TWO HOLE DATUM FOR ATTRIBUTE GAGE

[Wicsteve]

We’re building an attribute gage for production and incoming inspection. Two holes on the part are selected as a datum. There is both tolerance on the hole sizes and positional tolerance on the two holes. Should the gage pins for these two holes be sized only for variation in the hole size or a combination of hole size tolerance AND positional tolerance. … if another feature control frames references this datum, is the positional tolerance for these two datum holes to be taken into account?

 

[HGMorgan]

Gentlemen,
A basic alignment and measurement by CMM would not be able to take into account a "shift" factor, how would it know which way to "shift"?

This is why Best-Fit was created, and, why most CMM software includes Best-fit. If not, Best-Fit can be an add-on package.

All the Best-Fit packages we have seen and used allow operator control on XYZ rotation & XYZ shift. Thus the MMC value can be input. This allows a CMM to be a "virtual floating check fixture", that will reject bad product and accept good product.

 

[dingy2]

HGMorgan:

Yes the best fit packages allow the Operator to take many readings on a diameter and the CMM will come up with the best fit diameter discarding the largest and smallest readings. It can also calculate the tolerance using MMC on the feature BUT allowing a shift on a datum at MMC???

We could, in one case, have a datum shift tolerance between the actual size and its MMC size while the other datum feature could be between its actual and virtual condition size. If the software allows one to rotate and shift the part manually, I love to watch a CMM Operator explain how one shifts the datum to correspond with the requirement.

J-P

Using a CMM and having the datum holes at RFS while the drawing states MMC does not correspond to the requirement and the part is not inspected correctly.

Dave D.

http://www.qmsi.ca

 

[HGMorgan]

dingy2:
The Best-Fit process is not taking several readings on one feature (Dia.) and then discarding largest and smallest readings. Best-Fit is taking readings on ALL the product features and then letting the software shift them to the closest average values from nonminal.

The Best-Fit process:
1. Align product per Datums
2. Survey (take points on) ALL features.
3. Run Best-Fit by setting the limits on XYZ rotation & XYZ shift per Eng Dwg tolerances. This creates "Soft-Gage" zones.

For this example:
XYZ Rotation: -A-(Z) would be allowed to rotate with-in the FLAT 0.4 zone. B (XY)would have free rotation, the two holes are going to lock in the angle.

XYZ Shift: -A-(Z) would be locked, B (XY) would be allowed to shift TP Dia. and any MMC bonus obtained from the survey of the actual hole diameter.

This is "how one shifts the datum to correspond with the requirement".

On our software, this would be set-up one time, and then ran on all the separate products without any intervention from operator.

A "Virtual Floating Check Fixture" that buys the good and rejects the bad. This does work, and the need for it is why Best-Fit was created.

 

[restedup]

Wicsteve,

It seems unlikely that you would want the positional tolerance of the send of the tube to apply for the entire straight portion. Hence the question that I had posed earlier, with regards

 

[Wicsteve]

RESTEDUP

Background (or more than you really want to know- Extra credit reading) -- The application is for a small engine assembly. The tube end passes into an extruded hole of an exhaust muffler. In reality, about 12 mm of the tube end actually is engaged, even though our print calls for checking 25.4 mm of length.

We had an issue in our customer's assembly plant because the manifold is first assembled to the engine and tightened down prior to final assembly of the muffler. There appears to be too much allowable adjustment or swing in the manifold assembly due to the loose fitting flange holes and the assembly plant has difficulty in centering the tube. The customer doesn't have an attribute gage for this part but measured the part on their CMM using the upper flange hole as the origin. They declared the part out of tolerance even though it passed our attribute gage. We then got into a discussion with our QC people who checked the part on our CCM also using the upper flange hole as the origin for measurement. So went the discussion with multiple opinions as to which is correct, gage or CMM.

This whole posting was started to resolve the question of part measurement and to understand the issues between CMM measurements (for ISIR, PPAP) and the attribute gages. We normally supply our customer and our factory with CMM data at the startup of a program but check production parts with an attribute gage. I’m afraid that engineering, manufacturing, and QC aren’t sledding on the same hill in this regard. Also note that many of our manifold have two flanges (for twin cylinder engines). Common practice is to use attribute gages to check (all four holes are used). Yes, I know that only two holes should be used for a datum but the four hole method is sort of self-centering and more practical because “it’s how the part sees the engine”.

By the way, we probably will make the flange holes smaller to reduce the allowable swing. ... or the manifold might need to be fixtured while being assembled to the engine. The latter is more time consuming and harder to accomplish on our customer’s assembly line. The assembly plant has resolved the issue in part, by re-centering the tube to the midpoint of allowable swing before assembling the muffler (usually by the means of a swift stoke with a hammer). We're still working out a solution with our customer.

Finally, when this print was made we were attempting to comply with ASME 14.5, 1994 standard.

 

[restedup]

Wicsteve,

Tightening the holes would not seem to help the problem. I would think that to open them would aid the fit of the tube into the exhaust muffler. The datum identifiers with the dashes are not in compliance with 1994.

I would think it would help also to limit the extent of the gage for the tube end to cover the engaged portion only which can be done by limiting the length. (Chain Line comes to mind.)

To my knowledge not stated in any standard, but MMC works well with atribute gages whereas RFS is better suited for CMM.

Dave moderately hit on this idea in his post.

 

[Belanger]

As long as we're talking about the 1994 standard, it might be a good idea to use the "boundary" notation for the position tolerance on the tube.


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

 

[dingy2]

J-P:

The tube end is round and "BOUNDARY" is not required using the 94 standard. This termed is used on non-round features of size when reflected in positional at MMC.

Dave D.

http://www.qmsi.ca

 

[pmarc]

Wicsteve,

- I am just curious - you said that your customer declaired part of out the tolerance, but which tolerance from the drawing was actually out of spec?

- If I understood you well, normally you are going to use hard gage simulators for specified datum features to fix the part during the inspection, am I right? If yes my question is how are you going to assure that the part will not wobble when put on planar datum A?

J-P, Dave,
I think a good example of dealing with such tubular elements is again shown on Tec-Ease site:

http://tec-ease.com/gdt-tips-view.php?q=128

 

[Wicsteve]

Getting off the subject
Small holes -- The problem at our customer's assembly line is that when the flange bolts are tightened with an air wrench, the flange tends to rotate about the stud center (more or less). Making the holes smaller would, to some degree, limit the the amount of rotational movement possible. The manifold apparanently is not fixtured when assembled and can move. I know that the operator does, by hand, position the tube to some extent but can't hold it from shifting about when the nuts are tightened.

Datum identifier -- You know, I'm still inserting dashes with the datum identifier. Never noticed that they aren't to be there. Thanks, as I'd have gone on for years without making the change.

 

[Belanger]

Right Dave -- thanks for the reminder! MMC equates to boundary to begin with.



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