iSixSigma

Gage RR for buy-off new vision machines

Six Sigma – iSixSigma Forums Old Forums General Gage RR for buy-off new vision machines

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #42023

    howe
    Participant

    Hi,
    We are buying a new equipment (cutting system + vision system). We ran a Gage R&R on it, resulting on a R&R (% study var, Minitab)value of 52%. However, if I we add the tolerance (USL-LSL) the resulting P/T (% Tolerance SV/Tolerance, Minitab) is 8%.
    My opinion is that the equipment is good for the application since it will be able to sort parts at no risk since specs are loosen enough. However, continuos improvement efforts for reducing variation will faceΒ the vision system as the major source of variation, so considering is a brand new equipment I prefer to do some improvements prior to buy it.
    Any comment or idea is welcome,

    0
    #132470

    Scott
    Member

    Hi there,
    Very good question, but can you give some more information on how you did the actual gage R&R.
    The original SV of 52% is way out of line, and I do not really understand why just adding USL and LSS specs would have changed your final results so drastically, hence my question about how you did the actual study.
    If the major source of variation is your vision system, then it is indeed a wise idea to pinpoint that variation and do the improvements before you buy the tooling. It is great experience gain for your supplier because he will gain more knowledge about the machine and enable better products in the future. For you it is good because you don’t need to spend money on that issue later on.
    Regards,Rick

    0
    #132471

    Peppe
    Participant

    A very good system to improve R&R of vision system is,Β obviously, to have a software that manage masks on vision system to analize image at different levels (reject, acceptable, pass, etc…).We did it with very good results.
    Rgs, Peppe

    0
    #132472

    howe
    Participant

    HI,
    Thanks a lot for your answer,
    The SV%(R&R%) is the same no matter the Tolerance used. However, there is some discussion regarding usingΒ SV / Tolerance (P/T) or SV% (R&R%). Β 
    I believe both need to be below 10%, but some people says that they want to rely only on SV /Tolerance (P/T).

    0
    #132475

    Scott
    Member

    Peppe,
    Your improvement might very well work with static images, where the sample presented to the vision system is always positioned in the same way, and that is surely the system you are using or mentioning.
    In my factory we make carseats, these seats are never the same. We used to use a static system with reasonable results to check airbaglabels at covers. Now those labels have changed into little ‘flags’ sticking out of the cover in whatever angle you can imagine.
    We had to get rid of our old vision system and start with a dynamic one with learning capabilities. It is a Siemens based system who are using software that is conceived and written at a USA based university.
    This remark only to mention that you should always look at the usage of the system before deciding on what system to use. In my opinion investment in static system will not pay off anymore. Even in static environment the dynamic ‘learning” system score far better than the static ones. It’s really worth the investment. If we had known at our plant we would have started with the dynamic one right away. It would have saved us several ten thousand dollars on money that is now thrown away at an unused static vision system….
    Regards,Rik.

    0
    #132476

    Scott
    Member

    Mike,
    Could you please send me the output of your Gage study, because I am using Minitab too and I can’t seem to find the column or heading you are talking about in you question.
    It’s clear that the study variance SV percentage (%Study Var (%SV) column should read less than 10% on the Total Gage R&R line.
    Also look for the line ‘Number of Distinct Categories =’ it should be more than 6 too. Remember that you Range chart per operator in the graphical output should be in control (you don’t want data discrimination), and you XBar by operator should be out of control (you want more than the normal measuring range covered). A straight horizontal line on the value per operator sign is also a good thing…
    If one of those (except last straight line) is not the case than you can throw away your result because they aren’t worth a thing. The gage is totally unacceptable for the reasons I have summed up. Everything needs to be OK. Some people tend to find excuses in order to talk them out, but what they are doing is fooling themselves.
    I work for 2 auto constuctors in Europe (cannot disclose names), but the are all two German brands. I can tell you that the first time I would bring a result to them that didn’t fit the set out standards on all points, I would be butchered even during my presentation. Rules are ment to be followed and if one of them isn’t OK the total isn’t right.
    Remember you are going to use this gage in order to do DOEs for example. Ask yourself this questions. Suppose you will have to make million dollar decisions on the output of your DOE knowing that the gage you were using isn’t realy a good gage ?
    All I want to say is send me the output, so I can have a look. Preferably the graphical one too which you cannot send me this way, but tell me how they look like in words.
    I’ll try to get you out of this, with the knowledge I have. From what you have told me soo far your gage is unacceptable.
    Regards,Rick.

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

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