can a 90 degree feature be used as a datum? 6

[tk369]

I have a thermal strap, in 2 halves, which straddles a bellows fixture. There is an r51.5 mm feature which provides clearance around the bellows. the (2) halves make up a 180 degree segment, meaning each half is 90 degrees. i am detailing 1 segment, with the other half opposite hand. the drawing is attached as is.

The flat which bolts up to the bellows is -A-. i think the r51.5 should be -B-, with a size tolerance, perp to -a- so it locates the bolt circle pattern. As it is, what is the relationship?

Some say I can't use the arc feature as -B- because is's not a full hole. I say with a CMM, you can find the center axis and verify the size of it, and use it to locate the bolt holes.

Any suggestions? Other than dodge the bullets?

teddykaye

 

[pmarc]

Jim,

I am not a linguistic expert (especially that English is not my native language), my knowledge of geometry is quite good, but in this case I understand that users could misunderstood intentions of committee members.
I fully agree with you that reading more than was intended could lead into blind alley, but you must also admit that this is very often caused by insufficient precision of definitions or descriptions. I think in our FOS example simple sentence like e.g. 'Feature of size must have opposite elements' added to the definition would solve all the dilemmas.
And the last thought - drafters/designers also have some intentions when they specify GD&T on a drawing. If they do it correct, readers will not have any problems to figure out what they meant, but if they do something wrong or inprecise, probably only the author of the drawing will fully know what idea stands behind it.

 

[ewh]

I know better than to get into this, but a from a "cylindrical" surface, you can get two series of three points (if taken on the same planes) to define two arc segments which can then define a centerline. A complete cylinder is not required to achieve this. Granted, you can't use calipers on two opposing points to get this info, but this info is inherent in the surface.

"Good to know you got shoes to wear when you find the floor." - [small]Robert Hunter[/small]

 

[KENAT]

If it said 'surface of a cylinder' or similar I wouldn't argue. However, saying 'cylindrical surface' to me introduces an ambiguity, intentional or otherwise, which might allow use of a partial cylinder. Or at least explains why I and a few others might might percieve it's allowed.

Posting guidelines faq731-376

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

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

 

[fsincox]

Jim,
"A feature of size requires directly opposed points; pre-94, I believe that the caliper rule was shown."

I am not sure what you mean pre-94, the '82 definition was:
"1.3.8 Feature of Size. One cylindrical or spherical surface, or a set of two plane parallel surfaces, each which is associated with a size dimension", no opposed, no calipers.
Frank

 

[tk369]

ewh!
Someone else has the same perspective as i do regarding this surface.

Using a CMM, 3 points taken on the same plane gives you a center point. Another 3 points at a different elevation yields another centerpoint. These (2) centerpoints give me an axis from which the bolthole positions can be inspected.

No offense to the GDTP-S experts, but IMHO, the committee has to take computerized inspection and measurement into account regarding the standard. We are in the digital age and our thought processes should align themselves with it. Mentioning the '82 standard, at this point in time, shows that we are not keeping up and we have, more or less, rubber-stamped portions of the standard. Perhaps this is a portion the committee haggled about and let it fly.

I am forging ahead. Thank you all for your valuable input.



teddykaye

 

[dingy2]

teddykaye:

You only take 3 points on radius surface to find the centre point?? Would there not be a chance for error with this limited number of contacts? Why not take 8 or 10 points? Just asking.

By the way, I do agree that the axis can be developed this way but I question you limited number of contacts.

Dave D.

http://www.qmsi.ca

 

[Tunalover]

I've seen an internal cone e.g. 90deg csk used as a datum but never an external 90deg feature.


Tunalover

 

[rgb210]

Repeatability would would be the key issue in using the 90 degree cylindrical surface as a datum. Finding the center of a partial arc of less than 180 degrees is much less reliable than finding the center of an arc greater than 180 deg. The CMM could find the axis with 6 points, but if you checked the finished part several times, it would probably find a different axis each time. For that reason, I don't consider it the best choice for a datum.

