Are there issues with using load cells near the ends of their range? 4

[davidd314152wind4]

Feel free to correct my vocabulary here (not sure if spectrum is the right term).

Are there issues with using a load cell in the lower part of its measurement spectrum? For instance, a load cell rated for 0 N to 2500 N, is it less accurate or precise in the 0 N to10 N range than it is in the 1000 N to 1010 N range?

I understand it would not have the same resolution as a 10 N load cell itself; I am just wondering if there is reason to avoid using the load cell at the extremities like this.

Also, are load cells linear? I was thinking I could determine the resolution of a load cell by dividing its range by its bit width, would this be correct?

For instance, a 0 N to 1000 N load 16 bit load cell would have a resolution of 1000 / 2^15 ... ?

 

[sreid]

There is no reason to believe that a bonded resistance load cell will have any special problems near zero. The linearity should be as good as the transducer spec. Watch out for drift from many sources.

Your problem will be noise. You will likely need to take many samples and average them to see load changes in the one Newton Range.

 

[Compositepro]

Accuracy and resolution of your measurement are specified as a fraction of the full scale range or as a constant absolute value (e.g., "x" grams). If you only use the low end of a load cell's range your measurement will have significanly lower accuracy and resolution than if you used a smaller load cell at close to full scale. This applies to almost all instruments or sensors.

 

[MikeHalloran]

I would use a somewhat smaller load cell at _half_ of its full scale range if possible.

Mike Halloran
Pembroke Pines, FL, USA

 

[felixc]

I have worked in a previous life at a company that makes load cells. The transfer function of a load cell has the shape of an "S". The linearity versus the real curve is given as a percentage of the total cell range, as pointed by compositepro. You will also find some hysteresis because of the settling on the rest of the equipment used to fix the load cell to the rest of the setup. So repeatability problems, in addition to absolute value reading errors. In a given small range, the slope may be quite different from what the general sentivity will give. Temperature gradients through the cell will also create output peaks. Also, the sensitivity figure of a load cell is determined by an interpolation of not many measurements. I remember some of the cheapos taking only 3 measurements. The more measurements, the more expensive the load cell, as taking these measurements require time. Using only one percent of the range of a load cell is not a good thing at all. Everything will challenge your measurements. So if you can, try to match the load cell capacity to your measurement range.