the below link may be explain it for you , then it will make sense for you
The sigma level determined through the std dev. it has a levels from 1 to 6 , but the it is indicating the DPMO of the process.
So if the process performance at 2 Sigma the DPMO is 308,537 whoever it is 66,807 at 3 sigma
to determine at which level we should set our process we need to know what is the required specification , process capability (cpk) and the performance of the process at which sigma level . So after that you can determine to which sigma you can go.
hope this is make it sounds good , and make sense !,
No this is not what I'm talking about. The dpmo is just a fancy way of talking about a normal distribution but in any case that still doesn't help me. So, say, for example, a pack weighs 200g and the standard deviation of a given pack when passed over for a ZOI check (sigma) is an entirely acceptable 0.5g. The standard deviation is not controlled by anything in your process but simply the variability of your checkweigher. This might be due to all kinds of things which are controllable (decent set up) but mostly uncontrollable.
To have a decent assurance that the weight is real within the bounds of your variability introduced via the uncertainty in your checkweigher; you should really say a weight is correct passing over a checkweigher if it is within +/- 3 sigma which is approx. 95% of the data. That, however, includes weights from 198.5 to 201.5g. So if it is weighing 1.5g heavy, my check would work but an average weight on my line of 200.5g would actually be illegal because the real average weight was 199.0g.
What I'm saying, perhaps inarticulately, is that it's nothing to do with your process variability, that is not my question, what it is, is that due to the natural variability of checkweighers, you cannot do an accurate calibration check because you do not know what is due to checkweigher variability and what is to do with being out of calibration. The only way you can do it is to pass the same pack over the checkweigher again and again (i.e. do a ZOI check) and check the average corresponds with the weight you expect but that is a massive time commitment. Retailer codes of practice demand that you do a calibration check daily but how people generally do it is to do a static weight check which, as I've already explained, is a complete waste of time...