Unwinders and winders are used in almost all web-handling industries including paper, plastics, converting, ferrous and non-ferrous metals, wire, printing, and textiles. The winder is often the limiting factor in machine throughput when continuous winders and unwinders are not used. Moreover, close winder control is essential to quality production.
This article deals with the most common type — dancer-controlled, speedregulated center winders — and examines the difficulties of winder control and the strategies used to wind and unwind various web materials.
Need for close control
In general, the problems associated with winder control include obvious physical deformities such as crushed cores, air gaps, starring (caused by tension fluctuations), telescoping, and necking, Figure 1. There are also less obvious winder-related problems, such as scratches produced by relative movement within the roll (cinching, Figure 1) and web breaks. These are frequently caused by compression forces in a wound roll, which can crush some materials causing fractures. Web breaks become apparent on the machine that uses the roll, not on the machine that produces the roll.
Types of winders
Winders predominately fall into two broad categories: center winders and surface winders. In addition, there are winders that combine the characteristics of these two types. An example of a combination winder is a speed-regulated surface winder with a torque-controlled centerwind used to wind slippery materials.
Control considerations for a center unwinder are similar to those for a centerdriven winder. Therefore, almost all of the winder information in this article is also applicable to unwinders.
In a center winder, a motor drives the core or shaft of the roll being wound. Although the motor drive can use any technology, ac drives and dc drives are most commonly used on large winders. On smaller winders, servo drives can be used.
Center winders are usually controlled either in a speed mode or in a torque mode. When controlled in a speed mode, a speed reference is provided to the drive by the winder control. In such speed-regulated systems, the speed regulator commands torque in the motor.
In torque-regulated systems, a speed regulator is not used. The motor will run at the highest speed it can reach with the available torque.
Winder control system challenges
Winders are often the most difficult part of a machine to control, because both the diameter and inertia of the wound roll are continually changing.
Roll diameter. Consider two different times during the winding of a single roll, Figure 2. Assume, at both times, the motor is turning at 100 rpm. While the roll has a 10-in. diameter, the machine is winding at 261.8 fpm. However, the 50-in. roll is winding at 1,309 fpm. If you increase motor speed by one rpm, the 10-in. roll winds 2.6 fpm faster, but material on the 50-in. roll is wound 13.09 fpm faster. Thus, the same speed-reference command change causes different web-speed changes depending on roll diameter. This difference— a system gain change—can complicate the control strategy in speed-controlled winder applications.
Roll inertia. A second complicating factor is roll inertia. For example, a 3.5- in. aluminum shaft weighing 60 lb has a moment of inertia of 2.5 lb-ft2. If 50 in. of paper is wound on this shaft and the roll weighs 2,500 lb, its inertia is over 21,000 lb-ft2. This is an inertia change of over 8,000 to 1. Such inertia changes affect the tuning of speed-controlled winders, the inertia compensation of torque-controlled winders, and the web-tension regulation capability of brake-controlled unwinds.
The effect of inertia on speed-controlled winders can be considerable. For example, an empty spindle on a winder, if the drive is tuned properly, will closely follow commanded speed changes. However, a large-diameter roll on the winder, tuned at empty core, will be unstable and will not closely follow commanded speed changes. In fact, the roll may be wildly unstable and may rotate in one direction, and then in the other, even if zero speed is commanded. This instability results because the optimal integral gain at the core is too fast for the high-inertia load to follow.
Continue on page 2
Drive tuning. Similarly, if the drive is tuned with a full roll, the empty spindle will be unstable, because the full-roll proportional gain is much higher than an empty spindle can tolerate. Therefore, most winder drives are not optimally tuned for all roll diameters and inertias. Tuning is then a compromise. On a dancer-controlled center winder, this compromise tuning can be seen by observing the dancer position during two different situations. First, when the machine is webbed and the line is enabled at zero speed, the dancer will usually go to its center position if the winder is on a core. However, with a large roll on the winder, often the dancer will not go to its center position.
A second illustration of compromised tuning of the control is when the machine accelerates. Compromised tuning is evident if the machine accelerates without excessive dancer movement with some roll sizes, but the dancer moves excessively at other roll sizes. This effect can also be caused by an incorrectly sized drive system.
Speed range. A third complicating factor is winder speed range. This does not affect the design of the winder control, but it does alter the selection of the drive system. A machine designed to operate from 100 fpm to 2,000 fpm, has a speed range of 20 to 1. This is a speed range that almost any drive can attain.
The winder, however, must accommodate this speed range from core to full roll. If the core is 3 in., and the maximum roll diameter is 51 in., the winder has a build ratio of 17 to 1. The winder must meet the machine speed range regardless of roll diameter. Therefore, the winder needs a speed range of 340 to 1. Many off-theshelf drives have an effective speed range of only 20 to 1 or, at best, 100 to 1.
An established, cost-effective method of providing winder control, dancer-controlled winders are often selected for applications with a build ratio of over 20 to 1. Because the dancer provides some material accumulation, slight feed-forward errors are easily accommodated and corrected. Thus, this type of winder is appropriate when very fast accelerations and decelerations are required, and for highspeed, nonstop dual-turret winder applications.
Usually, some small dancer movement occurs during steady-state, constantspeed operation. During acceleration and deceleration, the dancer moves more.
A dancer’s position is determined by the difference in speed between the winder and the preceding web section. If the winder is faster than the previous section, the dancer rises; if it is slower, the dancer falls. Thus, speed control is a good strategy for dancer-controlled systems.
