When is an MSA should be performed

Six Sigma – iSixSigma Forums Old Forums General When is an MSA should be performed

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
  • #46723


    Does any body know when an MSA should be performed. Should an MSA be performed only on the X’s or could it be performed on Y’s also
    Thanks in advance



    In my opinion, an MSA should be performed on all projects. How do you know if you should base any business decisions on your analysis if you don’t even know if the data itself is good. The measurement systems analysis will tell you that.
    *At least*, every BB should think about this in every project.


    Adrian P. Smith

    Hi Frank
    MSA should be performed on the Ys and the Xs.



    The most important are the y’s. This reflects what the customer sees. To be honest, I have never done an MSA study on an X, and would be interested in an example.



    Ideally MSA should be conducted on all output & Input variables. But before doing that, it is better to look at all your X & Ys and find out source of data , if data is automated, you really dont need an MSA, but if there is manual intervention, or machine precisions involved, you should Ideally be conducting an MSA on the variable be it X or a Y…



    An MSA should always be done, but it does not always mean a full GR&R, etc.   The idea is too simply confirm the validity of your MS (hence your data) and the degree to which it adds variation into your measurements.  This can be a full blown GR&R along with stability and bias assessments or simply confirming that the data from your automated source is reliable…..but your MS should always be assessed….the method is based on what is most appropriate for your situation…..good luck.


    Jim Shelor

    In general, when starting up a new system, a full MSA should be performed on the measurement system that you use to confirm the final inspection of your product as a minimum.  This includes a Gage R&R as well as a gage study for Bias and Linearity.  For Example:

    Prior to the DOE to optimize the system parameters,
    Prior to operational testing of the system, and
    Prior to putting the system in actual operation.
    During operation, an MSA should be conducted,

    Following any maintenance on the measurement system, including gage calibration.
    Any time you have a change in your system capability that you cannot determine a cause for.  I did not say anytime you have a special cause variation, I said anytime you have a change in capability which means a change in your common cause variation.
    For other measurement systems in your process, it is a judgment call whether an MSA is required or not.  For example:

    An input parameter that has a specification limit (USL, LSL) and is important to quality should be considered for an MSA.
    An intermediate output/input (during the process) that is measured because it is important to quality should be considered for an MSA.
    If the measurement system for these parameters are determined to need an MSA, the same “startup and during operation” MSAs should be conducted.
    Contrary to popular belief, automated measurements also need MSAs.  Even in an automated measurement system, there is a sensor somewhere that can have the similar bias, linearity, repeatability, and reproducibility issues as measuring systems that are not automated.  These issues are generally smaller, but if you are using an automated system for your measurement system, you are generally looking at smaller specifications as well.
    It is not necessary to do an MSA on every parameter, but an MSA should be considered for all parameters and measurement systems that are quality effecting.
    I hope this helps,
    Jim Shelor

Viewing 7 posts - 1 through 7 (of 7 total)

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