【转帖】centroid

yang686526

centroid
i have two rings, one inside the other.  a material is injected between them that holds them together as a single assembled ring.  their surfaces lie on the same plane.
the outer ring has three holes on its face that are dimensioned using x and y dimensions and are unequal distances from the center of the rings, but about 120 degress apart.  the inner ring has two features that protrude toward the center.  one of the protrusions is at the 12 o'clock position and the other is 170 degrees, counter clock wise.
an assembly gets mounted in the center of these two rings.
the three holes in the outer ring are for alignment pins and their unequal dimensions from the center must be maintained.  these alignment pins are the only thing that locates the unit.  
currently we use basic dimensions to describe the unequal dimensions from the theoretical center of the rings to the center of the three holes.
here is the first question.  one surface is datum a.  if we use a true position to datum a for the three alignment holes in the outer ring, and call those three holes datum b, do the unequal basic dimensions describe the theoretical centroid?  if not, can a note drive that the theoretical centroid be derived from the unequal basic dimesions?
here is why we are thinking of doing this.  we want to allow the inner ring to translate in relation to the theoretical centroid described by the unequal dimensions from the alignment holes in the outer ring.  and, we want to allow the protrusions of the inner ring to be able to have a rotational tolerance.
in other words, say we want to allow the center of the inner ring to be able to allow translation of .010" for its tolerance in relation to the center decribed by the unequal distance of the alignment holes. and, we want to allow the protrusions of the inner ring to be able to have a rotational tolerance of .5 degree in relation to the actual center of the outer ring described by the three unequal spaced alignment holes.
i hope that is clear.
we have an alternate solution in mind, but i wanted to pick ya'alls brains first.
thank you!
(if we conserve energy, we're just letting the conservationists win.-homer simpson)  
find a job or post a job opening
well, you could name the three holes as datum features, individually, b, c and d. then you could proceed to use them in combination, b-c and d, to establish a reference frame with which to locate the protruding features on the inner ring.  use positional tolerance as required with the acceeptable locational tolerances.
i am wondering why you apparently do not use positional tolerances on the three outer holes.
this is rather brief, but i think it should help.
ringman,
"i am wondering why you apparently do not use positional tolerances on the three outer holes."
the three holes are called datum b.  that hole pattern being called out as a datum drives the datum to be the centriod.  however, the three holes are unequal distances from the theoretical center that we want to use as our datum.
so, do the unequal basic dimensions describing the three holes identify the theoretical centroid that we want to use as our datum or is the centroid of a hole pattern always the center distance of the pattern.  if so, can it be overriden with a note?
thanks for replying
ouch, no edit here.
the three holes are called out with a true position.
you can do do it that way but does it best imply what the function of this part actually does? what is the function of this is part?
using a pattern of holes seems like a smart thing to do however it has some luggage which might make it not as desirable as the alternative. by declaring all three holes as a datum what that does is generate a centroid of the three holes. that may or may not coincide with the desired center you have in mind. datums b & c are established from that centriod and all 6dofs are constrained. this is okay if you're making a lot of these parts and you can afford to build an expensive dedicated gauge to check them. there is a cheating compromise workaround for this if you're interested. personally i'd stay with the solution you've already got but add a datum target reference point or axis at the desired centroid. it would give you the convenience of using the center but still be traceable back by basic dimensions to the primary set of datums. again this might entail building a special guage to simulate that datum target.