【转帖】physical identification of datum features

yang686526

physical identification of datum features
does anyone out there ever use, or even aware of the need for datum feature physical identification on symmetrical parts?  does anyone know of a text that provides an example?
ringman,
i'm confused by your capitalization choice.    there's two types of symmetry in asme y14.5m:  symmetrical outlines and symmetry gd&t.  because you are talking about datums, i'm going to assume you are referring to gd&t.  
in one fashion or another, any gd&t needs to reference some real feature on the part.  tolerances of location (symmetry, positional, concentricity, etc.) even more so.  the datum for sym should be based on some feature or set of features.  if you have access to asme y14.5m, i suggest checking out figure 5-60 as the primary example for how to do this.
i hope i understood your question and gave the appropriate answer.
matt lorono
cad engineer/ecn analyst
silicon valley, ca
i think i get the question.  if one side of a symmetrical part is called out as the primary datum feature, then there are two different sides on the physical part that could be used as the datum feature.  in other words, the part could be lined up on one side for machining and lined up on the other side for inspection or assembly.  problems may result.  
in one of bill tandler's course manuals, there is an example that specifically mentions this problem.  the solution was for the designer to make a small change to remove the symmetry.  possibilities were putting a small chamfer on one corner of the part, and offsetting one hole of a rectangular pattern.
evan janeshewski
axymetrix quality engineering inc.
ringman:
i am assuming, as axym has done, that you are referring to something like figure 5-4 of y14.5, that has full symmetry. don't know of or have not seen any text to that regard, but two things come to mind.
1. if you're speaking of a secondary,tertiary datums, as in 5-4, does it make a difference? i guess only if one of them had a discrepant or out-of-tolerance feature. then the discrepant part could be non-permanently marked to distinguish the datums.
2. create a non-functional, non-symmetrical feature as axym has suggested.  
ringman,
   you have two choices when preparing drawings of symmetrical features.  
ignore the symmetry.  attach your datums to one side of the part.  your problem does away.  be careful preparing your drawings though.  the error conditions are not totally obvious.
attach the datum to the symmetric feature, as shown in asme y14.5m-1994, figure_5-4.  if you do not specify the datum at mmc, your fabrication and inspection fixtures become complicated, because they have to account for the varying width of the symmetric feature.  if you can specify the datum at mmc, fixturing is simple.  again, prepare your drawings carefully.
   a symmetric datum is a feature of size.  i try to avoid using features of size as datums, unless they are much more accurately toleranced than the features to be controlled.  
   a fixture that locates to a sloppy feature of size, regardless of feature size (rfs) has to be adjustable.  it has to grip on and centre the feature.  such a fixture is not simple.  is this your problem?
   section 4 of asme y14.5m-1994 explains datums very clearly, and in extreme detail.  it will tell you how to interpret any weird drawings you have been sent.  
                         jhg
perhaps a clarification of my original question is in order.
i was refering to section 4.3 of the y14.5 standard which states that if a datum feature is not readily descernible, it should be physically identified.  this has been the case for at least 26 years; see section 4.2 of the 1982 standard.   
ringman,
   okay, now i understand your question.
   handy dandy design tip #1:  if the feature is to be a datum, make it not identical to or symmetric with anything else.  i don't think i have ever thought about this.  there is a significant potential for making a mess.
   i am not aware of any literature on this.  if all fabrication and inspection are being done at one location, you can make the datum feature with a felt pen.  if you are a fabricator vendor and you anticipate that the customer do inspection, then your life gets complicated.
   at some point, you should advise your customer that the drawings are ambiguous.  they could drill a small hole or other id mark next to the datum, so that both of you are using the same feature.
                      jhg
this is an excellent observation to what the standard says.  i had to think about this for a bit. in fact i even looked to see if they plan on addressing it in the draft of the upcoming version and it appears that they don't, because the same wording exist.
i don't believe i have ever been in this situation because if i have a feature that is so similar that it would cause me to physically mark it. i would question whether or not it was the best datum candidate in the first place. when dealing with symmetrical components i would select the feature that mates to the next level component and then i would use the mid plane derived from the fos composed of identical feature.  also, if there is a pattern of features this would also be a robust selection for a datum.  here is a tip look for the features that have the least amount of "float" or constrain first when assembled.  
here i am somewhat contradicting myself about being in this situation because i now recall.  in any case, what i  would imagine is happening is if the component is a plate and you want to use a planer datum instead of mid-plane.  i believe some one mentioned it earlier but at this point i would specify a break/chamfer/radius on only one side/edge of the component with out compromising the function.  so that you can differentiate the two.  i have also witnessed where the part# has been specified to be stamped/etched or labeled in some permanent way on the surface opposite the datum.
just my thoughts, otherwise ringman this was a good catch and has led to some thought provoking thinking and responses.