OkiePC
Lifetime Supporting Member
I think using good winder control at the drive level is the way to go for you, but you should get a good handle on what is going on at the drive level and how it works, all the parameters involved, so you could in the future, use a generic drive and move more of the control to the PLC.
I have done winder control with a PLC for a line that had frequent stops and starts. The pure PID approach is not the best until your line speed is stable, and a PID is reactive, so I prefer to use line speed as the main source of the signal, and have some math to allow the calculated winder ratio to get the dancer under control when it is outside a deadband, and only let the PID have a small influence over the correction of the diameter calculation.
The dancer needs to have the most influence over drive speed when it is at the extremes, so it does not "top out" or "bottom out" after a stoppage and restart. At all times the inches per second of the winder is within known limits so just let the math make corrections to this magic ratio and multiple it by your line speed from the nip roller.
The main factor is the line speed, so when the line starts and stops, your dancer does not have to move to cause a PID to react and then cause motion, the winder will pick up where it left off with the known ratio.
The ratio is clamped between the known core size and max diameter, and allowed to "find" the actual diameter by adjusting the calculation to get the dancer error within a small deadband and letting the PID drive it slowly from there. Be sure to stop the PID when you stop the line too. This allows you to start a partial roll, for example, or use a larger core and still have good control until the math quickly catches up and the dancer moves right toward the target with very little overshoot.
You want the line speed as accurate as possible. In most really good slitters and winders, this is a quadrature pulse or a frequency connected from the nip or slitter drive to the winder drive and the drives have the smarts to make it a simple matter of entering parameters. This is the course you took, letting the drives do it, and the better performing machines I worked on were all done like that.
I have done winder control with a PLC for a line that had frequent stops and starts. The pure PID approach is not the best until your line speed is stable, and a PID is reactive, so I prefer to use line speed as the main source of the signal, and have some math to allow the calculated winder ratio to get the dancer under control when it is outside a deadband, and only let the PID have a small influence over the correction of the diameter calculation.
The dancer needs to have the most influence over drive speed when it is at the extremes, so it does not "top out" or "bottom out" after a stoppage and restart. At all times the inches per second of the winder is within known limits so just let the math make corrections to this magic ratio and multiple it by your line speed from the nip roller.
The main factor is the line speed, so when the line starts and stops, your dancer does not have to move to cause a PID to react and then cause motion, the winder will pick up where it left off with the known ratio.
The ratio is clamped between the known core size and max diameter, and allowed to "find" the actual diameter by adjusting the calculation to get the dancer error within a small deadband and letting the PID drive it slowly from there. Be sure to stop the PID when you stop the line too. This allows you to start a partial roll, for example, or use a larger core and still have good control until the math quickly catches up and the dancer moves right toward the target with very little overshoot.
You want the line speed as accurate as possible. In most really good slitters and winders, this is a quadrature pulse or a frequency connected from the nip or slitter drive to the winder drive and the drives have the smarts to make it a simple matter of entering parameters. This is the course you took, letting the drives do it, and the better performing machines I worked on were all done like that.