Load Cells

[GlennP 0]

Hi there, We are building a weigh station that is using two single point load cells(10kg) attached to a plate approx 150mm Wide by 1200mm Long. The information we have about mounting single point load cells indicates, finding the C of G of the plate and centering that over the centre of the load cell. My question is this, do we use the same thinking when using two load cells on the same plate? i.e. find the C of G and then divide that exactly between the two load cells. Thanks in advance, Glenn

[paulengr 44]

In theory, yes. However either way, you are going to have problems. Load cells are not really designed for lateral loads. To do what you are proposing, the standard arrangement is to use either 3 orr 4 load cells, unless you are using the "beam" type which support the mass and provide some lateral support as well. You can use 3 cells but then you have 2 cells at one end and 1 at the other, and the single cell will see twice as much force as the other two. If you don't do this, then you are going to end up with check rods somewhere in there to keep the beam squared up to the cells and this will result in accuracy issues. I really despise check rods...they seem to cause me more grief than any other load cell problem.

[GlennP 0]

Thanks for you reply, The only reason we are using two load cells is the plate is too large for the load cell to support. The weights are small from 30grams to 3kgs, but can range up to 1200mm in length. We have a transmitter that incorporates the two measurements from the load cells and sends out a single 4-20mA to my PLC We are going to set one up to test it. I suppose that if we can get repeatability then that's a success Cheers

[paulengr 44]

Load cells are just strain gauges configured in a Wheatstone bridge arrangement. They are typically driven by a 10 V excitation signal but load cells are very forgiving in this respect. You can sum the voltages together passively very easily but if you are doing custom equipment, I prefer to just buy a millivolt-input card for a PLC and read it directly. This is the same card that you use for reading RTD sensors. Just ADD readings to make it work, either electrically, or preferably, digitally. This arrangement is simple and cuts down on the amount of hardware involved in troubleshooting the system. You are not gaining anything in terms of calibration since the goal in the end is still to translate a digital reading into a "grams per volt" type unit with a simple regression curve (Y=MX+B). This requires a 2-point calibration at a minimum unless you want to do more points or you feel you can safely make assumptions about the "+B" factor (I don't). The nice thing about doing it this way is that if you have a load cell fail, you can both easily troubleshoot it since the PLC can detect the bad load cell (readings go out of range), and that you can also operate in a "degraded" accuracy mode if you so desire. Half the troubleshooting involved in load cell systems is just guessing which load cell to replace. It's easy with just 1 or maybe 2. It is much harder if for instance you have a truck scale with 8 sensors. This is speaking from experience. If you have excessively long wire leads (>50 feet / 15 meters), then the 6-wire arrangement is advised. You run 6 wires from the transmitter to the load cell. 4 of them are the standard function (2 excitation, two sense). The other pair are shorted at the load cell terminals to the sense wires. The transmitter/PLC card uses the second set of wires to estimate the impedance in the wiring. I avoid them if I can (more things to go wrong), but it definitely makes a difference where it becomes necessary. It can take the nonlinearity that you sometimes see in a 3-point calibration completely out.