Geometric Mean,or Nominal 1

[etrant1]

For several decades, I have been modeling CAD data to the geometric mean. The number in the dimensiuon is the nominal, and in the case of tolerances that are even ( say +/- 1) it is also the geometric mean.

If I have a dimension of 10 +2, -0, thenthe geometric mean would be 11, and I would model it at 11. We use the CAD data to build the mold for the part. I have a seasoned high level Engineer that insists that the CAD data always match the nominal. His example is: 1000 +.5, -1.5. He states that he wants the part dimensional distribution to be centered about 1000, not 999.5.

Here is the problem. There is a note that says in case of difference between CAD data and dimension on drawing, the drawing shall take precedence. When a NC programer sees this and the dimension, the programer will alter the tool paths to shoot for 999.5, plus other compensations for tool wear etc.

I seem to remember in 14.5M appendix A3 stating that the value of the dimension shall match the CAD geometry.

Does anybody have a standard that defines the correct action. Thanks!

 

[diamondjim]

Using min and max as you tried to
do seems to be the best approach.

Often the nominal dimension on one
part is the max dimension on the
other part that it will mate with.
The nominal dimension seems to be very
popular in Europe. Most shops prefer
the mean dimension for programming.
Using the nominal would produce more
scrap. That should be enough of an
argument to satisfy your high leveled
engineer. Engineers often prefer
max material conditions to draw to and
the shops prefer mean dimensions.
Both make sense in the way they are used.
The engineer uses max material conditions
to ensure that clearance will be in final
assembly. The shop uses the mean for
programming. I too was guilty of preferring
max material condition for the above
reasons and still prefer it for customer
or application drawings. With the advent
of data bases I do not necessarily see a
conflict between the two as long as the
shop and engineering operate separately.

 

[Hush]

You'll find the older the engineer the more likely he or she will prefer max matl cond. Comes from more exposure to manual equipment. If your ID is 100 +/- 2 an older type like me will tolerance it as 98 +4/-0. The thinking is, if the machinist screws up and misses the nominal by 4 the piece is still salvagable, I.E. more room for error. Of course this doesn't apply to CNC equipment but old habits die hard and I figure the programmer's smart enough to pick his own nominal. Of recent I've switched to limit dimensions, sort of a happy medium.