Load Cell Insight

Can someone give me some detailed understanding on load cells in web tension control please.

This is a thick film application and the winder speeds are controlled by a S warp stand that has a tension bar that has 2 1000 Lb rated load cells which are half bridge and from the best I can tell are series wired to make a full bridge.

We also have some others that are full bridge load cells that connect to the tension controller parallel.

I assume that since it's 2 1000 LB load cells the rage is 0-2000 LBS correct?

We normally only run 700 LBS of tension but we are blowing a lot of the half bridge type load cells on the same line and have changed them multiple times as well as replaced the wiring.

So I am trying to determine what the problem might be but at the same time gain some insight as to how these setups are working and why you would choose full bridge over half bridge and why some are series and some are parallel wired?

Any Insight you could pass my way would be greatly appreciated.

Some reading to help:
https://www.transducertechniques.com/wheatstone-bridge.aspx

http://www.allaboutcircuits.com/textbook/direct-current/chpt-9/strain-gauges/

I doubt the half bridges are wired in series but it is possible.

Tim,

The normal load is not the cause of the problem. Typically load cells fail from shock / sudden impact on the load cell. What many would consider a minor impact on the load cell can be 6-10 times the amount the load cell is rated.

In your case, the assembly is constantly moving trying to allow for flexibility in the web (thick film in your case) to prevent breakage. The sudden release of the web and then stopping is the shock to the load cell.

We had a burnish machine that wore in seat track assemblies that drove us nuts until we had a technical discussion with the load cell manufacturer. we multiplied the load cell rating by 5 and the problem went away. Since then, we have always had to take into account the load cell shock factor.

regards,
james

load cells convert tension and compression into an electrical signal via a bridge circuit. they are usually rated as mV/V i.e for a given input voltage there will be a linear mV output. some will have built in shunts for calibration, but others will come with an external shunt resistor. when calibrated the load cell will provide feedback to the servo system which in turn regulates the web tension via speed or torque control.

You need to determine the failure mode.

Are they mechanically breaking?
Is the zero point shifting?
Is the strain gauge going open circuit?

Load cells are pretty robust devices and will fail by either going non-linear (sprung load cell), This is where a force greater than then maximum load has been exerted and physically damages or deforms the load cell. the other failure I have seen is zero point shifting. this is an indication that the load cell calibration has changed. zero point shift can usually be calibrated out and may or may not require the replacement of the load cell. I have never seen an open circuit failure but this is not to say that it can't occur.

JY

have a look at the voltage that is used for excitation and the amps on the same line.
check with ohm meter the faulty ones.

load cells are voltage devices. amps has little to do with it other than ohms law. but to your point (shooter), checking with an ohm meter will help diagnose the bridge. in the end if the offset is large I would just replace the load cell.

Originally posted by jeffyech:

load cells are voltage devices.

Click to expand...

This isn't exactly true. A loadcell is a resistance device. How you choose to determine that resistance is completely up to the designer. While the vast majority of load cell amplifiers used for tension measurement are based on the voltage change at the load cell one could certainly use current feedback just as effectively. There are analyses that show this is a more accurate method, although somewhat more complicated.

Keith

I'm sure one could use a load cell as a resistive, current or voltage device. I've used thousands and only a voltage feedback device in the form of mV/V.

With a problem like this (and I have had my share), you need to start by going through the entire system. As mentioned above, ALWAYS check for a mechanical issue first in any weighing system, but since you mentioned that you are going through load cells like crazy, there are a few things you can do.

Since you are getting feedback to a servo controller, we really need to know if this system is using a trimming(summing) board or a weight indicator. Chances are, since you have two load cells you will be landing the cells on a board. This board takes the signal from the load cells and combines the two for a more balanced reading across your scale.

I won't give advice based on assumptions, so update the thread please for more specifics.

There are a few things to check and think about:

Has any welding went on in the area around the scale. This could be any kind of rail/mount system or on anything that is tied to a local ground. Go around checking all the grounds in the area for potential difference to each other. This issue is pretty rare, but I have ran across it once.

Double check each and every wire in the weighment circuit (you should check excite wires particularly) for proper polarization. A small strand of wire shorting could cause voltage spikes in the load cell but not be enough to trip a circuit protection device. Typically load cells are rated for about 50V tops. Any more and you can short out the bridge.

Is there moisture in the area that could be making its way up the load cells?

Are your shortening the load cell wire? This CAN be done. I have done it and seen it work, it calibrates out, but the wires are generally sealed pretty well, and honestly it is better off finding a cable organization scheme to avoid this. I have never shortened wires without being prepared to change a load cell shortly after doing it.

All the above things are pretty tedious, but basically you are looking for 3 things:

Mechanical damage to load cell (over-torquing bolts can cause this)
Moisture getting inside load cell
Inadvertent electrical damage from bad ground, poorly landed wires, defective equipment or otherwise.

Good luck!

As for "series or parallel wired load cells" I am not exactly sure what you mean. All scales I have seen with multiple load cells use a PCB to combine the signals and adjust out differences in the load cells.

Another question I wanted to ask is: Why do you think you are loosing load cells? What are the symptoms?

You generally would use a half bridge to save time/cost vs accuracy.

If you are looking for a low cost application when accuracy to the nano volt is not required you will find (as long as the gauges are placed on the load cell in the correct locations) that the output will not be too dissimilar between a full and half bridge. When this does present a problem is when load is applied unevenly.

If the gauges are in a shear web configuration as you suggest, and the application prevents uneven loading then a half bridge may well be used.

Main causes of offsets/incorrect outputs:

1) overloading. If built well a load cell should take 150-200% of SWL before it offsets. Overloading causes an offset to the output with no load applied but can normally be corrected by a shunt resistor.

2) shock loading. Already been covered but can cause an offset corrected by a shunt resistor (depending on how high the shock was).

3) Uneven loading. Causes the material position to change slightly effecting the output. But more importantly, strain gauges are always positioned so when a load cell is loaded correctly, the gauges see the maximum strain. Therefore, if a loadcell is loaded unevenly it sees less strain, so the perceived output is lower, and people generally apply higher loads than they think they are applying.

Another common cause of failure which many over look:

mechanical shock, if the load cell is hit with a hammer then this shocks the adhesive between the gauge and the material, not the material itself. This often gets overlooked can cause non-repeatable outputs.

As someone has already said, if you are expecting to see shock loads, you must uprate the load cell to one capable of dealing with the shock loads you are seeing.