There are three basic disciplines of control.  Discrete control which generally relates to making a product or dealing with sequential and event driven logic, process control which deals with the conversion of raw materials into more complex bulk products, and real time control of things like electric motors.  In general, discrete control is not really time based, although there are exceptions. Process control is based on longer time periods due to the nature of the large batches of material that are being processed and the associated thermodynamics.  The hardest of all real time control in the case of electric motors which requires nanosecond capability from the embedded control system to achieve the performance needed by energy conserving systems.  As a by product of the different time bases, each technology has grown into it’s own discipline and control philosophy.

Occasionally the line between mechatronics as the design of mechanisms in discrete manufacturing and applications that are more process oriented blur the neat categories of the major control disciplines. More and more control system requirements involve the blending of 2 or 3 different types of control into a single architecture.  This creates subtle problems in order to properly architect the system so that the final effects are achieved.

Polishing and grinding, for example, appear to be positioning applications.  A grinding wheel or buffing wheel must be brought into position to make contact with a workpiece.  So the normal control system behaviors must be dealt with in order to achieve position.  But positioning the tool is only the beginning of the process.

How do we measure the process of grinding or polishing?

And most importantly, how do we know when it is done?

The process of grinding or polishing is a matter of torque in the application of the working tool to the workpiece it is in contact with.  Generally through an electric motor that is turning the tool.  By measuring the torque, which is current in the motor, we can know that the actual process is being achieved.  It may require empirical measurement to determine how much torque is required to achieve the proper surface finish, but there is a direct correlation.  Too much current means the tool is buried in the part, too little current and there is no work being done.

But at this point, there is a process that can be controlled.  If the proper torque level is applied through the motor the runs the tool, there is also a corresponding value as the contact is reduced that indicates the completion of the process.

This behavior is completely separate from the position of the tool.  However, if there is reduced contact with the workpiece due to the tool wearing out, that is, the size of the tool has decreased slightly, then the positioning system has to be updated to compensate.

These are simple concepts, but they are often overlooked.  Ironically, there are many applications that require close consideration of the mixed control methods.  Chemical mechanical planarization of silicon wafers suffers from similar difficulties with the need for extraordinary precision in polishing the surface of the wafer.  Do we really know when the process is done or do we just leave it running an extra 20 minutes just in case?

There’s always room for improvement.  And some of the recent control system innovations are delivering significant performance that should be considered as we pursue new applications.