kamenges
Member
Again, give it a try. It would be relatively easy to do. You will need to make a force profile since the friction most likely changes as you move along the rails.
However, my guess is that you won't see much of a gain. By the time you detect the need for the offset the tension disturbance already occurred. Let's take the example where tension is a little below setpoint. Your control system will be increasing the torque command to try and get more tension into the web. But none of that torque is going into the web; it is going into breakaway force for the dancer system. Keep in mind that tension is the result of strain. You need relative displacement of sometning in the web path to create a change in tension. Once that breakaway force is exceeded the dancer will transfer the increased force into the web by moving slightly. At that point the tension will make a step up, since there is more force available than is required for tension. This is the first point that you can detect the need for anything. By the time you can react the need for the reaction is probably gone.
Now, if you are using this dancer as an accumulator also and the dancer makes extended unidirectional moves, then the friction compensation you are talking about is much more valuable. In that case a good friction map and speed and position knowledge can go a long way toward a good disturbance rejection scheme. But during 'stable' operation as a dancer, not so much.
Keith
However, my guess is that you won't see much of a gain. By the time you detect the need for the offset the tension disturbance already occurred. Let's take the example where tension is a little below setpoint. Your control system will be increasing the torque command to try and get more tension into the web. But none of that torque is going into the web; it is going into breakaway force for the dancer system. Keep in mind that tension is the result of strain. You need relative displacement of sometning in the web path to create a change in tension. Once that breakaway force is exceeded the dancer will transfer the increased force into the web by moving slightly. At that point the tension will make a step up, since there is more force available than is required for tension. This is the first point that you can detect the need for anything. By the time you can react the need for the reaction is probably gone.
Now, if you are using this dancer as an accumulator also and the dancer makes extended unidirectional moves, then the friction compensation you are talking about is much more valuable. In that case a good friction map and speed and position knowledge can go a long way toward a good disturbance rejection scheme. But during 'stable' operation as a dancer, not so much.
Keith
