The short answer is 'Yes, but ...'

Dancers are one method of closed-loop tension control, often preferred because it can compensate for variations and upsets that open-loop control methods—such as draw, speed, torque, and differential shaft control—cannot. A closed-loop system can counter variations that include bearing and gearbox friction changes while a machine warms up or that vary with age. The most common machine tension upset is merely changing speeds. Roller inertia causes a tension change that cannot be remedied with most open-loop systems.

However, dancers are not the only way to close the loop. More and more often, load cells are performing this duty. Load cells have several advantages, including the ability to indicate tension and the ability to resolve (though not necessarily control) smaller tension variations. It may be tempting to ask why you would ever want to use dancers when load cells are clearly superior in these important ways.

One reply has been that yes, dancers absorb tension variations. However, the damping benefit is not universal. With low-frequency variations, the dancer neither absorbs nor exaggerates upstream tension disturbances. At very high frequencies the dancer absorbs tension variations, especially if the roller inertia is high.

Unfortunately, however, at intermediate frequencies the dancer actually EXAGGERATES the tension variations. Simply stated, at or near the mechanical resonant speed of the dancer system, the downstream tension upsets are worse than the upstream. This contrary behavior gets worse with higher roller inertia. Thus, we can extend the range of the beneficial high-frequency response of the dancer by increasing the mass of the arm and roller, but we do so by making the detrimental intermediate frequency resonance worse.

Leaving theory aside, let's talk about application. There are situations where the tension variation absorption benefits of a dancer are clearly desirable. There are cases where the frequency of the upset is high. Examples may include flying-splice unwinds and turnups on winder turrets. In both of these applications the shock of the changeover from one roll to the next is one with a lot of high-frequency components. Another good application would be rotary diecutters. Another might be some cyclic advance-and-stop type machines such as platen presses fed by an unwind.

Contrary to common sense, however, you do not want friction in the dancer system. Not only does this not substantially help damping, it spoils the resolution of the dancer. A good dancer design would have a breakaway friction much less than 10 percent of the minimum tension to be run on that machine. In my book, The Mechanics of Rollers, you will learn how to easily calculate maximum acceptable friction and how to measure it.

However, for most garden-variety machines the dancer would need to be movable with your little finger. For thin webs and/or narrow machines, the dancer should be so friction-free that you could move it with a BROKEN little finger. Many if not most dancers fail this test. They almost always fail for the same reason: too much cylinder seal friction. Rolling diaphragm or glass cylinders are often required to reduce friction. Excessive friction of any type on the dancer will hobble the drive so that fine tension control is simply not possible.

The primary reason we use dancers, however, is not to absorb friction. It is to allow tolerance to a less than stellar drive. If you bought your drive when it was on year-end closeout from your local hardware store, you will not be able to use load-cell control because the drive is not going to be responsive enough. Moreover, if your drive guy is the local "Anything Electrical or Electronic" programmer, you will not be able to use load-cell control because he will not know enough about what he is doing to make even a great drive work.