【转帖】can a datum feature reference itself in y14.5m 9gdt0

yang686526

can a datum feature reference itself in y14.5m (gd&t)?
if i have a part with 3 datums, datum a is the surface on which i have datum b - a hole and datum c - the width of a slot and i want to tolerance the length of the slot can i use a positional tolerance that references a|b|c or can i only use a|b because the slot is c?
if i do tolerance the length to a|b|c what would that physically mean?  how could the length of the slot vary, positionally, with respect to the width of the slot?
thanks,
j
(originally posted in ansi standards forum)
jmarkus,
   your datum c controls the rotation of your part.  until this is controlled, there is no rotational datum.  your slot's width is controlled by datums_a and_b.  once the slot is in place, you have three datums.  
   if there was a feature you could use to control your slot, *it* would be datum_c, wouldn't it?  
                          jhg
   
the rotation is controlled by the width of the slot, so datum c is only the width.  the question is the length is controlled back to a & b, but can it be controlled back to c when c is a different dimension on the same feature?
j
we do this.  this is effectively breaking the slot into two features, length and width.
imagine if you were dimensioning a different slot.  you could give each dimension (width and length) its own positional tolerance.  no difference.  completely legal, just not common.  don't forget to omit diamter symbolfff">.
i could be the world's greatest underachiever, if i could just learn to apply myself.
you can't orient to something which does not already exist. that is simple rule to re  
but physically, how could the length of the slot vary, positionally, with respect to the width of the slot?  to me this seems impossible and therefore dimensioning the length of the slot to the width is meaningless.  so if i position the length of the slot to |a|b| only, am i missing something?
because the boundary callout pins the virtual condition boundary. the virtual condition boundary ensures the worst case fit of the mating part is line on line contact only unless you want an interference. makes perfect sense that the shape of the tolerance zone of a slot type feature needs to be rectangular. tolerancing system must be able to accomodate the needs of building real parts in an concise way subject to only one interpretation or it's rendered useless.
quote (jmarkus):
but physically, how could the length of the slot vary, positionally, with respect to the width of the slot?
picture that the slot position is constrained along the slot centerline (the line symmetrical between the two linear portions).  the entire slot can "slide" along that line until it is constrained by another dimension and tolerance.
orientation by definition is directing towards a point.  my question is: can a line or plane be 'oriented' with respect to another plane?  just something to ponder.
this with regards to plane c as the orienting feature.