iSixSigma

PPM and DPMO

Six Sigma – iSixSigma Forums Old Forums General PPM and DPMO

Viewing 16 posts - 1 through 16 (of 16 total)
  • Author
    Posts
  • #30168

    susu
    Member

    Are Parts Per Million (PPM) and DPMO (Defects Per Million Opportunity) the same thing? If they are not, what is the difference between them? In addition, how can I convert a certain Cpk value to a Sigma value that is used to calculate DPMO?
    Thanks!

    0
    #105461

    Harmendra
    Participant

    PPM and DPMO are the same in fact PPM is Synonymous for DPMO.

    0
    #105493

    Gabriel
    Participant

    I was about to say “Absolutely NO!”
    But then I read Harmendra’s post and he/she sais they are the same.
    Then I looked in the iSixSigma dictionary (in the bar at the top of this page) and both under DPMO and PPM it sais they are the same thing.
    So I will change my speech and say: “Absolutely NO!, and Harmendra and the iSixSigma dictionaries are wrong”
    For me, DPMO is defects per million of opportunities and PPM is defective parts per million parts.
    The differences are clear: A defective part can have one or more defects, and a part can have one or more opportunities.
    If we call X the number of defective units and N the total number of parts, and p the fraction of parts that are defective, then
    PPM=p*1,000,000=X/N*1,000,000.
    If we call D the number of defects (D equal or greater then X), Ot the total number of opportunities, Op the number of opportunities per part (Op equal or greater than 1), and DPU the average number of defects per unit, then
    DPMO=D/Ot*1,000,000=(DPU*N)/(Op*N)*1,000,00=DPU/Op*1,000,000.
    But I can be wrong.

    0
    #105498

    Nanni
    Participant

    Gabriel is totally correct!!!
    Actually PPM is NOT a six sigma metric.
     
    To caculate Cpk value to a Sigma value, you have this relationship:
    3*Cpk = Zst  (where Zst is Short Term)
    3*Ppk = Zlt   (where Zlt is Long Term)
     
    Then, Z is used to calculate DPMO or DPPM
     

    0
    #105500

    Mikel
    Member

    3*Cp = Zst

    0
    #105501

    Trev
    Member

    Gabriel,
    Where did you learn this understanding and difference between PPM and DPMO? What it on your job or in a course? Not that it matters, but I’m just curious where it originates from.
    Trev

    0
    #105504

    Sigmordial
    Member

    Gabriel,
    Nicely done! May want to consider requesting a change to the iSixSigma dictionary to include this. If you have time, you may also want to generate a technical paper — at least for iSixSigma — that provides further elaboration.  Include things like:

    The basics, like the differences between defects and defectives
    Why we need to distinguish between the 2
    Some potential issues with the current “just dpm/dpmo” mindset
     

    0
    #105505

    Schultz
    Participant

    I second that, Sigmordial. I’d love to read more from Gabriel on this topic.

    0
    #105506

    Gabriel
    Participant

    The concept of PPM is there well before we started talking about opportunities, DPMO and Six Sigma. It means Part Per Million and is allways used as Parts (with a specific characteristic) Per Million (of parts in total). Being a parts/parts ratio it is dimensionless (unlike DPMO which is defects/opportunities). I think that the first time I came across the PPM concept was during the secondary shcool, in the Chemestry course. You know, a solution of 250 PPM of alcohol in water means 250 parts of alcohol in 1,00,000 parts of solution. The concept remains unchanged to refer the ratio of defective parts.
    I came across DPMO about 2 1/2 years ago, when I visited this site for the first time and learnt that there was something called Six Sigma which included something called DPMO. Today I still find it too ambiguous because of the ambiguty to define what is one defect (if a board with 100 resistors gets burnt in the soldering process, how many defects are there? one or 301?)  and what is one opportunity (you can go from one: either it is right or wrong, to as many as your creativity goes).

    0
    #105513

    John H.
    Participant

    Gabriel
    Good Post. My preference would be a metric that is linked to the Taguchi Loss Function say for example Loss per failure. A product fails when it does not perform according to its intended function with  the Total Loss being distributed across the product components. In the Pharmaceutical Industry a the key phrase mentioned by the FDA in their regulations is that Drugs and Devices be “safe and effacious”. Just some thoughts.
    John H.
     
