If my customer request a specific product characteristic spec (say < 0.3um),how do we use SPC or other stat tools to decide product spec appropriately? By the way,this product is a new one so that lacking historical data base.

If my customer request a specific product characteristic spec (say < 0.3um),how do we use SPC or other stat tools to decide product spec appropriately? By the way,this product is a new one so that lacking historical data base. Your customer determined the product spec. You now must develop a process to meet the customer requirement. You can use statistical techniiques to determine if your process is in control and capable after you have the data from the process. You might want to do some searches in the cove and google for additional information.

ysdennis:

This is a kind of "chicken or the egg" question - which comes first?

In order to have an idea what your process can truly do, you have to actually run the process, to get some historical data. Then, you can do Cpks for various specs until you get one you are comfortable with. I generally like > 1.33 as a minimum, with at least 3 for a good comfort level.

That said, we have a few specs that are 100+, and I won't offer to change them unless the customer brings it up!!!

So, would it be possible to get at least SOME data from the process? That would be a start, anyway.

No single index can be universally applied to all process and no given process can be completely described by a single index.
If the specification is inappropriate effort may be wasted in analysis

To identify the real characteristic condiction the better way is to identify the natural limit, average +- 3 sigma followed by reliability study the specification can de definded appropriately
Thanks
JAIRO BRANDAO - BRAZIL

No single index can be universally applied to all process and no given process can be completely described by a single index.
If the specification is inappropriate effort may be wasted in analysis

To identify the real characteristic condiction the better way is to identify the natural limit, average +- 3 sigma followed by reliability study the specification can de definded appropriately
Thanks
JAIRO BRANDAO - BRAZIL


Hi, Jairo, and welcome!

You make some valid points. I would point to one-sided specs and specs with natural limits (where a value less than 0 is impossible for instance) as examples where using "one size fits all" techniques can lead to trouble.

I ysdennis' situation, though, with little or no history, it will be difficult to do any meaningful analysis, agreed?

I would like to re-iterate the point about the specification is separate from SPC. Dr. Wheeler uses the phrase "Voice of the Customer" in regards to the specification. It is the customer's request, has nothing to do with SPC. Then comes the "Voice of the Process" which the SPC chart represents. That lets you know how well and how reliably you are answering the specification. It will also tell you the effects of changes on the process with respect to achievement of the specification.