Robert Bohot
GDTP-S

 

[KENAT]

Best choice, certainly not. I agree that even if you accept it's not forbidden then it's not a first choice option and there would have to be a fairly compelling functional reason to choose it.

Posting guidelines faq731-376

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

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

 

[MechNorth]

Reality of physical geometry supports the use of opposed points vs random points.

Per Y14.5-94, Section 4.5 Establishing Datums, (4.5.3 and others) "A machine element that is variable in size (such as a chuck, mandrel, vise, or centering device) is used to simulate a true geometric counterpart of the feature and to establish the datum axis or center plane." For a cylindrical feature of size, the true geometric counterpart (TGC) is a perfect cylinder; if the datum feature of size is referenced RFS, then you find the largest inscribed cylinder which makes maximum contact with the wall of the hole. This is repeatable.

While understanding the industry norm of using CMMs of various types, the fact remains that there is a greater inherent error in a CMM-established datum from the raw feature than there is by using a TGC at least as the basis of establishing a CMM datum.

Anyone that has ever done precision metrology on radial elements has seen that a seemingly "perfect" radial segment in fact is composed of multiple radial segments. Using the 3-point method (or 6 or 10, or whatever) to find the center of the arc means that you will end up with multiple arc centers and radial measuremnts. Which, then, do you select? The one closest to what you want, regardless if it is representative of the majority of the feature? Then, add a second set of 3 points to establish the axis of the "cylinder"?
I've tried to illustrate this in the file at the end of this link:

http://www.profileservices.ca/files/tidbits/thd1103_267281.pdf

I was at the knife-point of a project where this was a critical factor, costing significant $$s.

I don't dispute that CMMs are useful tools, even necessary in many cases, however I do get rather frustrated at the CMM salesman's mentality / pitch that you just need a few points to simulate a surface adequately to represent its functionality ... and THAT is what ASME GD&T is intended to do ... represent the design intent / functionality. There is considerable ongoing dialog on this site & in other venues as to whether the design intent or manufacturing process or the inspection process is to be predominant in the GD&T application. The standard specifically talks about engineering and omits reference to manufacturing and inspection. Perhaps, in that light, metrology should be focusing on how to achieve the intentions of the design documentation as established with GD&T rather than expecting GD&T to accommodate the inconsistencies of metrology. I recognize that the two aspects are symbiotic in nature, but within the ASME standard, design is paramount. ISO offers the opposite, a GD&T standard that focuses on inspection rather than design intent.

The fact that a center point or axis established by a radial segment without directly opposed points is unrepeatable should guide people to recognize that it is a fundamentally flawed process as far as ensuring design intent. As with any written work (standard or literary), the text itself gives you a substantial part of the knowledge, but you have to extend, combine and contrast ideas to get a full understanding of the content.

OK, I've beaten my head enough for today.

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

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[dingy2]

Jim:

You stated about the use of a CMM "Using the 3-point method (or 6 or 10, or whatever) to find the center of the arc means that you will end up with multiple arc centers and radial measuremnts. Which, then, do you select?"

Most modern computer programmes on a CMM automatically select the "best fit" centre. It is not up to the CMM Operator to say "I like this centre" since I am hoping the product is non-conforming." The higher number of contacts, the better the centre but it is the programme that selects the centre. The calculated arc and centre are developed using the mid-point of all the contacts rather than the theoretical true geometrical couterpart.

I do agree with you that a CMM has error and the repeatability (and reproducibility) would be suspect unless one contact on the exact points which is unlikely. The error would certainly be reduced if one had to find the centre of a hole (feature of size) but, again, more than 3 points should be taken.

You are correct that the standard does not cover manufacturing, inspection or, in fact, suitable applications for GD&T. It is up to the Designer to apply GD&T primarily based on the part's "function and mating relationship" being aware manufacturing capabilities.

Just putting in my 2 cents worth.

Dave D.

http://www.qmsi.ca

 

[MechNorth]

Agreed for the most part, Dave, but the operator DOES select the points (or sampling routine), so in effect the operator does select which radial segment is being used. Then, the operator also selects the fitting algorithm that processes the data points, again deciding the outcome.

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

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com