Question about Independency Principle 1

[pmarc]

Imagine that fig. 2-7 from Y14.5-2009 shows a pin instead of the rectangular block, and the flatness callout has been replaced by straightness FCF. Everything else stays the same (of course the diameter symbol is added to 10.7-10.8 dimension).

Knowing this, what is the minimum possible diameter of a perfect cylindrical boundary that the pin would never violate?

Thanks.

 

[CheckerHater]

Belanger,
If perfect form required at MMC, ding is allowed.
“Flat” cross-section has no “material”. Do you believe part with perfect round cross-section that is also bent is at LMC?

Pmarc,
While I am on the look-out for ISO 14660, could you at least clarify which one of your point-to-point dimension is a diameter?
After all your OP was about DIAMETER 10.7-10.8?

 

[pmarc]

My picture shows theoretical as-produced geometry of nominally round shaft. I know, it does not look realistic, but we are debating on what can be possible from theoretical point of view.

 

[CheckerHater]

Based on what theory exactly?

we can say that my picture follows ISO definition of actual 2-point local size.

In ISO the size can only be defined on features of size.

So, let me rephrase my question: "could you at least clarify which one of your point-to-point dimension is a feature of size?

 

[pmarc]

Oh man, this is getting difficult.
Isn't the perfect pin a feature of size? How can point-to-point dimension be feature of size? What do you exactly mean? Can't you see opposed points in my example?

 

[CheckerHater]

pmarc,
You keep avoiding my questions.
You are calling your dimension “sizes”
ISO says the size can only be defined on features of size.
Dimension between two random points on random doodle is not a size.
I am asking to show me features of size and associated with them size dimensions on your drawing.
Am I really that difficult to understand?

 

[pmarc]

Apparently you are, because I do not know what you are asking for.
How can I show you features of size (plural) if we are talking about single pin produced as a 9-arm star? What do you want me to show you? Shall I draw infinite number of opposed points and connect each pair with the line passing through the red point? Are 9 pairs and lines not enough? Aren't the 9 point-to-point distances local sizes of the pin in a single cross section?

 

[CheckerHater]

Just show me 2 points that you'd consider "opposite"

 

[pmarc]

Here you are:

http://files.engineering.com/getfil...65272059f&file=circularity_independency_2.png

The two orange points are opposite, that is they lie along a line passing through the center of associated circle (red point).
The same applies for 8 other pairs of points and for infinite number of pairs not directly shown on the picture.

 

[Belanger]

If perfect form required at MMC, ding is allowed.

This illustrates my point...
You say MMC requires perfect form (even at an individual cross-section).
Then you say that the MMC part can have a ding in it.
So we have a discrepancy at hand: Either the dinged part is not at MMC, or it doesn't have perfect form. Which is it?
Based on this answer we can hop over to my LMC analogy.

I don't mean to derail the main discussion (still thinking about pmarc's 9-pointed star!), but this is important in that it ties in with the relationship between size and roundness.


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

http://www.gdtseminars.com

 

[CheckerHater]

Thank you Belanger,

I think our major disagreement in in terminology than anything else. We have to clearly distinguish between “drawing demands” and “part actually is”
Yes, part at MMC has perfect size and shape.
If perfect form at MMC is demanded on the drawing, part at perfect size and shape with ding on it is still a good part, as long as ding doesn't violate tolerance.

Area where I am less comfortable is to have “local” LMC limited to roundness, but allowing longitudinal variation.
Imagine part made to “LMC” where cross-section has perfect size, is perfectly round, but axis of the part is allowed to bend. Can we bend axis so much that “LMC” boundary will actually violate MMC boundary? Is that a good part? Is it really LMC part?

So I am leaning towards interpretation that in “perfect form required at MMC/LMC” “form” means roundness and straightness combined.

Actually straightness seems to be the only thing required.
When we say “perfect form at MMC/LMC required” we imply that we may have MMC/LMC without perfect form, which is perfect size, but not straight.

I hope I didn't muddy the water even more.

 

[Belanger]

I do agree that it's a terminology thing, because if a cross-section is truly at MMC or LMC, it can't do a darn thing -- it's walking a very fine line in terms of circularity.

