Angle of Cone as a Datum

[Sora5]

Would appreciate any input on how to interpret and how to have PCDMIS interpret the following position callout on the inside sphere per ASME. A is the face of the part, which is fine. But B is the angle of the taper? with MMB?

Which would be the most accurate way for PCDMIS to dimension what's intended:

Autocone and use the cone as the datum -- easiest
Constructed circle intersected at the gage point
Constructed midline between two tangent lines somehow simulating the angle of the cone? MMB how?

Angle_Datum.jpg

[louisd]

Welp, is your diameter a basic, or is your offset distance from A plane basic for your gage point? Relevant info was cropped from pics.
Also is the gage point defined or linked to the definition as B datum?

[louisd]

Knowing you have a basic offset defined along the axis, to clarify datum target, and target diameter is toleranced:
Create your cone via auto-cone, make sure one pass around the cone is really close to your gage point diameter.
From there, construct a generic plane, parallel to A datum, offset in Z by that defined BASIC value of the gage point.
Then I would construct an intersect circle between the basic offset plane and the cone.

I agree that B datum (the entire cone) is considered the datum, but for the intent of attaining MMB shift, and applying communicated BASIC offset to a "gage point", it seems that the gage point should be used as B datum X/Y origin for the sphere's position tolerance.

Measure your sphere. within XACTMEASURE, reference datum A to level and zero Z.
Use intersect circle of B cone and BASIC offset plane as your B with MMB applied to the nominal ±0.001" tolerance.

This will enable MMB shift, and inspect the part as the engineer has defined (via gage point), so if the angle of the cone is not perfect (but within 8.5° ±0.04°), it should still function as designed.

I feel confident in saying the below posts aren't necessarily an accurate application of the drawing (using entire cone or cone axis as datum), knowing the BASIC offset in height for the gage point and a toleranced diameter are clarified within the print.

[louisd]

Gotta tosss it out here... I started to dive into exactly what I described in the above comment, within my original post... but realized without knowing the presence of BASIC offsets, it is speculation. I deleted all of my instruction, and instead asked for input/clarification. Providing direction without knowing the path can very easily lead to misdirection.

[Sora5]

What you described is what I'm used to doing when the Datum is attached to the diameter at the gage point which is what I am used to dealing with in the past.

I was unsure of using the same approach for this specific case where the Datum is attached to the angle of the taper.

Especially with the feature being positioned being a sphere, I do think an offset intersecting circle would handle the Datum shift correctly. I'll give that a go. Thanks!

[UKCMM]

Datum 'B' appears to be a diameter created at a specific gauge height in the cone.

Or it could be datum 'B' is the centre axis of the cone as the datum header is centred on the angle dimension header.

[JEFMAN]

datum 'B' is the centre axis of the cone as the datum header is centred on the angle dimension header.

+1 for this interpretation !

[Peter Fuller]

+1 also, I agree with this statement.

[Ego Murphy]

The print defines datum B as a Cone.

A Cone is a very powerful datum - it's the only prismatic geometric shape that can control 5 degrees of freedom all by itself: two rotations and all 3 translations.

That said, PC-DMIS sucks at cones. For years I have suspected that it does 2D probe comp perp to cone axis and not 3D probe comp perp to surface. I would honestly measure many rings of AutoPoints and Construct the cone BFRE.

[AndersI]

A cone does probe comp perpendicular to the surface, nothing else. But if you measure circles and construct the cone BF you will get an error like the one you describe.

[RandomJerk]

I've had huge inconsistencies with autocones myself, so much so that I only do cones with vector points now.

[vpt.se]

+1 for this interpretation !

+1

[NinjaBadger]

I'm with Ego on this one (about the cone being the datum) - there'd be no reason not to attach the datum letter to the gauge diameter.

I disagree about the probe comp, it's easy enough to check the hypothesis though Ego!

[JEFMAN]

I agree, the datum is the cone (MIN SEP cone here, I'm not at the cmm but I believe that PC-DMIS can't do it for the instant), so it fixes 2 rotations (cone axis) and 3 translations (vertex).

On the picture, it's used as a secondary MMC, so you should construct a MIN SEP cone perp to A, and use it's axis as a datum... Good luck with this... !!!!!!!!

[Rich P]

I'm with Ego on this one (about the cone being the datum) - there'd be no reason not to attach the datum letter to the gauge diameter.

