Profile question

It effectively gets evaluated as flatness.

You will still have a min and max, they will just be equally dispersed. So if you have a profile measurement of 0.010, your min will be -0.005 and your max will be 0.005, for example.

Here is what i don’t understand, if you you dont have a datum, which means that there is no actual nominal distance of any kind, there for the profile can be anywhere even an inch from the nom, as long as the profile is good to itself . Unless i am not understanding this correctly and regardless of whether there is a datum or not, the profile is still evaluated in both directions from the nominal.

Junkie, if it was flat yes, but what if the surface is nit flat, its a curve or a radius.

Yes, the feature can "float". If it's a circle, a form only profile would be like roundness.

So the profile will have both min and max huh

Yes, it will still have a min and a max. The nominal surface informs the tolerance zone, but thinking of it like your surface must be +/- .005 isn't really helpful. It's the tolerance zone that has the perfect shape of your nominal surface +/- .005, but your actual condition should have a min of 0 and a max of less than .010. The easiest way to think about it is a flatness tolerance on a plane. The exact same thing happens for a curved (non-size) feature.

Things get funky for features of size that have profile with no datum structure though.

Now that you mention it, I guess I've never seen a profile callout to no datums on anything other than flat surfaces. But yes I think your logic is right, it just needs to be good to itself.

I've played a little with Excel, on a theo profile Y=2*SIN(X).
If you realize this profile, you can imagine different defects of size, orientation and vibrations, so the measured hits are following a function like Y1=A*(sin(X)+ random value) + B*X + C.
The form only must fit this function, but the form + location just look at the measured hits.
This gives those results :

I entered A=1.96, B=0.03, C=0.05 + a random on Y value, the LS calculation givesA=1.961, B=0.026 and C=0.058.
The min max value for form only give ±0.080, but the differnece between hits and theo profile is +0.254, + 0.001.
The first result is the difference along Y between the red curve and hits, the second is the difference between green curve (theo) and hits.
The result should be different if I used distance along the normal vector (using LS + vector on PC-DMIS) instead of distance along Y (LS only).

Hope it can help, not sure of it, but it was a pleasure to play with a sinus !!!!!!!!!!

I was thinking of a overlay concept, where you have a perfect profile tollerance mylar and you overlay it over the part shape, and you can wigle it and move it around anywhere as long as the profile fits. In which case you can move so that the profile deviation will only be in one direction

Yes. A mylar is a great way to validate a "form only" profile, if it fits within the bounds, no matter how you must rotate or translate it, then it is good. The difference in the software is that the software has to start from somewhere. And where it starts from is 0. Everything else is (well, should be, there are some nuances) plus material.