross-section for runout, or between the Min/Max radii from all cross-sections for total runout. Naturally as the number of points examined increases the estimate improves.

I doubt that any of the modern CMM software uses the doubled differences between the resolved centroids of the “best-fit” circles and the DRF axis to estimate runout as is done with position RFS. That routine assumes that the surface is round! Some of the best software with DCC driven axes and scanning probes use continuous circirlar or spiral scans with thousands of points to estimate runout and total runout respectively.


Dingy2, I have to disagree with you that runout or total runout cannot be estimated using a CMM. As with all inspections and inspection equipment assumptions are made and error exists. Chucks are assumed to be oriented and located coaxial with the bearing surfaces…they are not. Mag surfaces are assumed to be square with bearing axes … they are not. You pay for what you get with bearing accuracy… roundness machines have air-bearing and swash-plate tables that generally provide the best precision and accuracy but seldom is that kind of precision required for runout checks. Inspection from centers is usually the cheapest and most repeatable way to check runout but assumptions are made that the adjacent journals that were ground from those centers are coaxial with them …they are not.

You can say that there is greater error in establishing an axis from geometric form regression equations applied to datum feature surfaces on a CMM but one always has to pay attention to the margin of their assumptions in relation to the tolerance specified. Volumetric accuracy, apparent datum feature form errors, point saturation, and tangent contact error of datum features as well as spindle accuracy, physical tooling orientations/locations, instrument rigidity, and trace abbreviations all have to be “checked at the door” to make reliable measurement estimations. No matter whether the measurement is accomplished with physical observations relative to artifacts or mathematical assumptions applied to data gathered from artifacts, inspection involves uncertainty. One is not better than the other they are just result in better estimations in the hands of skilled inspectors!

Paul F. Jackson

 

[dingy2]

Paul Jackson:

The word "cannot" that I used is too conclusive when expressing that the two (2) runouts cannot be checked on CMM. We can check anything with anything but it is the amount of error that we experience. I call this confidence in the outcome. Both the divider head and the CMM have error built in and then we have skill and repeatability/reproducibility.

The software that you expressed sounds wonderful but most companies have only limited scanning capabilities, if any, on their CMM. Some older CMMs still around do not even have computer capabilities.

Both circular and total runout are conducive to small cylindrical features that share the same centres and are best checked on a divider head or chuck.

I really don't want you to influence Designers that they should use the runout everywhere.

I remember one Customer who used circular runout on a counter bore relative to the bore. I knew the CMM Operator would have a problem so I asked the lady how she confirmed the requirement. She said that she didn't. Here, the more appropriate geometrical symbol should have been positional at MMC relative to the datum at MMC. A checking fixture would have been ideal.

Dave D.

http://www.qmsi.ca

 

[PaulJackson]

Thanks for the advice Dave.

In response to:
I really don't want you to influence Designers that they should use the runout everywhere.

One of the Y14.5 committee members asked, Should designers consider inspection methods? I responded...

http://tech.groups.yahoo.com/group/Y14-5_User_Group/message/1014

Paul