The most relevant measure of tension-control health is if very little waste or delay is attributed to tension variations with respect to time. Examples of waste which may be increased by tension variations would include going out-of-register during printing, as well as some types of web breaks on paper.

Unfortunately, while economics of waste and delay are the ultimate measure, the connection to tension quality could be difficult to establish in a specific situation. Thus, we can look only at textbook measures of health. The danger of fallacy here is that some products and processes, such as the winding of carpeting or paperboard, are tolerant of even extreme tension variations. Other products, such as thin materials, or other processes, such as multi-color printing, can be very sensitive to tension variations.

There are several indicators of tension health. The most obvious would be instruments such as load cells, dancers and drive ammeters. However, the product will also react to tension in ways that can serve as clues. An indication of variability by any measure, whether by instrumentation or eye, is enough to cast suspicion on a drive system.

Drive systems are composed of electrical items such as motors (or other actuators such as clutches, brakes etc.), motor controllers and control algorithms. Drive systems must also consider mechanical issues such as parasitic drag (nips, gearboxes and bearings) and roll/roller inertias. The mechanics of load cells, dancers and accumulators also plays a part in the drive system.

The load cell is the most useful tool for tension-control quality. Tension variations as read by fast-acting load cells should be held to a particular value. One common specification of control quality would be variations not to exceed five percent of setpoint during steady-state and 10 percent during speed changes (and perhaps a bit more with violent transients such as wind/unwind roll changes or engagement of nips).

The problem here is that the load cell may not be reporting fast enough. If the cell is read by a PLC or computer, chances are good that the display is too sluggish because of averaging and sampling. You may need to put a fast recorder directly on the output of the load cell amplifier. Alternatively, you could resort to the old-fashioned but responsive analog meter with needle.

Dancers give us different information from load cells. If the dancer moves we know that a tension variation exists. However, we have no idea how big the variation is. If, on the other hand, the dancer is steady, we can not assume that tension is steady. All we can say is that the tension variations did not exceed the friction of the dancer. Unfortunately, many dancers are not so free-moving (usually due to cylinder seal friction), so that variations can be large before the dancer breaks loose.

Alternative instruments include DC drive amperage, again with responsive analog meters. Unfortunately, most modern AC-vector drives do not give us such a convenient read of motor effort. What we look for is amperage that is a good fraction of rated load and steady during steady-state operation. During speed changes, we expect the needles of all drive points to move in concert without overshoot or being sluggish.

The web gives us another view of tension quality. Obviously, any web bounce is a bad sign. However, bounce could be caused by flutter or bagginess as well as tension. Sideways movement of the web is a similarly bad sign. The same goes for width variations or loss of registration tolerance. A web break could indicate tension variations, or it could be caused by other means.

The most demanding time for a drive system is during speed changes. If your registrations are lost during speed changes, tension variation is a likely candidate. If a speed change causes any ring or mark on the side of a wound roll, tension variation is a likely candidate. If you are not satisfied, you will need a drive expert who specializes in web handling to sort things out.