iSixSigma

DPMO

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #44313

    EBMA6 Sigma Guy
    Participant

    Can anyone suggest a method for a “weighted” DPMO? The reason I’m asking is we have several facilities most of which are manufacturing disposables but we are not. So our DPMO seems to look VERY large compared to high volume (we are one unit at a time). Any thoughts would be appreciated.
    Thanks,EBMA6

    0
    #141744

    boettler
    Member

    if it’s one unit at a time, why not focus first on DPU to make it more reasonable and less hybrid.

    0
    #141755

    villageidiot
    Member

    Its my understanding that DPMO can indeed to used as a weighted average.  Simply calculate it accordingly.  In addition, the DPU is a great alternative.  It is additive and thus, multiple measures from the different plants can be calculated and then averaged…DPU also allows you to determine RTY using the inverse log of your DPU…pretty slick and useful.  Either one works though. (double check this advice).  Good luck.

    0
    #141762

    Anonymous
    Guest

    EBMA6,
    Taoists observe the complementary nature of the universe, so that “the one that does the least is valued the most.”
    Now given the undoubted success of Motorla’s waferfabs in the late 1980’s, a reasonable person might suspect that Dr. Harry would be known in Austin! Yet, nothing is further from the truth!
    Therefore, we have to ask ourselves about the motivation of someone who couldn’t be bother to find out how world record yields were achieved. Did he report the great efforts in spatial yield analysis, failure analysis, and the use of multivariate methods – No!
    (Note: failure analysis included reverse engineering, ion beam probing, and cleave and etch-staining for SEM examination.)
    I wonder why? Yet, if you look at DPMO – what can you conclude? Only someone with the minimal amount of experience could ever come up with such a tin pot theory. Yet, even more arrogantly, they wanted to use it to measure the performance of other facilities – even those unrelated fields. Take the old Nasa thread for example, what a mindless waste of time. (A statistician called Post once referred to it as ‘mental master*** .)
    To give you some indication as to the size of this error, at one point in time a newly recruited Intel engineer reported that one Motorla wafer fab. produced 4 x more microprocessors than an equivalent Intel fab.
    Now to your question. DPMO assumes each defect is critical. It also assumes each defect is random and independent. The only type of defect that ‘fits’ this model is a random ‘point’ defect – one having no dimensions.How realistic is that?
    Therefore, I concur with the advice of the other posters. Don’t waste your time counting ‘opportunities’ – determine the area inspected.Count defects and categorise them into ‘point defects,’ line defects, and area defects. Define their seriousness – critical, reliability hazard, or cosmetic. Yes, it is challenging, but the objective is eliminate sources of variation.
    Andy

    0
    #141763

    Hans
    Participant

    Andy,
    Once again, very sharp! Thanks.

    0
    #141771

    EBMA6 Sigma Guy
    Participant

    WOW……Thanks for all the great resplies! I was thinking along the lines of “work centers / operation steps” so sounds like I may be on the right track. The down side…….I’m stuck wtih DPMO for atleast the next year (corporate mandate). I will also remember the “Taoist complementary nature of the universe!” Thanks Andy. Again thanks to all.
    Neil

    0
Viewing 6 posts - 1 through 6 (of 6 total)

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