【转帖】spacer problem

yang686526

spacer problem
folks-
i have the very simple scenario of a cylindrical pcb spacer used with conventional screws.  in the pcb design there are component keepouts sized to accommodate the maximum "float" of the spacer when the od comes at mmc and the id comes in at lmc.  common sense says that when the id comes in at mmc (providing minimal orbit about the screw), then the od should be allowed to come in at mmc.  as the id enlarges the float (=diametrical clearance + bonus tolerance) increases and we then need to make the id tend toward lmc.  it looks like positional tolerancing is the way to go so i assigned a datum to the id and controlled the od with a positional tolerance having two mmc modifiers.
after reading 5.11.1.1 of y14.5m-1994 i am led to believe that a positional tolerance for coaxiality control has to be stated with rfs or mmc modifier but never with an lmc modifier.  why not lmc also?
tunalover
if the spacers are shouldered, then the pcb hole id and the shoulder od and the spacer od and the coaxiality between spacer od and shoulder od determine where the keepout has to be, which is what i assume you are trying to check.  
if the spacers are plain, then the screw od and its clearance to the pcb hole get involved.
or maybe i don't understand the problem you're trying to solve.
mike halloran
pembroke pines, fl, usa
mikehalloran-
for illustration, suppose the od is .120-.130 and the id is .076-.081.  next suppose the feature control frame calls out a .001 diameter tolerance zone when the od is at mmc.  the id is a datum feature of size so we can capitalize on that fact by gaining bonus tolerances.
i'd like to invoke -g- at lmc while the od is at mmc.  in theory this provides the largest positional tolerance when the id is at mmc and the od is at lmc (.011).  this is the extreme where the locus of the spacer is at its smallest.  we can tolerate the largest positional tolerance at this condition.  on the other extreme, with the id is at lmc and the od is at mmc, then the positional tolerance is the smallest.  this makes sense because the spacer has the most "float" around the screw at this condition.
here are some numbers:
at od=.130 (mmc) and id=.081 (lmc), pt=.001.
at od=.130 and id=.080, pt=.001+.001=.002
at od=.130 and id=.079, pt=.001+.002=.003 and so on until
at od=.130 (mmc) and id=.076 (mmc), pt=.001+.005=.006
at od=.129 and id=.081 (lmc), pt=.002
at od=.129 and id=.080, pt=.002+.001=.003
at od=.129 and id=.079, pt=.002+.002=.004 and so on until
at od=.129 and id=.076 (mmc), pt=.006
to finish up:
at od=.128 and id=.081 (lmc), pt=.003
at od=.128 and id=.076 (mmc), pt=.008
at od=.127 and id=.081 (lmc), pt=.004
at od=.127 and id=.076 (mmc), pt=.009
at od=.126 and id=.081 (lmc), pt=.004
and so on until
at od=.125 (lmc) and id=.076 (mmc), pt=.011
to achieve the above, what i want to do is assign -g- to the id and invoke it at lmc while the od is at mmc.  i haven't been able to find where it says i can do this!  paragraph 5.11.1.1 seems to say that you can only invoke rfs or mmc, not also lmc.
tunalover
it seem that you could accomplish the desired results without any gd and t.  what if you were to tolerance the id and the wall thickness?  it would probably save a lot of time in the long run.
ringman-
tolerancing the id and wall thickness takes no advantage of variations in size.  to dimension and tolerance it this way would defeat the spirit of bonus tolerances.  granted this part is very simple, i work in a new company where there are likely to be many more custom aluminum spacers in future designs.  the part is drawn and toleranced once to accommdate perhaps dozens of different sizes (dash nos) on the drawing.
tunalover
not to put too sharp a point on it, but the future bodes ill for a company that:
- can't control part number growth.
- buys custom parts where commodity parts are available.
- pays you to shoot sparrows with a howitzer.
mike halloran
pembroke pines, fl, usa
mikehalloran-
allow me to help clear up some false assumptions:
a.  the spacer is for no. 0 and no. 1 hardware with non-standard lengths.  only a fool would design parts when cots parts will do.  the only cots source for non-standard spacer lengths is asm but they go no smaller than no. 2.
b.  there are two dash nos on the drawing.  that is not uncontrolled part growth.
c.  if you do the drawing and d&t once it never has to be done again except to add dash nos.
d.  conventional bilateral d&t provides no control on coaxiality (but we both know that if bilateral d&t is used we will in all likelihood get the parts we need).
e.  since i am salaried my company pays me for a 40hr week.  since today's business climate allows no time for professional growth and general research (not related to a particular project), all of my research and learning about things like this happens on my own time.  also i am working on a project that is badly behind schedule because of poor planning that i had nothing to do with.  so far this week i have 52hrs on the clock (including the weekend).  this problem was very educational and i am dedicated enough to do it all on my own time.
next time i suggest you think things through before you spin off with cynical comments based on assumtions.  we all know what happens when we ass u me.  i'm fighting the good fight and carrying the gd&t torch forward.  if you have a problem with that then next time don't waste your precious time responding to posts in this category!
tunalover
tunalover,
can you explain what how the 'bonus tolerance' would be applicable in this case.  i see no benefit in adding gd & t for the spacer definition.
i still am of the opinion that a id and min edge distance would suffice.
ron
the method adds a coaxiality control that varies with material condition.  as the od nears mmc and the id nears lmc the allowable tolerance is minimal.  as the od nears lmc and the id nears mmc the allowable tolerance is the highest.  this controls the locus of the od to lower the risk of shorts while providing the loosest possible coaxiality tolerances.

tunalover
okay, perhaps that was too sharp a point.  i apologize for any offense you may have taken.  and i salute your obvious dedication to your cause.
if this were anything but an educational exercise, i would (gently, i hope) point out that a screw machine, not a document, controls the locus of the od, and the locus of the id, cuts them in the same chucking, and usually keeps them coaxial within a thou or two, and that in the case of really small spacers like in your example, it's probably a swiss-type screw machine, that holds tenths, so loosening the coaxiality tolerance, or even including one, is probably not an effective use of your time.
i would go on to suggest, gently, that your time would be better spent trying to redesign the assembly so as to eliminate a single- function part like a spacer.
and, i would further suggest that you _not_ use this exercise as a presentation to management in furtherance of gd&t.  the obvious reaction, to a person accustomed to making assumptions and to making decisions with incomplete information, is along the lines of, "if gd&t makes it this hard to tolerance a part that has only three dimensional attributes, it must make it impossible to tolerance an average part."
since it is an educational exercise, i will make no such suggestions.
again, i apologize; i did not realize you were trying to tolerance just the spacer, and not the assembly of which it is a component.
mike halloran
pembroke pines, fl, usa