2 Replies Latest reply on Oct 11, 2012 2:13 PM by mhamel

    SPC Charts With Slow Moving Data


      I am trying to develop a Processbook display that shows a variety of data points from a machine. I'm using X-Bar charts, taking 90 samples before end time, and each sample has a 0:05 calculation period and a 0:02 sample period with a sample size of 2.


      My goal is with 90 samples and 0:05 calculation period, my charts are all approximately a 450 second slice of time.


      For data points that change very little over a long period of time, I get a straight line, but if there is any sudden change in value, instead of drawing a line from the last data point to the new, lower or higher value, it shifts the entire line down or up.


      I think this occurs when the change is very small, for example a change from 0.255 to 0.251. My SPC chart is set up with a range that can display such changes, but instead of charting them it's shifting the entire line up and down - making the previous datapoints inaccurate.


      Could this be somehow related to the fact that the value isn't changing enough to produce an actual "raw" PI data point due to compression? Is it a bug in SPC charts?


      Thanks for any insight you can offer.



        • Re: SPC Charts With Slow Moving Data

          Ahh...it looks like if your tag compression deviation is set too large, these charts won't work. You need the raw PI data points.

            • Re: SPC Charts With Slow Moving Data

              @Ryan: The sampling process takes account of the weight of events due to the compression. If you get long period of time with no changes, the last period of time will have more "weight" than the others and will affect your X-Bar and R calculation. The goal of the X-Bar chart is to show average values of samples taken at regular interval from the process which are linked together by preserving the chronology of events. This process creates a kind of filter (like compression) to better evaluate how the process performs. If you apply the filter after another filter, what you see might not be the same. This is why you came to the conclusion you need to get the raw data mostly.


              Just to add up more information. This type of chart is only valid if the within-sample variability is constant. You need to validate the R chart before to see if the process is in control. If it is the case, you can examine the X-Bar chart to see if the sample mean is also in statistical control. If on the other hand, the sample variability is not in statistical control, then the entire process is judged to be not in statistical control regardless of what the X-Bar chart indicates.


              I hope this helped.