A motion controller is the core of any motion control system. Motion controllers essentially take an input command, which can be a motion profile consisting of position commands, and issue torque commands to a drive which powers a motor and moves a load.

Choosing the right motion controller for your application begins with knowing the requirements for your specific application. Perhaps the most important is the complexity of the application. For instance, a fairly simple application is one that involves only a single axis of motion and where the speeds involved are relatively slow. On the other hand, a more complex application may involve multiple axes of motion with highly synchronized coordination between axes. Each of these systems requires different controller types as well as decision about general control architecture.

One of the most basic architectural decisions to be made in a multi-axis motion control design is the trade-off between centralized and distributed control. This is not a one-or-the-other choice, but rather a spectrum of possibilities. Obviously, the top-level executive functions that decide what is done must be centralized in a single processor, but systems can and do split out lower-level functions to distributed components (if at all) at various levels.

A typical hierarchy of functions would be:

1. Executive planning (sequencing)
2. Motion equation generation from move descriptions
3. Solution of motion equations (interpolation)
4. Position loop closure
5. Velocity loop closure
6. Motor phase commutation
7. Current loop closure
8. Power modulation

At any stage in the hierarchy, the information can be split out from a central controller to distributed devices. The further down the split is, the greater the communications bandwidth required and low latency required. The further up the split is, the less possibility there is for tight coordination between axes.

Digital networks for connecting different hardware modules are becoming increasingly common. The advantages of these networks are often obvious – just connecting all of these modules on a fiber or Ethernet string or ring can have huge advantages over traditional wiring harnesses. However, because these networks employ serial communications, there is an inherent delay in the communications through the data stream. When high-frequency feedback loops are closed across the network, this transport delay has the potential to detract from the stability of the loops.