Raymond – I’ve never yet used one of these, but have been able to track down the following references which may be of help (or they may not) “Implementing Six Sigma” by Forrest W. Breyfogle III has a large-ish section on CUSUM charts. MiniTab’s Help section also goes into some detail (with worked examples) if you have access to a copy of that.
As I understand it your centre line should be a zero value, i.e. you are looking at the cumulative sums of the deviation of your samples from a target value. No deviation from target, no problem, hence value = 0.
UCL and LCL are +/- 3 sigma from this 0 value, and the sigma is estimated from MRbar/d2 (as per usual guesstimates of sigma) so you will need some historical (hysterical??) data to start with to get a current sigma value to work with. The moving range is of length 2, since consecutive values have the greatest chance of being alike.
None of the formulae I’ve looked at refer to ‘alpha’ but I’m guessing here that it may refer to an acceptable quality level – so is there an acceptable reject level as well in your formula? If this alpha is the case, then you are probably using the chart to determine incoming parts quality.
Now I’ve stuck my neck on the block, is anyone else out there prepared to offer any further constructive comment to help Raymond out on CUSUM charts??
James A
.
