dimensional Tolerance

[southpaw]

I am in the process of reverse engineering a part. i have measured key areas of 4 each of the same part... i have come up with the standard deviation, and the mean of these dims...
how would i go about determining the manufacturing tolerance of each dim???i is it just the mean, the STNDDIV or some other method??

thanks,
southpaw...

 

[GregLocock]

Well, you are definitely in the wrong forum for this question, but I'll give it a go.

If a process is stable and normally distributed then the ratio of the tolerance range to the 6 sigma range is called Cp. If this is greater than 1 then you have a tolerably good process, 1.33 is preferred and 1.67 is what we currently aim for.

So a starting point for your dimension would be

mean +/- 3*stddev

But that really is a lousy way to do it! (tolerances should be the result of a tolerance study, not a function of the process that makes the part) Cheers

Greg Locock