In a simple dancer-controlled winder, Figure 3, the dancer position error is used by a PID (proportional, integral, derivative) control loop to establish winder drive speed. A PID loop maintains a process variable at a given setpoint by adjusting an output variable to minimize the error between the process variable and the setpoint. This type of simple control can be used on winders with build ratios of 4 to 1, or less. For example, if the core diameter is 3 in., then a maximum roll diameter of 12 in. can be controlled. The 4-to-1 build ratio for this type of control is just a guideline. Fast acceleration rates or large inertia changes may make a 4-to-1 build ratio unobtainable.
Similarly, if the material being wound has a low density, or if the roll is narrow, build ratios of 10 to 1 or even 12 to 1 may be obtained.
A more sophisticated dancer-control system, Figure 4, has three sensors that are used by the winder control to calculate the drive-speed reference. The three sensors determine line speed, winder speed, and dancer position. Two of the three sensors are used to calculate a feed-forward control term that anticipates the speed that the winder should be running. The feed-forward term is determined by dividing the line speed by the winder speed to calculate the roll diameter. The line speed is then divided by the roll diameter to provide the correct winder speed. If there are no errors in the feed-forward calculation, and if the drive follows the speed reference without delay, the dancer will always stay in its center position.
But, in the real world, nothing is so exact. Time delays, electrical drift, mechanical variances, drive sizing, and out-ofround rolls contribute to inaccuracies in the feed-forward calculation. To correct these inaccuracies, the dancer position error is fed to a PID loop. The output of the PID loop modifies the feed-forward value to provide the drive with the correct speed reference.
Systems such as those in Figure 4 can typically be used on winders with build ratios of up to 10 to 1. For example, with a 3-in. core, a maximum roll diameter of 30 in. could be obtained. The 10-to-1 ratio is a rule of thumb and larger or smaller build ratios can be obtained depending on several factors including material density and gear ratio.
Continue on page 3
Dancer-controlled winders with build ratio’s larger than 10 to 1 require a more sophisticated winder control. The block diagram for these winders is the same as shown in Figure 4. However, the winder control algorithm is more complicated. These winder controls have adaptive gain functions that compensate for changing roll diameter. Such adaptive functions also change the gain as the wound roll inertia changes.
Maintaining tension. In a typical pneumatic system used with a dancer controlled winder, Figure 5, a pneumatic cylinder is tied to the dancer. Both the top and bottom of this cylinder can be pneumatically loaded. A manually adjusted regulator, which provides air to the bottom section of the cylinder, is adjusted to counterbalance the dancer weight. The pneumatic loading of the top cylinder section, controlled by an electric-pneumatic transducer, provides the required web tension, which is based on a tension reference. This is usually provided by the control system.
Note that accumulators in the air lines to both sections of the cylinder enable the piston to move without compressing air in the system. If dancer movement should compress the air, thus changing the air pressure, web tension or the counterbalance pressure will be incorrect. A rule-of thumb is that the accumulators should provide at least ten times the volume of the air cylinder. This will normally keep web-tension variations due to air compression to less than 5%. If tension variations of less than 5% are required, use larger accumulators.
For large rolls, tension is usually linearly tapered as a function of roll diameter, Figure 6. It is not uncommon for winders to have nonlinear tension taper functions. The exact taper function required depends on the material. Some materials do not require winder taper tension. Unwinders typically do not have a taper tension function.
Dancer design. Correct dancer design is not a trivial exercise. Every time the dancer moves, it causes a small (or possibly a large) tension variation on the web. Although it is often not obvious, the winder control system does not directly control web tension. Rather, the winder control provides a tension reference, but the control itself regulates dancer position. Thus, tension variations are often caused by dancer design.
Theoretically (although not possible), the dancer mechanical and pneumatic system should be designed so dancer movement does not cause web-tension changes. However, steps can be taken to minimize the effect of the dancer on web tension.
First, use only low-stiction cylinders. The force required to start the cylinder moving shows up as a web-tension change. Cylinders that have some air leakage around the piston work best, and they are often the least expensive.
Second, make the dancer as lightweight as possible. The dancer has inertia, even though it may be counterbalanced, and the force required to overcome this inertia will cause a web-tension variation.
Third, use air pressure to counterbalance the dancer, not a weight. Weight adds inertia and this is not desirable, as explained previously.
Fourth, the web path should not change as the dancer moves. If it does, be sure the resulting tension variation is within acceptable limits. For example, with a pivoted dancer, Figure 7, web tension, which depends on dancer position, varies as the cosine of the cylinderdancer- arm angle. Because of its low cost, this is a common dancer design.
Dancer disadvantages. There are, however, disadvantages to dancer-controlled winders. Dancer design can affect web tension. In light-tension applications, dancer stiction and inertia can cause major problems. Also, good winder control is necessary to prevent instabilities, which can be encountered with large inertia changes and large speedrange requirements.
Finally, since dancer-controlled winders regulate dancer position and do not have tension-measuring devices, web-tension problems often must be solved mechanically, not electrically. However, the problems related to dancer movement can be minimized by good winder control.
One way to eliminate dancer-related problems is to use a load-cell to control the winder. This approach will be discussed in the final article in this series appearing in the next issue.
To obtain more information on winder controls by Amicon, Charlotte, N.C., please circle 421 on the reader service card.
If this article is helpful, please circle 422 on the reader service card.
Mark S. Dudzinski is chief executive officer of Amicon, Inc., American Industrial Controls, Charlotte, N. C.