How to select the requirements for Continuous Sampling Plan (CSP-1&F), Mil-Std 1235C?

[CK lau]

Continuous sampling plan, CSP. Can anyone explain to me by illustration? Thanks in advance.
Cheerio!

[Ajit Basrur]

In Reply to Parent Post by CK lau

Continuous sampling plan, CSP. Can anyone explain to me by illustration? Thanks in advance.
Cheerio!

Can you pl let us know what is i & f ?

[Tim Folkerts]

I am also not quite sure either what "i" and "f" are, but I can explain the basics of a continuous sampling plan. In a nut shell, a continuous sampling plan tries to make decisions with the least possible samples.

All sampling plans are basically a compromise between the producer (who wants to see that good lots are rarely rejected) and the consumer (who wants to see that bad lots are usually rejected).


For example, suppose you wanted a plan that accept accept lots with 1% defects 95% of the time and reject lots with 5% defects 95% of the time. In other words:
Consumer: Only on rare occasions am I willing to accept a lot that is 5% defective.
Producer: Only on rare occasions will you you reject a lot that actually is 1% defective.

One way to do this is to choose a basic single sampling plan. In this case, you could end up with a plan where you sample 190 items, accepting the lot with 4 defects and rejecting the lot with 5 defects.

But there is no need often to do that big of a sample. As a very simple example, if you found those 5 defects after looking at 20 samples, there is no need to test the other 170 items. You already know you can reject the lot.

For a continuous sampling plan, you make a decision after every item tested. The idea is something like this ...

Item 1
Producer: If you find 1 defect after checking only 1 item, that would be pretty rare if the defect rate is 1%, so I will allow you to reject the lot.

Item 2
Producer: If you find 1 defect after checking only 2 items, that would be pretty rare if the defect rate is 1%, so I will allow you to reject the lot.

....

Item 6
Producer: Hold on now. If you find 1 defect after checking 6 items, that would not be so rare if the defect rate is 1% (it could easily happen 5% of the time), so I won't let you reject the lot. You would have to find 2 out of 6.

...

Item 35
Producer: If you find 2 defect after checking only 25 items, that would be pretty rare if the defect rate is 1%, so I will allow you to reject the lot.

Item 36
Producer: Hold on now. If you find 2 defect after checking 36 items, that would not be so rare if the defect rate is 1% (it could easily happen 5% of the time), so I won't let you reject the lot. You would have to find 3 out of 36.

.....

Item 58
Consumer: If I find 0 defects after checking 58 items, that could still happen occasionally if the defect rate is 5%, so I am not yet willing to accept the lot.

Item 59
Consumer: If I find 0 defects after checking 59 items, that would be rare if the defect rate is 5% (it would happen less than 5% of the time), so I will accept the lot.

...

Item 82
Producer: If you find 3 defect after checking only 82 items, that would be pretty rare if the defect rate is 1%, so I will allow you to reject the lot.

Item 83
Producer: Hold on now. If you find 3 defect after checking 83 items, that would not be so rare if the defect rate is 1% (it could easily happen 5% of the time), so I won't let you reject the lot. You would have to find 4 out of 83.

..... and so on ......


In summary
1- 5 items sampled, reject with 1 defect
6 -35 items sampled, reject with 2 defects
36-82 items sampled, reject with 3 defects
83-137 items sampled, reject with 4 defects
138-199 items sampled, reject with 5 defects

59 or more, accept with 0 defects
93 or more, accept with 1 defects
124 or more, accept with 2 defects
152 or more, accept with 3 defects
181 or more, accept with 4 defects

Once the two numbers agree, you can stop. In this case, both the consumer and producer are willing to accept 4 defects and reject 5 defects once you have tested 181 items Up until that point there is a gap - a lot might be neither good enough to accept or good enough to reject.

ADVANTAGE: fewest possible items to test; easy to design the sampling plan.
DISADVANTAGE: more to remember for people doing the testing; hard to design the sampling plan.


Tim F

[CK lau]

Thank you.

Please refer to MIL-STD-1235C 1988.
There are 5 different types of continuous sampling plan, namely;
1. CSP-1
2. CSP-F
3. CSP-2
4.CSP-T &
5. CSP-V

I need to apply is CSP to my subcontractors/Suppliers, who the companies are ISO certified and the quality level is rather high. By our current practice, ie. ANZI 1.4 AQL 0.65%(A0/R1) and 2.5% (A1/R2) Lot size range from 6 to 14,000 pcs per lot, & 10K~14K lots every month, the sample size is 20pcs/lot, we have tabulated that 5 to 10 PPM were detected in Jan~June 2008. And I need 8 QC inspectors round the clock operation.
Do you have any suggestion?