Hey guys. Could one of you explain this for me. I thought probe compensation was correct prior to a part alignment. Because even without a part alignment there is still a coordinate system within the CMM. Why wouldn't it be correct without a part alignment?
It will 'never' be correct to your part, unless you part is perfectly flat and square to the machine axis.... FOR SINGLE POINTS.
A plane will get the correct probe comp. A sphere will as well.
BUT, a line measured without an alignment will be comped to MACHINE plane, not part plane.
Oh ok. i think i understand now. Thats how i explained it as well. If you hit perfectly in the X Y or Z axis the compensation should be correct relative to the machine axis but it will not be a perfect hit with a perfect 90 degrees from the surface.
without an alignment, all probe comp is done to machine axis, except as I noted for true 3D features (plane & sphere). Once you have an alignment, all probe comp for MANUAL features will be to the alignment. This is why 'they' recommend that you measure a plane, then level, before measuring lines/circles so they they comp will be in that plane.
THAT BEING SAID.....
Almost all my parts are checked on custom, part-specific 3-D fixtures.... that are 'supposed' to be all within +/-0.002" (0.050mm) everywhere. A plane I measure on the fixture base SHOULD be parallel with the machine axis (example: ACTL/900.643,326.325,-598.326,-0.0002622,0.000067,1) and that is close enough to measure a line or circle (while still in machine alignment) and not have to worry about it, not to mention that a DCC iterative alignment will then be done on the actual part locators (net pads and locator pins).
In my opinion this question has more to do with probing hit vectors.
Probe compensation is relative and subject to a tangent-surface probe hit.
Any time you are probing a component that isn't perfectly normal to the surface, you induce COSINE error.
None of the above applies, because without a robust part alignment, you will always have some amount of cosine error.
With a scanning head, selecting "3D" mode in the measured point (instead of work plane), selecting 'fine manual probing" in F10 should give a compensation not to far from the right one (IMO)
