Getting the most out of your Control Limits

Control limits are very important tools in winning the battle against process variation.  First let’s look at the different limits available in Quality Window and their roles.

1. 

   Targets (green) – the value you would ideally like to maintain for quality, performance and cost reasons.
2. Warning Limits (yellow) – these indicate values that are now closer to a Control Limit rather than the Target value – the first symptom of variation.
3. Control Limits (red) – usually based on Process capability and act as indicators of when a process is outside of its capable/normal performance range and should be reviewed to determine cause and action to take.
4. Specification Limits (white) – indicates a value is outside acceptable limits and immediate action should be taken.

Now, each limit within a Quality Window Application can be set as one of the following:

• Blank – no limits
• Calculated – typically using a multiple of the Standard Deviation of the Population but with options available (see below)
• Fixed – set manually

This discussion will focus on the options available for calculated limits and espeically when used for Control Limits.

Part of any ongoing continious improvement process is to get better – minimize variation – and then adjust or re-fix your control limits accordingly.  This can be a long and ardurous process if you are monitoring hundreds or thousands of variables in your process or product.

As I mentioned earlier there are three types of limits and I would recommend setting Control Limits to calculated.  Now, this alone is not enough to ensure reasonable and realistic limits are derived.

A setting just above the Limits window labelled “Calc. Limits Based On” gives you the additional capability to calculate the limits more accurately.  The options here are:

• Population
• MR Estimate
• Low Variation
• High Variation
• Low Average
• High Average

I would suggest selecting Low Variation.  What this will do is scan the data for the variable and select 20 consecutive points that have the lowest Standard Deviation and then use that value to calculate the control limits.  (+- 3 times Lowest SD)

This means your control limits always reflect the best recent performance you were able to achieve which in turn means that you don’t have to keep manually adjusting your Control Limits every month or so.

The following charts demonstrate Control Limits set with Low Variation and the second with using the Standard Deviation of the Population.

	Low Variation
	SD of Population

The Low Variation option provides you with Control limits away from the Specification limits thus giving you ample warning of process upsets as an Out of Control situation rather than Out of Specification or reject level.