Surface Profile help

[Skullbuster]

surface profile ex.jpg Pcdmis 2019 R1 build #375 - Hi, everyone! I have never measured Surface Profile before and honestly have no idea how to accomplish this. I have included a rough pic of what needs to be done. Please excuse the drawing as I am not allowed to display the actual print. Am I just scanning the area with the probe? Unfortunately I don't have a part to play with yet. Sorry if this seems basic to you guys but I really have no experience do this type of measurement. Thank you for any input!

[Matthew D. Hoedeman]

8-SIDED nut basically, right? (or 6 sided, why the X6? if 8 sided) Scan? I wouldn't. I would measure 4-5 points per flat (or more if it is big, no idea how big it is). Dimension each point "T" axis, +/-0.1 (1/2 of the profile in the upper), to ABC alignment. Then do a 2D best fit of all those points, ROTATE ONLY!, and do the same again, but +/-0.05. Then construct 8 lines from the points on each flat, dimension the flatness.

[louisd]

functionally speaking, A and B datums should be flipped.
B-datum should be the level plane, A datum has such a short axis, that leveling to A is nearly impossible.
If you want your parts to pass and functionally meet spec, level to B, Origin to A, rotate about a line from centroid of A to centroid of C.

[Peter Fuller]

The way I see it that cylinder controls where the head of the part exists in space especially provided that the cylinder in question is precision ground and fits into a mating bore. How would B control the profile surfaces position in space if you rotate two axis to it?

[louisd]

Presuming this is essentially a shouldered nut:
A datum is a 'ground' (it could be turned with a lathe just as easily, I don't know why you think it is precision ground) cylinder, B is a perpendicular interface plane with whatever it's going to be installed onto.
-The axis line of that shoulder (defined as A) is less than 1/5 the height of the part; which means any inaccuracies in the measured cylinder's vector/axis will be compounded/projected 5x upon the axis of the part.
-The plane that derives B is a very large and flat surface. Functionally speaking if this is threaded, the threads will have typically >0.008" of play in which the axis can tilt/wobble within, once torque is applied to the true axis-vector-controlling surface (B datum).
--When you torque this nut down, it will self-square to the B datum and its respective mating surface, so as the threaded ID permits, with little regard for that tiny A datum shoulder.

[louisd]

Maybe our error here is a result of inconsistent interpretation of the napkin-sketch.
-Do you see Datum A as a small undercut shouldered cylinder beside/coaxial with the wrench flats?
-Do you see datum B as the planar face of the part?

[Skullbuster]

It actually has 10 sides. 6 sides were easier for me to draw lol. I agree with the Datum structure as another lab is trying to measure this as well and they are complaining about it. also they seem to keep getting 0.01 for all their results. But you are suggesting points instead of closed line scan? and the part is pretty small approx ø50

[Matthew D. Hoedeman]

I would do points as a scan, which should be re-learn, will not have the same points on the same flats if there is a little rotation, hard to pick the points for the straightness callout.

[Skullbuster]

I can't seem to get good results on this. Getting a couple hundred microns above spec. I was using Auto-Plane feature (the area is small - about 20mm x 30mm) to create a plane (6pts) and then use XactMeasure and dimension Profile of a Surface. Dimensioning a single point will be in spec but I need to represent the form of this flat yes? I'm stuck....

[SingularitY]

Correct me if I'm wrong, but I thought if the Primary and Secondary Datums are repeated in a lower segment of a feature control frame, then the pattern is only held rotationally relative to the Datums, not translationally since the upper control already defined translation. I attached a snippet from the standard.
Top FCF level and origin to A, Origin to B, rotate to C.
Bottom FCF Level to A, B really wouldn't do much, then best fit 2D both translate and rotate to the profile points/scans.
3rd callout seems like flatness would be better. But you could do straightness I guess.
that being said, you could repeat all 3 datums in the same order in a lower FCF and it has the same concept. Relating only rotationally to the FCF.

[Skullbuster]

do i have to dimension all of the features in quesiton simultaneously (like hole position)?

[SingularitY]

ABC doesnt matter since all 6 degrees of freedom are constrained. AB is a pattern and should be checked simultaneously. Straightness you dimension each face separately.

[Skullbuster]

Thanks, Singularity. Right now I am leveling to B (Z+) because it just makes sense I guess. How would I level to A (X+Y)? I have no experience measuring Surface Profile and I am well past being embarrassed about it. LOL. Thanks again for your help!

[SingularitY]

Probe the longest possible cylinder on Datum A without running off or hitting unwanted surfaces. Level to A in the direction of the axis vector. If you look at the vector direction of the auto cylinder, the IJK, level the same way. IJK= 0,0,1 That means level Z+. IJK= 0,1,0, level to Y+. Etc. Origin to the cylinder with the 2 axis you didnt use for the level. Create a line through Datum A and Datum C and rotate in the direction of the line vector using the same method as before, about the same axis as the level. Origin to B. Dimension profile of a surface to all surfaces.
Then created a best fit full 2D alignment, about the same axis as your level, selecting all individual points or scans of the profile, select vector least squares, compute, ok to create. Then dimension profile of a surface to all surfaces.
Dimension straightness of each surface separately.

[Peter Fuller]

+1 Can, May, Must. Your Cylinder must control 2 translations and two rotations. I know that cylinder surface on A is small but I highly suggest at least 3 rows of data with at least 7 hits per row. Personally I measure every diameter I measure with 13 hits and every cylinder with 13 hits per level unless I'm directly instructed not to for the sake of time.

[Skullbuster]

Thanks guys! Don't have results yet but I'm trying it out...

[Calvin.Korver]

I was reading through this.
Do you remember what you went with?