But you don't have to be uncomfortable with an LMC part that is bent; if we go with the envelope rule then the part can bend until the envelope hits MMC. The part still has the "least material."

On the other hand, if an external diameter is at MMC cross-sectionally, the tiniest bend already puts it outside the envelope. This is why MMC requires perfect form (longitudinally) but LMC does not require perfect form (longitudinally).

But I must admit I am still having trouble with this 9-pointed star (see next post)...

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

http://www.gdtseminars.com

 

[Belanger]

Pmarc -- your latest graphic shows the orange points, and they indeed seem to cross through a center point. And each of the other dots illustrated would also pass through that same center.

But I'm having trouble with all the other infinite pairs of points; those along the sides of the spikes. Suppose we take one orange dot and move it down the side just a little. Is it correct that the opposing point also moves down its angled side, maintaining the same local size, and the line connecting them still passes through the center?

I'm not doubting it. I just don't feel like doing some sort of geometric proof. But that would have to be true to make me feel better about the "actual local size" truly being a constant.

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

http://www.gdtseminars.com

 

[pmarc]

J-P,
Yesterday when I was posting my sketch I did not have access to my CAD software, so could not really verify if the 9-pointed star was meeting the requirement for having all local 2-point sizes equal to MMC size. Now I have the access, checked it and my answer is NO - the star does not and actually cannot have 10.8 dimension for each pair of opposite points (I am talking about my very initial geometry of the pin).

But my another attempt is the shape similar to shown in attached picture. In my opinion this could meet the requirement (for simplicity I just shown 5-pointed-star-like contour instead of 9-pointed). All 2-point local sizes passing through the center established by minimum circumscribed circle will be equal.

Now, if you imagine that the root circle is approaching dia. 0, the diameter of circumscribed circle will be approaching dia. 21.6 in order to keep 10.8 local sizes everywhere. This will result in 10.8 of maximum possible circularity error. So repeating the answer to your question, in absence of explicit circularity control the maximum possible circularity error will be limited to half of MMC size of shaft's diameter.

Of course all said above may be true only if we assume that all actual local 2-point sizes intersect a common center and not when interpreted in accordance with muddy definition proposed by Y14.5.

http://files.engineering.com/getfil...2d7-9932-f57665dbf156&file=5-pointed_star.png

 

[CheckerHater]

Pmarc,
For some reason you ignore 2 facts:
1. Far shorter measurement can be taken, as it is shown on the copy of your own drawing.
2. Both measurements should be falling within the size tolerance.

 

[pmarc]

I am not ignoring anything.
First of all, I would suggest to stop using my first sketch, because I already admitted it was wrong.
Secondly, if you look at my most recent sketch, you will not find such "shorter measurement" that would pass through the center of the circle. All 2-point measurements passing through the center should return the same value.

Maybe it was you who ignored my last post?

 

[CheckerHater]

It is still possible to take “long” and “short” measurement in your new drawing.

 

[CheckerHater]

As long as you try to keep your measurements within tolerance you will find that circularity error is shrinking together with tolerance.

 

[pmarc]

It is still possible to take “long” and “short” measurement in your new drawing.

This may be surprising for you, but in 2D model I have the lengths of both lines drawn by you are equal to 10.8.

As long as you try to keep your measurements within tolerance you will find that circularity error is shrinking together with tolerance.

I would not agree with this. Again, you are using incorrect shape to prove your point. If your most recent bunch of pictures was showing "flower-like" contour proposed by me, there would be a chance to observe that even for root circle diameter very close to 0 there would be a contour for which all 2-point distances would equal MMC size, so actually in theory there would be no size tolerance needed at all.

 

[CheckerHater]

It may be theoretically possible to create a shape that will measure exactly the same in any direction AND have out-of round error equal to the “diameter” itself, but I am afraid this shape will have purely academic value.

 

[pmarc]

Of course. But we have been speaking theoretically. The original question was solely about theoretical interpretation of the independency principle. J-P's question about maximum possible circularity error was also of a theoretical nature (I suppose), because in reality no one will ever produce a flower when the cylinder is desired.