iSixSigma

Focused Question

Six Sigma – iSixSigma Forums Old Forums General Focused Question

Viewing 15 posts - 1 through 15 (of 15 total)
  • Author
    Posts
  • #36501

    H82BLATE
    Participant

    Which is most important in terms of improving quality:
    Capability, Complexity or Control. Capability can be
    expressed as a Z value and control as the extent of shift
    and drift, while complexity can be described as the total
    number of opportunities. Which of these factors is most
    important in terms of reducing total defects per unit? And
    why?

    0
    #105448

    Mikel
    Member

    The answer is yes.
    And wow – yet another screen name for the elder, bald statesman for the desert.

    0
    #105452

    Gabriel
    Participant

    The answer is it depends. What is spoiling your units more? Lack of capability, lack of control, or too much complexity? The answer will not be the same for every product/process.

    0
    #105454

    Mikel
    Member

    Go to chapter 12 of The Vision of 6S (cartoon version) for a shallow discussion of the same.

    0
    #105458

    Anonymous
    Guest

    I only use ‘opportunities’ in the sense that I would like to know how many parts there are in a batch process, or on a wafer. In other words, to determine ‘yield’ or the percent defective. Trying to count defects is frought with danger because they can be higly correlated.
    For example, do we count every ‘contact’ on semiconductor wafers as an ‘opportunity?’ What happens when an etch is incomplete and whole areas on the wafer are ‘open’ or have ‘high contact resistance’ which affects device speed? Should we report defect density in the millions?
    If this by itself is not sufficient reason not to use ‘opportunties’ in the sense of complexity … it is worth noting that complexity can actually improve yields and quality – Japanese engineers have been doing it for years!
    Examples of how increased complexity can actually improve quality and yields are abundant in Taguchi Methods case studies. One well-known example is how two extra transistors in a circuit ( a potential divider) can make products less sensitive to manufacturing variation in gm or hfe.
    Basically, it is my belief that many of the tools and methods within Six Sigma are questionable and are in desperate need revision – particularly those introduced after 1990 by ‘you know who ..’
    Good luck,
    Andy
     

    0
    #105468

    Mikel
    Member

    Andy,
    I have been on my good behavior lately, but it’s tough with such baiting comments.
    Just a clarification on the tools and methods. The statistical methods are valid – it’s the dogma associated with Six Sigma especially as defined by ‘you know who ..” and espoused so dogmatically by ‘you know who and now his alter ego daughter ..” that need to go.
    A statistically literate workforce is good. A common language is good. Minds open to new tools where needed is good. That is Six Sigma at its heart. Doesn’t have anything to do with the nonsense marketed these days as Six Sigma.
    By the way, I like the Lee book.

    0
    #105472

    Anonymous
    Guest

    Stan,
    It’s always good to hear from you .. whether it is ‘for’ or ‘against.’
    I’m glad you enjoyed the book .. I wonder if it does explain why kanbans work ..’an excessive backward flow of cards to cause a forward flow of parts’
    As you know, I’m quite interested in quality complements and it’s interesting to note that both our governments have increased interest rates, at a time when manufacturing is still struggling. Of course, unlike obvious complements such as: ‘refraction and reflection’ It can be quite difficult to prove that ‘lending and risk’ are complementary.
    I guess the bottom line is whether or not there might be a theory of quality complements, and whether or not it can be applied to ‘decision analysis’ at some time in the future.
    Best wishes,
    Andy

    0
    #105479

    Reigle Stewart
    Participant

    H82BLATE: As I understand your question, you can
    “discover” an answer by first defining each of the 3C’s at
    two levels (low and high) and then conduct a factorial
    experiment on the eight combinations. The response
    variable would be TPDU. For example, capability can be
    set at Z.st = 3 and Z.shift = 1.0, thereby providing a long-
    term expectation of Z.lt = 2.0. Next, convert Z.lt into a first-
    time yield value (Y.ft) using a table of area under the
    normal curve (or use “=normsdist(Z.lt)” in Excel).
    Following this conversion, take the resultant yield value
    (Y.ft) and opportunity count (M) to establish the rolled-
    throughput yield. This is done by computing Y.rt = Y.ft^m,
    where M is the defined number of opportunities. After
    this, simply compute the quantity TDPU = -ln(Y.rt). Do this
    proceedure for each of the eight combinations of the
    factorial experiment. You will discover that when Z.st is
    low, the influence of Z.shift and complexity is high. But as
    Z.st increases in value, the TPDU becomes robust to shift
    and complexity. Also pay attention to the interactions. By
    examining the output tables and graphs (in Minitab), the
    conclusions will be fairly self-evident. Furthermore, it is
    interesting to extend this experiment using other DOE
    designs.

    0
    #105480

    Mikel
    Member

    As always, it is fun to watch you ask and then later answer the question. Maybe on to the other identities can ask you an interesting follow on to your “interesting” response.

    0
    #105481

    Anonymous
    Guest

    What nonesense …

    0
    #105482

    Mikel
    Member

    Nonvalue added as well especially since he asked the question in the first place.

    0
    #105483

    Reigle Stewart
    Participant

    Perhaps some fact-based insights can be realized by
    running the suggested experiment. In this way, you too
    can advance your application and theoretical knowledge
    about Six Sigma and not be so dependent upon the
    research and work of others. You might event find or
    discover some new knowledge. Judging by your
    responses in this thread, it would seem you still don’t
    know the answer to H82BLATE’s question. Maybe you
    have a better way to analytically investigate the question
    and set forth insights (other than qualitative opinions). If
    you need help running the experiment, please post and I
    will guide you through it. Respectfully, Reigle Stewart

    0
    #105485

    Heebeegeebee BB
    Participant

    heh heh…I’m gettin’ some popcorn…
    this outta be entertaining!
    Go Stan!

    0
    #105488

    Mikel
    Member

    Oh but we do know the answer to the question –
    Answer 1 – What a stupid question, we seek to have the least complexity needed to give the desired functionality and reliability, we seek to have the best capability possible given the resources we have, and we look to have excellent controls in place and fire any operations person that cannot do better than the 1.5 shift you are so fond of.
    Answer 2 – What a stupid question and that DOE answer is pretty stupid too. It is sad that you set up questions just so you can come back later with some confusing answer to have the less informed be in awe of your knowledge.
    You may want to note that I called that it was you asking the question under yet another name 8 minutes after you posted it.

    0
    #105490

    Anonymous
    Guest

    Stan
    This has to be one of the worse cases of armchair engineering I’ve ever come across. As for the ‘projection’ of someone with  limited practical experience on to you is all the more ironic given the substance of the reply and the recent confusion between the quality complements: anima and animus.
    Cheers,
    Andy

    0
Viewing 15 posts - 1 through 15 (of 15 total)

The forum ‘General’ is closed to new topics and replies.