【转帖】olerances for basic dimensions, continued

yang686526

tolerances for basic dimensions, continued...
israelkk,
don't think of it as a debate.  the standard is a complex one, and i'd say that most of the people who use it don't understand it fully.  i definitely don't, but i do have a basic working knowledge of it and turn to the standard often to improve my understanding of it.  the worst thing we can do if we don't understand why something is toleranced the way it is is to blindly accept that because it's been used that way before, it must be correct.
  you are correct in that you can use basic dimensions without establishing features, by using other dimensional tolerances. examples are angularity and taper.  i have gotten into the habit of establishing datums before i dimension, thus my erroneous comment.
  however, to properly use a geometric tolerance on a hole, you must establish what that tolerance is relative to, which would be datum(s).  it can't be relative to itself.
for true position, per y14.5m-1994, para 5.2.1.3, "it is necessary to identify features on a part to establish datums for dimensions locating true positions."  for your example to be correct, one of the holes would have to be a datum, and the others located relative to that.
  i am definitely open to correction on this, to better understand the standard myself.  i won't accept "because that's how it's done" as a valid reason, but will gladly accept any argument documented by the standard.
eric
the bolt pattern (or more specifically, the pattern of bolt centerlines) can form its own implied datum, as in bolt circles or basic dimensions.
btrueblood,
could you direct me to where in asme y14.5m-1994 that implied datums are referred to, defined or used?  my copy of the standard is missing the index.  thanks!
eric
(blush) y'know, i was worried i'd get called out on that one!  i don't have a copy of the 1994 standard, but base my knowledge on years of using the 1982 version.  
take a look at examples of position tolerancing of bolt holes in a circular pattern.  the tp tolerance is applied with no specification of a datum (i.e. there is no "a" or "b1" etc. called out in the gdt block), and instead is taken as meaning (essentially) the tp of the bolt holes relative to one another.  i may be wrong, and the new standard may have abolished the practise -- if so, they've nullified the specifications of a lot of other standards, wherein the datum is "any and all the bolt holes in aggregate" in order for the wheely thing to fit onto the axle-thing.
by the way, on all our prints, we still specifically call out ansi-y14.5m-1982 as our standard.  
ewh,
   you can apply geometric tolerances without specifying datums if you confine yourself to self contained qualities such as flatness.  when you control qualities like location or profile, you need reference datum.
   asme y14.5m-1994 specifically describes a dual location tolerance where you specify a location of 1mm diameter with respect to datums a, b and c, and a 0.1mm diameter with respect to datum a.  this means that the pattern must be accurate, but you do not care where it is located.  a positional tolerance without a datum does not mean anything to me, at least as per asme y14.5m-1994.  the 1982 standard is not close at hand at the momemt.
   there is no taper tolerance.  my copy of the standard uses a profile tolerance to control this.  both angularity and profile require datums.
                           jhg
drawoh,
i agree with what you have posted, but need to clarify my statement regarding angularity and taper.  while a datum is required when specifying angularity in the geometric tolerance block, you can also control it using a basic dimension and a toleranced dimension (no geometric control block or datums involved)(fig. 2-14 of the current standard).
there is no geometric control symbol for taper, but again, it can be controlled by the use of a taper symbol, basic and toleranced dimensions (no geometric control block or datums involved)(fig. 2-15).
profile is a handy and powerful control.  it lets you control the form without multiple geometric control blocks, and is the simplist way i know of to control non-standard shapes, such as splines and lofted surfaces.  in the new standard asme 14.41-2003 digital product definition data practices, the profile tolerance has been updated so that a unilateral tolerance no longer has to be specified in a true view using phantom lines, but is specified using a modifier in the geometric control block.  
ewh,
   you might have me on this one.  i once did a drawing of a 45 degree elliptical mirror in which i specified the angle as a basic dimension.  the mirror was a piece of round glass bar cut at 45 degrees on both sides, and i did specify datums.  i need to specify one surface as parallel to the other.  
   i put a tolerance on the diameter.  the mirror would have been inspected, admittedly with some difficulty, by jigging it on a 45 degree surface and testing the diameter.  there was no need for an angle tolerance.
   if i had not cared about parallelism, i could have done without a datum.
                           jhg
well, now...i have a copy of asme y14.5m 1994 standard in my lap, fresh from the printers...
there is nothing in the standard that requires a t.p. tolerance to have specified datums.  in fact, the definition of true position (1.3.36) is the position of a feature related established by basic dimensions, not datums.  furthermore, basic dimensions are defined (1.3.9) so that they may or may not reference a datum.  my case regarding bolt circles stands - you can dimension the bolt circle without a datum (provided you as a designer don't care about the position of the bolt circle relative to other features).  the spacing can also be defined by basic dimensions if necessary.  at t.p. tolerance frame can then control the location of the holes relative to the basic dimensions, without reference to any datums required.  in such a case an "implied datum" or whatever you want to call it, is formed from the features themselves, and the basic dimensions used.
btrueblood,
read a little further on.
  para 5.2 states "basic dimensions establish the true position from specified datum features and between interrelated features."  "a positional tolerance is indicated by the position symbol, a tolerance value, applicable condition modifiers, and appropriate datum references..."
  para 5.2.1.3 states "it is necessary to identify features on a part to establish datums for dimensions locating true positions."  in other words, when using true position, you need to establish what the true position is relative to.
  i still have not found where "implied datums" are referred to in the standard.
  i am glad to see that you have the standard and are attempting to understand it.  keep at it and let me know when i misinterpret it.
ewh,
i have read the whole section.  i know this is confusing, since most of the time (possibly "all the time" for some of you) the t.p. tolerance based on datums is exactly the type of control required to establish fits and alignments on a simultaneous basis between mating parts.  my point applies to a special case, and i'm trying to make sure i know how to handle this case (since it occurs occasionally in the work i do), in a manner which results in the least confusion to people here, and thus hopefully to the least confusion to shops that are going to use the resulting drawing.
first, to counter-argue your points:
5.2 does not require a basic dimension to reference a feature, it can also apply between features.
the verbage "appropriate datum references" holds only when such datum references have been defined.
5.2.1.3 is troublesome, i agree - but what do you do when you don't want the pattern controlled tightly relative to other features, or at least not as tightly as you want it controlled for the features relative to the other features in the pattern (i.e. i don't care about the alignment of a pipe flange blank o.d. relative to the bolt pattern to better than 1/4", but the bolt holes must line up to the other part within .03").  i would maintain that 5.2.1.3 simply states that "if you are going to establish a datum, it must be related to features on a part" - i.e. you can't identify a centerline as a datum, but should show the feature(s) that the centerline is to be gauged from.
the definition of t.p at 1.3.36 takes precedence over the statement in 5.2.1.3; at 5.2.1.3 the example is given where the bolt circle is to be controlled (concentric to) an inside diameter as the datum - critical when alignment of some mating part at the i.d. is required, but not always required for all parts.
that said, i know what i'd do now (to avoid the confusion that most of you express in this forum on this topic), is to define the bolt circle as a datum.  would this get the point across to most shops, and avoid confusion regarding the position tolerance requirements as the apply between bolt holes?
i'd worry that the result of the above is an over-dimensioned print (since the bolt circle also needs a basic dimension).