Process Monitors and Controls

     

Process Variability and Capability

         

Every process varies.  If you try to bake ten cookies using the same mold, mix, and procedure, none of them will come out of the oven identical. To a certain extent they will all look similar, but no two of them will be exactly the same. This inherent tendency of your cookies not to look identical is known as your process variability, while the ability of your baking prowess to make your cookies stay within predictable limits of similarity is your process capability.

      

 Books for you: 

Using the same process to bake cookies over and over again will result in many batches of cookies that look similar, but once in a while you'll get a batch that's quite different. One time you may get a batch of darker cookies, and another time you may get a batch of sweeter cookies. When this happens, the cookies are outside the normal variability of the rest of your cookies, so they are called 'outliers.'

              

Something special outside of your defined process happened that resulted in such outliers, e.g., you probably set your oven too high or you inadvertently added more sugar. Your process is said to have gone out-of-control because of special or assignable causes. Special causes are unpredictable and should be avoided. In processes that are more complex than baking, the occurrence of out-of-control incidents can be very costly. 

       

In semiconductor manufacturing for instance, a single misprocessing incident due to an assignable cause can mean losses in the hundreds of thousands of dollars. This is why in this industry, it is everybody's responsibility to keep the process in control.

              

This can only be done if there's a means of observing the process for any abnormalities and to correct the situation before the process goes out of control. Observing how the process takes place and behaves is known as process monitoring, while responding to changes to keep the process within its normal variability is known as process control.

      

Process Monitoring

       

Thanks to the invention of statistics, nobody has to stare at a process all the time to know how it is behaving.  It is just necessary to look at how a process looks like at specific intervals and a fair assessment of how the process generally behaves may be achieved. In fact, it is not even necessary to scrutinize all aspects of the process at these intervals.  One simply needs to check the process aspects that matter most.

      

Process monitoring consists of observing and measuring the critical parameters affecting a process at pre-defined, regular intervals and recording the observations and results. The critical parameters monitored are chosen in such a way that they collectively represent the state of the entire process.  Since monitoring costs money, it is necessary to limit the number of parameters monitored to the minimum required to ensure that the process is meeting the company's quality standards.

      

One way of reducing the parameters that need to be monitored is to look for correlations between parameters.  If the behavior of one parameter can be reliably predicted from the behavior of another parameter, then only one parameter has to be monitored.  The process owner should eliminate all redundancies in process monitoring.

      

Monitor results may just be jotted down on record books, but the simplicity of this method has a big drawback.  Numbers written on paper are difficult to analyze visually and will not catch the attention of the process owner when an anomaly arises.  This is why many companies employ control charts instead to record measurements from their monitors.

       

A control chart plots the measurement data for the parameter being monitored, usually with time on the x-axis and the measurements on the y-axis. These charts are special in the sense that they show the historical behavior of the parameter in terms of the mean and variance of past data. The variance is expressed in terms of upper and lower control limits, which are three (3) standard deviations away from the mean of the data distribution. 

                              

During process monitoring, the process owners looks out for anomalous trends in the control charts.  For instance, any measurement outside the control limits is an automatic cause for alarm, because this is an outlier.  Four (4) or more consecutively increasing or decreasing points form a trend that is not normal, and therefore deserves attention.  Six (6) consecutive points on one side of the mean also deserve investigation. When such abnormalities are observed, the process owner must take an action to bring the process back to its normal behavior.

                  

Process Control

        

Process Control, which is the means by which a process is kept stable within its normal behavior, comes in many forms.  It can be as simple as assigning a person to monitor the progress of the process (e.g., watching the cookies as they bake inside the oven) and responding in accordance with the state of the process observed (turning off the oven when the cookies are brown enough).  Or it can be as complicated as computer-based real-time monitoring of many parameters at the same time and automated control of equipment based on what the parameter readings are.

      

Again, cost and reliability play an important part in determining what process controls to implement on your manufacturing line. Remember, sophisticated computer systems need expensive maintenance, and a single downtime can result in total breakdown of your process control system. All things considered, many companies today (including the ones from the semiconductor industry) prefer to use Statistical Process Control (SPC) to keep their processes in check.

      

 

See Also:  SPC Quality Systems Document Control The ISO9000 Standard;

Metrology and Calibration

 

   

Books recommended for you:

 

 

HOME

         

Copyright 2001-2007 SiliconFarEast.com. All Rights Reserved.