iSixSigma

DPMO and Sigma Level

Six Sigma – iSixSigma Forums General Forums General DPMO and Sigma Level

Viewing 13 posts - 1 through 13 (of 13 total)
  • Author
    Posts
  • #53577

    Rohatgi
    Participant

    Let us assume a PCB has 1000 solder points.
    10000 such PCB makes it 1-million opportunites.

    For 6-sigma process there would be 3.4 defects / million.
    So if there are 3 solder defects in 10,000 PCB the process is 6-sigma.
    But these 3-defects can happen in any PCB which means there is a possibility that there are 3-PCB found defective which translates to 3-defects per 10,000 PCB.

    So, if we look at PCB level, the process is no longer at 6-Sigma Level.

    What are your opinions ?

    0
    #190755

    Mikel
    Member

    My opinion is this is a stupid question.

    0
    #190758

    Rohatgi
    Participant

    Questions are never Stupid.
    Yes ! Answering can be in a Stupid ways

    0
    #190762

    Strayer
    Participant

    Not a stupid question! It depends on what you’re measuring. If you are measuring process sigma (DPMO where O is derived from the number of opportunities for defect to occur in each iteration of the process, e.g. 1000 solder points) your example would be a 6 sigma process. If you are measuring product sigma (DPMU where U is the number of output units) you find that it is not six sigma since you have a smaller denominator. If you are measuring defectives, you could have 1, 2, or 3 defectives per 10,000 units. All three defects could be an a single unit giving you just 1 defective, but still more than 3.4 defectives per million units.

    DPMO is internal process focussed. What the customer really cares about is defectives. Accordingly I insist that defectives per million units is the most critical measure of quality. There are plenty of examples where the producer and the customer differ on meausurement of quality. The producer may be able to show that they have a six sigma process but the customer is seeing many more than 3.4 defectives per million units. Don’t argue with the customer.

    0
    #190764

    Rohatgi
    Participant

    Hi Straydog,

    Thanks for your reply. It clarifies DPMO , DPMU but it raises one more question.
    Customer is interested in DPMU which means Sigma Level should be measure at the Last Process Step.

    Which makes it similar to RTY.
    That means in a 10-Stage process each prior stage should be greater than 6-Sigma level.

    End product to customer is combination of many set of products. e.g. TV+Remote.
    So if TV & Remote assembly line produces 6-Sigma Level,
    Then the final sigma level to customer would be ADDITIVE or MULTIPLICATIVE of both these Sigma levels

    Awaiting you reply.

    0
    #190765

    Mikel
    Member

    There are lots of stupid questions, don’t let anyone kid you about that.

    Your question and your further questions show a absolute lack of any research on your part. You just sit back and dream up crazy questions and instead of learning you throw them out like you deserve respect for lack of effort.

    Go learn something and come back with rational questions.

    0
    #190773

    Strayer
    Participant

    There’s a reasonable assumption that reducing DPMO must also reduce DPMU. You may not be able to say whether it’s additive or multiplicative since mulitple defects may appear in a single unit (DPU). If you yave 3 delivered defects you could have anywhere from 1 to 3 defective units. If you really want to understand the 6s-DPMO metric, pick up Armand Feigenbaum’s book “Total Quality Contol”. For a criticism on whether it makes sense see Arthur M. Schneiderman’s posted article: http://www.schneiderman.com/The_Art_of_PM/six_sigma_metric/six_sigma_metric.htm
    I emphasize customer focussed metrics. In the end, all internally focussed metrics are delusional if they differ from what the customer is seeing. From your original question I can guess that this is also your concern. Cheers.

    0
    #190774

    Rohatgi
    Participant

    Straydog,

    Thanks ! You are right, we should look at it from customer point of view.
    So focus on those process steps which leads to customer perceived defects

    Anupam

    0
    #190776

    Mikel
    Member

    Man you recommend boring reading.

    Focusing solely on customer metrics is delusional. It is driving looking only at your rear view mirror. The data is slow and unreliable.

    The reason for the internal focus is to have a more timely indicator. The things you see in your operation are the things your customer sees. You balance this by making sure that as you drive internal metrics the external metrics follow (but lag).

    It’s QTC – Quality, time and cost.

    Internal quality – DPMO
    External quality – returns and complaints

    Internal time – cycle time
    External time – delivery complete to customer’s requested time

    Cost is total cost – you have to comprehend the entire value chain

    0
    #190777

    Mikel
    Member

    rabos,

    You’ve got it all wrong. Focus is on steps that are value added to the customer, it has nothing to do with defects.

    Your approach has people focusing on containment – the same stupid s**t people have been doing for 50 years.

    0
    #190784

    Strayer
    Participant

    Stan, thanks for contributing your wisdom. I actually agree with you and I apologize to anyone else who may have thought I was recommending DPMU over DPMO. Complaints about delivered defects/defectives only tell us that something is wrong. The quandry in the original post was that there can be a disconnect between our internal quality metrics and what the customer sees — the point that Schneiderman makes and explains in his article. So let’s accept that 6s based on DPMO may not mean that the customer is seeing that level of quality and listen to VOC to tell us we still need to do better.

    0
    #190785

    Mikel
    Member

    I agree 100%

    0
    #190786

    Breytenbach
    Participant

    I’m a new subscriber to iSixSigma. Would like to read the original article. Where is this, and other articles, published?

    0
Viewing 13 posts - 1 through 13 (of 13 total)

You must be logged in to reply to this topic.