What kind of controller would you call this?

[DragonPetter]

I have a motor that is driven rail to rail. So, in forward its +12 volt, and backwards its -12V, and it can also be at 0V.

Can I call it a bang-bang controller, or is there some other name for it since it has 3 different control values?

It is also very stable and accurate steady state, which I'm trying to understand since its just proportional with very large gain. My only guess is that the thing its coupled to has very high friction.

 

[Floid]

I suppose you can call it whatever you like. I might call it a state based controller since your output is based only on deciding which of three output states to use.

But you are right, the only way it can be stable and have an accurate steady state is for one of two situations to occur (and probably both).

1.) Your motor is way to weak so that you have such a slow response time that it is difficult to see the oscillation in the system.

2.) Your method of measuring the system error is inaccurate to the point where you only see large errors so that whenever you are in the ballpark of being right your controller switches to the 0V state. This is why there would seem to be little steady state error.

 

[berkeman]

Maybe just "Bidirectional Controller"?

 

[DragonPetter]

I suppose you can call it whatever you like. I might call it a state based controller since your output is based only on deciding which of three output states to use.

But you are right, the only way it can be stable and have an accurate steady state is for one of two situations to occur (and probably both).

1.) Your motor is way to weak so that you have such a slow response time that it is difficult to see the oscillation in the system.

2.) Your method of measuring the system error is inaccurate to the point where you only see large errors so that whenever you are in the ballpark of being right your controller switches to the 0V state. This is why there would seem to be little steady state error.

Yes, its a BLDC motor with only 3 hall sensors that gives me 30 degree increments of position, so I think its safe to say that the resolution is so low that the controller stops providing the command and the friction stops the motor before it ever reaches the next 30 degree position. I was reading about bang-bang controllers, which can include a hysteresis band, its on page 6
http://faculty.washington.edu/lum/aa448/lecture_notes/lecture13.pdf

Would it be safe to say that these positions between the 30 degree position increments are acting as deadbands to reduce the "chatter"?

This would make sense, because when I put a delay on my control loop(it wouldn't apply the error until a delay after sampling the feedback), I noticed some chatter.

Thanks for the replies.

 

[Floid]

Sure, from that description you have a bang/bang controller. I can't remember ever hearing that term before.

Yes, what it sounds like is happening is that when you enter the correct 30 degree increment you are transitioning to the 0V output state and the system stops in this band before it is able to overshoot in the next band. In the system you have your controller is a good one as there is nothing gained by a more complex control algorithm.

When you add the extra delay then your system is overshooting into the next 30 degree band which reverses the controller direction and results in oscillation.