Profile of a Line?

[John Riggins]

I have a rather long part that has a Line Profile specification.
Imagine a banana (keep yer mind outa the gutter graiger).
This banana is positioned with the long axis in Y.The Line Profile is specified in the X axis, in the short, curved direction.
So I measure a series of points all in a row running in line with the X axis.
I create a set of these 12 points.
Dimension the Set Profile of a Line, Form Only, +.010", Bestfit in YPLUS or YMINUS.
I have a series of these lines running the length of the banana.
Due to some deburring irregularities, some of the sets are showing out-of-tolerance.
First problem: graphical analysis has arrows pointing in Y+ and Y-. I expect them to point in Z+ or Z-. I can get them to do that if I change the bestfit to X or Z, but the results are VERY much worse then.
Second problem: When I try to generate a Dimensional info box, I cannot get the deviation values to match the deviation values reported in the textual analysis for the set.
Maybe there is no way to do this?

[cmmguy]

Dimension the Set Profile of a Line, Form Only, +.010", Bestfit in YPLUS or YMINUS.

I would say that the bestfit should be in the Z axis only. The X is the direction of travel and the y is the "fixed" axis. The error that you seeing now is only the error of the cmm positioning as it moves along the path.

I am not sure that I understand the second part of you question.

[\v/inston]

profile of a line should be on the edge of the part.

profile of a surface should be on the surfaces of the part.

looks to me, from the picture you uploaded, that you are trying to measuring profile of a line on a surface.

[cmmguy]

That's a good call Winston. We would need to see the drawing to be sure but your probably right.

[bob mappes]

a "line" can be measured on a surface...

1) say the surface to be measured is vertical to the table in the "z" direction
2) now do a "linear open scan" along the surface in "x" and or "y"
3) the scan results in a "line" not a surface (like an edge, but not necessarily an edge)
4) you now can do "profile of a line" of the scan set

if you were to include more points that were moving in the "z" direction as well , then you are applying a third axis and "profile of a surface"

pg 170 of ASME14.5M-1994 gives an example of a line profile

[John Riggins]

Here is a screen shot.... no drawing

profile of a line should be on the edge of the part.

profile of a surface should be on the surfaces of the part.

looks to me, from the picture you uploaded, that you are trying to measuring profile of a line on a surface.

It is specified on a surface. My interpretation is the "short" direction on this part is what has the line profile specified.

[\v/inston]

i dont see the point of calling out profile of the surface and profile of a line on the surface tighter then the surface profile, it should be one or the other. kind of redundant.

pg 170 of the ASME says "each line element". there could be a million lines within that surface and that is not feasible to measure unless you are using a metris laser scanner.

do they have a criteria for how far apart the measuring points should be?

on our ford prints they give a general profile of a surface callout and a general profile of a line. for the general line profile i measure all the trim/part edges for the general surface profile i measure all the surfaces...

[John Riggins]

i dont see the point of calling out profile of the surface and profile of a line on the surface tighter then the surface profile, it should be one or the other. kind of redundant.

pg 170 of the ASME says "each line element". there could be a million lines within that surface and that is not feasible to measure unless you are using a metris laser scanner.

do they have a criteria for how far apart the measuring points should be?

Frankly I don't see the point either. But trying to communicate with our asian customer is a bit of a challenge. Our only criteria is to meet the requirements of AS9102A.

[jon_dav_uk]

I have seen this sort of thing before

Profile of a surface calling out three datums- this covers both shape and position
Profile of a line calling out no datums - this covers shape only, ie you would be able to best fit the scans

Not sure if this is relevant to your part but my interpretation would be that the designer wants the shape to be good but he is less bothered about the position of that shape relative to his datums

[cmmguy]

The purpose is to control the surface tighter in a smaller area and looser overall.

It is sort of like flatness between mounting holes being tighter than overall flatness on the mounting surface.

[\v/inston]

you still have two tolerance zones on a 3D surface. the points you measure are compared to a CAD model that is hopefully aligned using the primary datum scheme.

there are two schools of dimensioning profile:

1. creating a feature set and using the pc-dmis 'form only' or 'form and location' of a line or surface. form only measures the shape and forma and location measures the shape and location.

2. dimensioning the T vaule for each individual point.

3. i guess you could use both.

you cannot measure profile of a 3D surface without CAD.

whats the best way? whats the right way? whats the wrong way???

i don't know i normally go with #2, i don't really like the pc-dmis profile dimensions they seem to confuse people too much.

[John Riggins]

Back to my original questions:

Should I be bestfitting around "Y"?

If so, is there any way to get the graphical analysis arrows to indicate deviation in "Z"?

[cmmguy]

Back to my original questions:

Should I be bestfitting around "Y"?

If so, is there any way to get the graphical analysis arrows to indicate deviation in "Z"?

I think my original answer stands...

[Paul Sarrach]

The way I see it, is its asking for .030 to the datum scheme and .010 to itself. So I would try zeroing out on one point and then dimension the T Value for the rest of the points. And which point? Use your discreation. Kind of like a True position to a hole pattern back to itself.

[cmmguy]

To me, the .010 allows for each individual profile in question to be best-fitted without regard to the datums, but overall, all of the profiles cant be over .030 with regard to the datums when evaluated together.