Calibration is required of all sensors simply because none can be trusted. Consider the ubiquitous scale for weighing produce or the flow meter on gasoline pumps. Regular calibration with documentation is required by law even though these sensors are simple and seemingly bulletproof. One of the tenets of ISO is to calibrate important gauges. I strongly disagree that an “indicator” can be deemed not important enough for this effort. All gauges should be important and should be trustworthy, or they should be removed. Trustworthiness is established by regular calibration.

Load cells should be first calibrated at the design stage by calculation. The designer should understand enough about the mechanical and control details that the zero and gain can be set close before the equipment is even shipped. Upon installation, the zero and gain should be calibrated again, but this time physically. The zero is easy; turn the knob until the reading is zero. If there is more than one cell, each must be zeroed independently.

Gain is a little more involved. The procedure: hang weights upon a wire (small machines) or strap that follows the web run in that local area as seen in the figure (above right). The amount of weight should be similar to the full-scale design capacity of the machine. If the machine is rated at a maximum of 2 PLI (lb/in.) on a 40-in. web, you would need something like 80# available. Note that the proper English units are force/unit width or lb/in. This complicates the procedure because one must specify the current width for the correct units to be computed by a PLC. Stand-alone load cells do not usually have the ability to accommodate regular changes.

Technique is very important. The setup must be as friction-free as possible. This means that you run the wire over as few rollers as possible, and all must be free-turning. If the roller adjacent to the load cell is motor-driven, you must use a pulley to simulate the geometry without touching that high-friction roller. The wire or strap should be as thin as possible to avoid bending hysterisis. It's easy to check your setup. Reset the weights, and the reading should be the same. If not, it's almost always a setup problem having to do with friction. Similarly, the reading must return to zero when the weight is removed. Some people check a mid-span load. However, linearity is seldom a problem with modern load cells so efforts would be better spent elsewhere. Finally, it's important to have the wire in the middle of the rollers so that each cell shares the load and calibration equally.

Why recalibrate? Three reasons. The reading does not return to zero with the web off; if there was an accident in the area which might have overloaded the cells; and for precision processes, recalibrate at least once a quarter. If maintenance finds that the cell needs frequent calibration, you have a design error. Re-zeroing the load cell would be about as lasting as fixing a flat tire by putting air back in. The two common load-cell application errors are where electronics drift and with cells that do not have mechanical overload protection and are thus easily damaged.

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