How to Determine Max Test Specification Limit

Six Sigma – iSixSigma Forums General Forums Tools & Templates How to Determine Max Test Specification Limit

This topic contains 5 replies, has 6 voices, and was last updated by  Roger Hill 1 month, 3 weeks ago.

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
  • #56115


    Hello quality friends
    I am pretty green on quality and need your help on a particular issue. I am trying to determine the maximum breakage force allowed on a component during IQC check. The question I have is that should I conduct a breakage capability study on the sample quantity and select the minimum value as the max allowable breakage or what would the right way to determine that spec.?



    Spec limits are set by the customer, or by the next step in the process. Your question should be about the best way to measure it, not how to determine it.


    Chris Seider

    Here’s a thought….IF you have failures related to the component downstream in the process, do a process capability (if possible) of the component that was in the failed part/process and see what it speaks to you. Compare that to the process capability of your process. This approach MAY assist you in helping the process determine some initial specs if no other knowledge exists.


    Peder Skov-Hansen

    As mentioned above you should determine what level of failure you (or your customer) can accept. The capability index Cpk is one way. You may want to look into what the part is used for, if there are other considerations that are equally important other than just minimum breakage force. Check up on MTBF and DPMO on this site. But it sounds like you just need a simple breakage strength test so don’t overengineer it.

    But first you need to perform a measurement system analysis (MSA) in order to assess the influence of your measurement system. Google or isixsigma is your friend here. And Minitab :)

    For your test plan the trick is to make sure you include all plausible sources of variations, so things like mould batches or production machines will not give reason for higher variation in the future.

    You want to determine a good sample size for testing, because you can’t sell tested parts. Here you can check the power of your analysis, again Minitab is your friend. Power gives you the strength of your prediction. Lots of good links to articles here on that too.

    During production you should test the strength regularly. Here SPC is a good tool, it can help detect variation that is not caused by common causes.

    A note: Breakage force often follow the Weibull distribution. That is for your calculation of variation.

    BR, Peder



    @truong, if we think about an output Y:

    -The Upper Specification Limit (USL) is the highest level of Y the customer will accept, based only on customer requirements (here, your customer seems indeed to be the next process step). It is not obtained through calculation.
    -The Upper Control Limit (UCL) is the highest level of Y at which we consider that the process will still be in control. It is a calculation based only on the values of a set of consecutive data points and the moving range.

    Your UCL needs to be lower than your USL to make sure the output Y is within specification.

    It looks like your current approach is that you are looking at your UCL to define your USL – you should instead be defining your USL based on customer requirements and making sure your UCL is lower than it.


    Roger Hill

    Just a comment that echos what others have said. Is this effort for new product introduction? I worked at a company where design engineering regularly designed product made with “special processes” like bonding and welding. I could never get them to define the strength requirement (or get their management to see this as a problem). They expected manufacturing to “just figure it out and make good parts”. It all starts with the requirement (what the customer needs).

    Is this part of new product development or is this a test to monitor supplier manufacturing performance?

Viewing 6 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic.