Control Systems - Feedback

sample imageA simple definition of 'Feedback' is simply just telling a system how you want it to behave.
A cart rolling do a hill will mostly likely get there.  But was it going too fast?  Was it going straight?  Did it stop where you wanted it to?  Pushing that cart down the hill without a feedback mechanism would be a demonstration of an 'Open Loop' system.  Now, add a governor to regulate speed , opto sensors to see where it is going, and tactile sensors to feel the wall to stop, and you have a 'Closed Loop' system.

An open loop control system is one which does not use sensors and relies on calibration to achieve the desired action. Let's say you want to use a pair on electric motors on your cart.  The speed curve for the motor states that 12V will give you 100RPM.  So, you wire up the motors, and you slap in the battery, only to find the cart is going in circles.  Why???  Mostly like one of the motors is slightly out of spec. and is turning at a different rate than the other one.  Now, take that same setup and add wheel encoders which are wired to a controller and you have a closed loop feedback system.  As the wheels move away from their set-point, the encoders feed the speed information back to the controller which adjusts the voltage (or PWM signal) to compensate.  

Advantages of closed loop systems over open loop are that it improves accuracy and gives the system the ability to automatically make continuous adjustment. Without feedback may systems would be uncontrollable.  Many actuators have built-in characteristics which makes them erratic and/or non-linear. These mechanisms could overshoot their desired target positions, move at an unintended rate, or not move at all.  Feedback will make the uncontrollable, controllable.  Feedback can also be used to change the characteristics of a system, making a linear response to a non-linear input, or non-linear to linear.

Feedforward Control, is essentially dealing with a disturbance before it enters and affects the system.  This is a very accurate way of dealing with systems error.  However, the feedback method assumes the disturbance was known in advanced and mechanisms/models were put into place to handle it.  The problem, here you must know and design-in the feedback system for all anticipated problems.  The configuration assumes that that the disturbances are known in advance, and that they will have sensors associated with them.  It also assumes the disturbances modeled will be the only disturbances.  Due to this added overhead, feedforward control tends to be more complicated and expensive, and requires it requires a better understanding of the system being monitored.

Remediates the disturbance before the process is affected
No affects to the stability of the process (system)

No ability to eliminate steady-state offset
Requires a sensor and model/design for each disturbance

Feedback Control, is looking at the result to see how it has changed, and then changing the input variable to compensate.  This method is tried and true.  If the controller detects a difference or error between the output value and the desired value of the process, the controller will then take the necessary corrective action to return the process to the desired value.

Very simple, and easy to build
All disturbances corrected
Offsets are easy to eliminate

Waits until output deviates from set-point before correcting
Affects closed loop stability