Calculating True Position CPK's
	Post Reply		Forum

Replies to this message

Re: Calculating True Position CPK's
Re: Calculating True Position CPK's -- geobe2480	Post Reply	Top of thread	Forum

Posted by: Cragyon ® 12/05/2004, 21:44:55 Author Profile Mail author Edit

Understanding process capabilites is normally derived by experience, direct measurement, or understanding the manufacturing process employed and predicting the result.  As allways, good tolerance design and fits should be well within the manufactruing process capabilites.  If your just placing a nut on a plate and using a tape measure to locate it, then weld, I would apply a tolerance of something greater than 1 measurement unit on the tape, 1/16 - 1/8 of an inch. If you are using a fixture which perhaps locates the minor diameter of the nut with respect to a couple of references on the plate, then I think 0.5 mm would work fine.

Post Reply | Recommend | Alert

Rate

Where am I? Original Top of thread

Re: Calculating True Position CPK's
Re: Calculating True Position CPK's -- geobe2480	Post Reply	Top of thread	Forum

Posted by: traingdt ® 12/05/2004, 19:41:05 Author Profile Mail author Edit

For the most part, Cpk's on a geometric tolerance are no different that for traditional tolerances (assuming that the position tolerance is RFS). But realize that the distribution will be only half of a "bell curve" since position measures the distance from the perfect location in any direction (it's an absolute value; there is no negative position tolerance). Also realize that most position tolerances are diameters, which may be confusing if your gage readout only tracks x and y deviation. You would have convert the x-y reading to a radius by calculating a diagonal (Pythagorean theorem). Lastly, the Cpk calculation does get difficult if the MMC or LMC modifier is shown after the tolerance amount, because now there is a "bonus tolerance". There are three responses to this: 1) Why is Cpk data needed on a tolerance that allows bonus? The two ideas contradict each other--bonus implies that the fit is not critical, while Cpk data is usually for critical/significant items. 2) You could ignore the bonus effect and calculate Cpk using the stated tolerance only. 3) If you really must have Cpk data that represents the actual tolerances, then you can track how well the virtual condition is being met, because that is the one thing that IS constant for an MMC tolerance. Sorry to get so long-winded, but I wanted to cover all the bases! Feel free to post a reply if you would like more info on any of the terms or concepts mentioned...

Post Reply | Recommend | Alert

Rate

Where am I? Original Top of thread

| |