     

    0
    #105515

    Charmed SIX
    Participant

    I think Grosby (before  some  50 years) has created the ZD  concept,long  before the SS Motorola  concept.But unfortunatly NO BODY BELIEVES IT IS PRACTICAL AT THAT TIME.

    0
    #105518

    A.S.
    Participant

    In a mechanical enginerring industry practically we are assuming the oppotunities as one ie PPM is equivalent to DPMO since people will have confusion for deciding the no of oppotunities.I felt that this is as a practically feasible approach.
    In some cases considering the more no of opportunities will dilute the quality level ie artifically it will show high sigma level.
    Please go though this example :
    Situation 1 : Considering PPM
    No of cars produced —1000
    No of defectives ——–100
    PPM ——————–100000
    Sigma considering PPM—-2.7
    Situation 2 :Considering DPMO
    No of components in a car—–500
    No of opportunities for defect—-500 (min one defect per component)
    Total no of defective —-100 (Assuming only one defect in each car)
    DPU——100 / 1000——0.1
    DPMO —0.1 /500 *1000000—–200—–5 Sigma level
    The process is worst while considering PPM = DPMO but the same process is better while considering DPMO as a different one.
    DPMO with multiple oppotunities will mislead by showing the high sigma level.
    Extra care is required while talking about DPMO.
     
     
     

    0
    #105528

    Gülcin Kaygisiz
    Participant

    I know since 1996, that PPM and DPMO are not equal.
    If in a unit one defect, it’s enough to accept it, as a defect part. That’s why customer complaints are following up as a Metric PPM.
    DPMO is an oppurtinity. In one unit could be many defects. Because to produce one unit, we have to make many operations. And in all operation there are oppurtunities to produce defects. After completed a unit, during processes could be produced many defects.

    DPU= -LN(FTY)

    DPO= DPU/(#of OPPR)

    DPMO= DPO*1000000

    DPM= DPU*1000000
    For example:

    Significant Carac.

    Cp

    Cpk

    Zlt

    FTY

    p

    DPU

    DPO

    DPMO

    1

    2,59

    7,77

    1,00000

    0,00000

    0,00000

    0,00000

    0,00

    0
    #105529

    Anonymous
    Guest

    Dear Gülcin,
    In my opinion, the yield equation should be -LN(FTY)= DPU – LN(Y0)
    where Y0 is the yield loss due to systematic defects (correlation defects.)
    This equation was used by Motorola waferfabs in the mid-1980s to determine low yielding product families.
    Broadly speaking, the defect density in a facility is reasonably constant. Therefore, one would expect similar products to yield similarly, but this is not ususually the case, and this formula allowed us to identify many such cases and the causes through detailed failure analysis – causes such as such as reticle (masking) defects, design rule violations, incorrect alignment and registration, etc. Do yourself a favour and try it …
    It is easy to check on some grid paper, and by casting defects on to the paper – beads for point defects and matchsticks for line defects.
    Best regards,
    Andy

    0
    #105533

    Mikel
    Member

    It was Phil Crosby and it was about 45 years ago at Martin Marietta in Orlando, Fl.
    Nobody believed it because it lacked tools and methods.
    Crosby’s writings are valid and everyone should read Quality is Free.

    0
    #105541

    John M
    Participant

    Gabriel,
    I sit here shocked and stunned that people think this is a major breakthrough. Common sense should tell you parts per million refer to a part and dpmo is talking about opertunities per part. This is the most basic topics to understand what standards are people getting taught out there?
    If I manufactured two items
    1. cordless drill 2. a pencil . If  I had more parts per million defective in my drills it would be grossly unfair to say it was worse than the pencils.
    To gain a like for like comparrison I would then use DPMO This would take the complexity of the drill and evenly balance it on the scale with the pencil allowing me to make judgements on the items.
    You are 100% correct but  shocking that that is not widely recognised.
    All The Best
    John

    0
Viewing 16 posts - 1 through 16 (of 16 total)

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