Feedforward v. Open-loop

Status
Not open for further replies.

petersonra

Senior Member
Location
Northern illinois
Occupation
engineer
I don't know exactly what an open loop controller would be. maybe like a thermostat? Not a commonly used term.

A feed forward controller uses some knowledge of the process to affect the loop output separately from (sometimes instead of) the error signal. It can be as simple as putting in what is called bias, where you know that the process will always need a minimum amount of output so instead of starting at zero, you start at whatever the minimum output is.

it can be a fairly complicated mathematical model where the output is mostly calculated based on the input variables, and then trimmed by the error (feedback) calculation.

Most PID loops (feedback control) only care about the error signal, and the output of the loop is adjusted by a mathematical formula based on the error.

A simple example. Take a steam heat exchanger. Normally you would use a feedback (PID) loop controller on it. The controller looks at the error (the difference between the set point and the actual temperature) and adjusts the output accordingly.

You could add feed forward to the system if you know how many BTUS/hour you need downstream at any one time and set the bias to open the valve to a point where that many BTUs/hour are furnished, and use the temperature feedback only to make minor trims to the output. Not a real common FF type application, but you might get the idea.
 
Last edited:

Jraef

Moderator, OTD
Staff member
Location
San Francisco Bay Area, CA, USA
Occupation
Electrical Engineer
Layman's version:
Feed Forward: You give a command signal, the output follows your signal without regard as to what is happening in the load. Example: you step on the gas pedal in your car, the engine gets more fuel. Done. What happens after that is of no concern to the simplistic feed forward control system.

Closed loop: Feed Forward, but with a sensor on the output that detects the RESPONSE to the feed forward command, then adjusts the command signal response to affect the proper OUTPUT. Example: Cruise Control in your car. You set the cruise speed, the closed loop comes from a speed sensor off of the speedometer. If the speed drops, i.e. you start going up hill, the sensor knows it, tells the closed loop controller that, and the controller automatically increases the throttle to give the engine more gas.

By the way, in the first scenario, THE DRIVER is actually acting as the Closed Loop controller. So for that first scenario to be correct, assume the driver is blind, deaf and has no sense of movement, as in being blind drunk. :eek:
 

Besoeker

Senior Member
Location
UK
Cruise Control in your car. You set the cruise speed, the closed loop comes from a speed sensor off of the speedometer. If the speed drops, i.e. you start going up hill, the sensor knows it, tells the closed loop controller that, and the controller automatically increases the throttle to give the engine more gas.
It's an apposite example of closed loop and one that I've used before but I'd have called that feedback rather than feed forward.
You have an input demand signal that asks for 70mph output. And you have a sensor that compares actual output and feeds that back to the input of the controller where it is compared with the input demand. Any error acts to correct the difference.

My lecturer for that subject had a nice succinct definition for a closed loop system.
"Error actuated and power amplifying."
That was four decades ago then some. But still valid.

We do a fair bit of work in the variable speed drives area. Some are retrofits using existing motors where closed loop control is required. Those can be a bit messy. The old speed feedback unit might be an analogue tachogenerator, subsequently out of production and no easy means of fitting an encoder. We have manufactured toothed steel wheels to fit and used an inductive proximity sensor to count pulses. We have a pretty good machine shop so we can custom manufacture them as needed.

We have a retrofit project going through at the moment where we need to detect speed. It's an existing speed controlled application with eight 200kW vertical shaft wound rotor motors. Speed feedback is currently computed from slip frequency and rotor voltage. That's fine. The system has been in operation for over twenty years. Recently we have been experiencing problems with dirty contacts causing the rotor contactor to drop out. The machine stalls with the stator still energised. The shaft mounted fan no longer provides cooling and the motor cooks.

A shaft mounted speed sensor would seem like an easy solution. But there's no way to mount that. Nor our toothed wheels. I think we can fit proximity sensors to detect the cooling fan blades. We are already undertaking the work. All the cabling is in place. We'll complete one to prove the system works. Access to do that has been an issue. It's a pumping station for London and production takes priority.
Their engineering wants it done. We want it done.
Frustrating doesn't have enough syllables...
 
Status
Not open for further replies.
Top