iSixSigma

Calculate Control Limits from the sample itself?

Six Sigma – iSixSigma Forums Old Forums General Calculate Control Limits from the sample itself?

Viewing 14 posts - 1 through 14 (of 14 total)
  • Author
    Posts
  • #38736

    Ken Braithwaite
    Participant

    In our 6-sigma training we were taught that when using control charts the control limits are calculated from the sample you are testing. So if you are testing over a 1 day period you collect that test data, calculate the CLs from that data and then apply the tests using those CLs. Tomorrow you repeat, thus calculating new CLs with every data sample. This seems quite wrong to me. I would calculate CLs from tests taken when I thought the process si in control and compare subesequent samples to those calibrated CLs.  Does anyone have an authoritative answer?

    0
    #116394

    Mikel
    Member

    You are correct, and if you know a target value, your x-bar-bar should be set to the target value.
     

    0
    #116397

    Darth
    Participant

    I don’t know of any valid training that would tell you to keep computing control limits with every sample.  You collect data, establish some degree of stability and then continue monitoring against that baseline.
    Stan has this thing about monitoring against the target/spec rather than the calculated values from the data.  Mini gives you some options in this regard.  Personally, I would prefer to monitor against the actual process and do process capability as a separate exercise to determine how we are doing against the target.  While I agree that you might be in control but the process is way off target and therefore would be tempted to monitor against the target but I have found less problems by first stablizing the process and then trying to bring it to the target if needed.
    Stan, I assume you are back home so I will go wait by the mailbox for the promised Necter of the Agave.

    0
    #116402

    Mikel
    Member

    When are you going to learn about the target. Do I need to send you to remedial Loss Function school?
    Yes, I am back and since you sent the message to surprise you, I got you some stuff that was 80 pesos per quart. Is that okay? ;-)

    0
    #116403

    Darth
    Participant

    I am not familiar with the exchange rate…is that good?  If I recall it was once 1 peso = 1 dollar.  My sense, knowing you, is that it is closer to 1 dollar = 11.227 peso so 80 pesos equals ??????  What!!!! You are soooo cheap!!!!!  As a joke, my son-in-law did bring me back some white lightening that was about $5 a jug.  You know, it wasn’t bad….after a while.

    0
    #116404

    luke skywalker
    Participant

    It’s possible that your training was correct – if the idea was to allow the limits to be recalculated as additional samples are added to populate the chart. A control chart with a few plotted points is a rough guide – if it’s out of control, you can feel pretty sure that it is. If it appears to be in control, that’s harder to argue. As data is added via subgroup means (X-bar), the estimate of 3 standard errors gets better, so the control limits stabilize.
    That part of it focuses more on process variability, rather than ‘on-target’. Stan’s observation of loss function might suggest a process improvement methodology that simultaneously addresses being on target, with minimal variation.
    Hey Pappy, is Don Emilio the same as Don Julio?

    0
    #116407

    Darth
    Participant

    Hey Sonny, how be you?  If Don Emilio and Don Julio were the same they would be spelled the same, wouldn’t they?  Don Julio is distilled by Tequila Don Julio (NOM 1449) and located in Atotonilco. Don Emilio is produced by Tequila La Parrenita (NOM 1115) and is located in Guadalajara..  If you get a chance try their Penca Azul…very nice and a great looking bottle. 
    Interesting news about General Veers and the Maulster.  If interested, I can give you the inside scoop.  Maulster is with us now.  More on the way, I hope.

    0
    #116412

    luke skywalker
    Participant

    Interesting is a good word. Oddly the Star Wars thread has Luke as the last Jedi…

    0
    #116413

    Darth
    Participant

    You and Yoda may well be the last to die.  Can’t wait for the new episode to come out.  I have to see where I went wrong and how I turned into such a bad guy.  Wait it wasn’t my fault…it was the evil Princess of Overhead.  But as they say, one door opens as another one closes.  And this one is so much more professionally fulfilling if not as lucrative.

    0
    #116415

    Mikel
    Member

    Psssst – don’t tell anyone but I got you something that is very nice – about 1350 pesos.
    Now go learn about the wisdom of target instead of x bar bar or it willl be Everclear in the bottle by the time you see it.

    0
    #116909

    Sridhar
    Member

    Hi, Ken,
    I am a novice to six sigma but have been using control charts for quite some time. Let me attempt to understand your query.
    The very name of control chart implies it is for a process that is already in control. Having a specified tolerance limit from the customer (specification limits – SL), we can set an internal limit for our process parameter that is more stringent than the SL. Control limits (CL) are generated out of the natural variation in the process the output of which we track using the control charts. Tracking and comparing present performance using CL’s generated out of a standard or initial sample does not account for the present variation in the process.
    Any thoughts?
    -Sridhar
     

    0
    #116931

    Jonathon Andell
    Participant

    If Yoda lived so long, how come he never learned good grammar?Getting back to the charts: there are some charts for which the Control Limits are recalculated for each sample, because sample sizes vary from one subgroup to the next. However, the overall mean should no be changed. Instead, the control limits get wider (small sample size) or narrower (larger sample size) about a consistent mean.If your training suggested a new mean for every subgroup, get your money back.

    0
    #116943

    Ananthasubramaniam G
    Participant

    The missing link is, the process control study details  are also evidenced from the daily details of Measurements to assure you of process conformity.These are the same details used for Calculating Control Limits Too.
    As you are aware that  after every stoppage, Setting,& Tool Changes , the consequent  product measurements are used to verify the process .
    Hence,every time you start off tomorrow it necessiates verification of not only Process but also the Control limits

    0
    #117149

    marklamfu
    Participant

    The samples(subgroup) used to calculate control limit need come from different shift /operator , at least 20 subgroups is need, so, one day ‘s test-data is not enough.
    It is need to remind the control limit need to modify (CL may advise) if process capability is changed.
    Best regards
    Mark

    0
Viewing 14 posts - 1 through 14 (of 14 total)

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