pyromech,

Where do the statistics come from? When I apply tolerances to drawings, I am specifying what I want. I hope the fabricator can do it. My tolerance stack-ups are based on specified tolerances.

I don't have statistics unless I inspect a batch of pieces.

--
JHG

I saw a demo of some tolerance analysis software a year or 2 ago. The demo was very interesting; the software takes your 3D models and varies the parameters randomly within the specified tolerances then reports if the assembly can be built or not. It is supposed to help zero in on areas where the tolerance needs to be tightened or can be loosened without ill effect. I don't use the software myself, so I cannot comment on how it performs in real world scenarios.

cowski,

If you do tolerance stacks manually, you assume all the worst cases. If you specify a positional tolerance of Ø0.4mm, your inspector must accept parts when the holes are 0.2mm out of position. What is ±3σ? Maybe the fabricator's process is Ø0.1mm! Maybe it is Ø0.6mm, and their inspector is doing 100% inspection and rejecting a significant number of parts. You don't have statistical data until you inspect stuff. You cannot assume that the statistical distribution has any relationship to your tolerances.

In mass production, there are all sorts of work-arounds, but you need to know the capabilities of your fabricator.

--
JHG