I disagree about the probe comp, it's easy enough to check the hypothesis though Ego!

As a conical feature you have an axis for datum, as a circle only a center point (2D feature) and would not constrain the all degrees of freedom, would need another datum defined
I think cone is the correct call on this
Excellent interpretation discussion

[AndersI]

... I believe that PC-DMIS can't do it for the instant

PC-DMIS seems to lock 4 DOF with a cone, leaving the DOF along the cone axis free. Quite understandable - if we have measured a cone with sample points, what should the locking criterium be? The sample surface, the cone diameter at a certain point (which one?) or the full cone surface (we don't have it)? What if the actual cone angle is wrong - what's really the locking criterium? As I see it, it depends on what the mating part is, and that may not be known for the current part program.

[Sora5]

Thanks for all the great input everyone.

I had a difficult time wrapping my head around how to apply the material condition on datum B, being called out on the angle instead of the diameter at the gage point. If I use a cone as a Datum, PCDMIS doesn't allow MMB on the FCF, so it's using the centerline instead of the cone size.

To allow Datum shift, I would have to create a intersecting circle, either at plane A or the gage point. I guess this is why I'm usually seeing the Datum being called out on the diameter at the gage point on other similar parts. Maybe the eng had an extra smoke that day.

Thanks all!

[louisd]

Yes! Reference my comment above. Intersect circle for the win

[JEFMAN]

PC-DMIS seems to lock 4 DOF with a cone, leaving the DOF along the cone axis free. Quite understandable - if we have measured a cone with sample points, what should the locking criterium be? The sample surface, the cone diameter at a certain point (which one?) or the full cone surface (we don't have it)? What if the actual cone angle is wrong - what's really the locking criterium? As I see it, it depends on what the mating part is, and that may not be known for the current part program.

I would say that the locking feature should be the vertex, even if it doesn't really exist... (and is not too much repeatable...).
But PC-DMIS doesn't respect ISO 5459 rules for the cones... !
I'm really happy to don't have a lot of cone to measure !

[VinniUSMC]

Seeing as the Datum Structure is A|B, and the cone is B, then plane A already constrains 3 of the degrees of freedom that the cone is capable of (can/may/must). So, all that is left for the cone is translation in X and Y. Then, 1 other feature is needed for rotation about Z. I agree with louisd , construct an intersect circle (on A though) for your datum, and construct a gage circle from the cone at the specified basic dimension (via construct circle, select "Cone" and tell it the height).

The cone is the datum, not the gage circle. The gage circle is the gage circle.

[VinniUSMC]

Around. Doing other things. I got tired of the forums... lol, and my mind has been else where.

How's life treating you KIRBSTER269?

[KIRBSTER269]

Can't complain, Very happy where I'm at now. Great owner, about 5 miles from my home. What about you? I keep saying I'm gonna head over to there and look you up. Just keep saying "I'll get around to it one of these days"

[JEFMAN]

Happy to "see" you again here, Vinni !

[AndersI]

I would say that the locking feature should be the vertex, even if it doesn't really exist... (and is not too much repeatable...).

That's theoretical ISO thinking :-)

As you note, it mostly doesn't exist physically, and is very sensitive to measuring errors. And in the physical reality, it's the cone surface together with the mating part that defines the locking - not really measurable.

[JEFMAN]

That's theoretical ISO thinking :-)

As you note, it mostly doesn't exist physically, and is very sensitive to measuring errors. And in the physical reality, it's the cone surface together with the mating part that defines the locking - not really measurable.

What I wanted to point out here is the lack of the correct algorithm for cones (something like sep min, Chebychev).
If you have a cylinder as secondary datum, a plane as primary, you can project hits of the cylinder on the plane (L2 constraint), the create a max inscr / min circ circle wich represent exactly the right datum (exactly ? exactly with measured hits only, not the real surface !)
If the secondary is a cone, I don't see how to create a cone with an axis perp to the plane, which minimizes the max deviation.
Just a thought...

[AndersI]

Speculating: Create a minsep (Chebyshev) circle from all the projected points - wouldn't that give a more or less correct center point? But then we are back at my old queston: How to easily project a whole lot of points to a plane?

Uh-oh: I just tested it, and the minsep circle seems to be correctly calculated, but there's also an error message that "Math failed...", I guess they consider the form deviation too large, even though the 'correct' result has been calculated :-(