There are even more reasons to calibrate dancers than load cells. Just like load cells, dancers are a tension indicator, and the native units are not proper tension units. In the case of load cells, the native units might be 4-20 mA or 0-5 V, which have to be converted to lb/in. In the case of dancers, the native units are cylinder-pressure units, which also must be converted to proper tension units of web force per unit of web width.

The zero for a dancer is defined just as with a load cell. It is the amount of pressure on the control side that would be required to counter balance the weight of the assembly. Whether a counterbalance is actually used is immaterial in this discussion. (As seen in the figure, a good design might use a counterbalance pressure on the piston side of the cylinder to mechanically zero the system. The control pressure applied to the rod side then would be used purely for tensioning the web.)

The gain for a dancer is calibrated the same way as load cells were described in last month's column. We merely hang weights corresponding to roughly the full-scale design tension of the machine. It may be good to use a mid-weight check as well. What the instrument technician has done so far is to create a table with three pairs of numbers. The X column is the load applied, and the Y column is the pressure required to balance that load. A line is fitted and a conversion formula is obtained.

There are two ways to implement this calibration. At the very least you can apply a stick-on overlay to a standard analog pressure gauge where the new force-per-width scale accompanies the pressure scale. Or, you could make marks on the glass with an indelible pen for the new scale in proper tension units. With PLCs, you can take the P/E transducer output into the PLC and compute a proper tension.

However, there is one distinct difference between load cells and dancers. Load cells have no significant hysterisis while dancers usually have debilitating lag in response caused by mechanical friction. The piston seal alone is enough to cripple the sensitivity of most dancers. Friction or uncertainty in a dancer can be measured various ways. 1) Find the amount of additional force required to move a balanced dancer. 2) “Weigh” the friction as the difference between raise and lower pressures. 3) As seen by the calibration procedure itself, resetting the weights would give slightly or significantly different answers. How this friction is best obtained is a matter of what is easiest to set up. However, how much friction is allowable is even easier to state. Friction should be far less than 5 percent of the lightest tension to be run.

Most dancers fail this, hobbling fine tension control. The drive is then in bang-bang control much like a furnace going on and off. With dancers, the tension wanders in the dead band of the sensor. If friction is excessive, it is usually necessary to replace the cylinder. The simplest replacement would be a bucket with weights, or a rolling diaphragm or seal-less cylinder might be used.

920/725-7671, drroisum@aol.com, www.roisum.com