Load Cell question

russrmartin

Member
Join Date
Aug 2002
Location
Eastman, Wisconsin
Posts
744
I have not had a lot of experience with load cell installations. I have a certain load cell which when installed, will be very difficult to calibrate. We have calibrated the cell in a press using all the existing hardware that will be used in it's working condition from the amplifier and PLC end. Using the press we have for this, the load cell was dead nuts, and very repeatable. My question is this. When installed, the load cell will see the load differently than it did in the press we used to calibrate it. However, this load will be applied exactly the same every time. I expect that when all is installed, due to existing hardware being attached to the load cell, the zero position will need adjustment. Can I simply create a zero offset in the PLC to "tare" the load cell to zero and expect that it will be linear and correct throughout the whole range? Or will the load cell react differently because the load is applied to it differently. If it should be off by the same amount, I'd rather not have to adjust zero and gains again, due to not having anything really reliable to use as a force reference.

Example. Now, 0-1000 lbs. is right on. When mounted, zero is off by say 75 lbs. Should I expect that when 1000 more lbs is applied that the cell will show me 1075? Or is it a likely possibility that because the load cell is recieving the load differently than it did in the calibration press that the whole scale may be off. Something like 0=75, 1000=875. The whole point of this would be that if this is linear, our calibration procedure would be to remove the load cell, calibrate in press, and reinstall. Then create what we consider zero lbs. state, and adjust offset to get a zero value.

Edit: FYI, this is actually for a nip roll. Right now, we have no way to apply a calibrated force to the sensing roll because of it's mounting setup. The load cell is mounted horizontally to a frame with the butt of a cylinder attached to it. The cylinder pushes on a rod, roughly in the center causing it to rotate around a pin. The end of the rod opposite the pin hosts the nip roll. There is one of these on each end of the nip roll. Due to other rolls and equipment for the process being in the way, there is no easy way to create a horizontal force by hanging weights or the like.
 
Last edited:
As long as you are actuating the transducer in a straight line, you can expect linear response. The difference between horizontal/vertical is too minor to worry about.
The range is still 0-1000 regardless of Tare. If you have 75 pounds of load, only the rest of the range is available. If you Tare at 75 pounds, your new, linear range is 0-925. (75 is subtracted from raw reading).
 
Thanks, but....

Thanks for the responses. I think the answers were a little basic for what I needed, but the feeling I get is that yes, these will be linear. I was already aware of the 0-925 effect, but am really concerned with the congruency of the load being applied on the floor, and the load being applied when installed. In a nutshell, will the load cell act the same with a load applied dead center as it will with a load applied say more towards the top right hand corner. And will it act the same if the load is applied in the lower left, etc. Again, the feeling I get from you guys would be that yes, unless the load is applied severly into one of the corners and off center, then the load cell should respond the same to any load placed upon it.
 
Russ,

A tared value should not affect the span or linearity. Any on axis load will be sensed in a linear manner. Off axis loads likely will not. Of course, if you can simulate the actual working conditions, check it that way.
Perhaps you could use your cylinder (f=p*a) for the in place checks?

Thomas
 
I guess I should have stated this, I am not worried about the tared value. This is irrelevant. The question is more, if I am off by 75 lbs at zero, can the expected error be 75 lbs at span, when the load has been applied differently than it was during calibration. How would you define off axis, simply not centered evenly? My guess is that the load is not going to be applied exactly dead center when it is installed. From your last post, I would gather that we really need to find a way to apply a known force to the system and calibrate it rather than calibrating it outside it's physical location and adjusting offset. This is what I expected, but thought perhaps there was a chance.

The p*a method is what we are using now, and it appears to not be accurate enough. We have found that the setup is extrememly sensitive to cylinder rod length. Meaning, when 1 cylinder is fully extended, if the other is not, we see a torqueing effect on the cylinders, which throws everything off. Last Friday we noticed an error in length of about 1/8" sent one transducer to zero and the other to max poundage feedback. This is the second time that I have seen where 2 power sources acting on the same rigid body makes life very difficult. I am in favor of replacing both cylinders and roll with a single cylinder setup and using the p*a method. Doing this would make the total pressure on the roll easy to control and calculate, but, that is not a financial alternative.Thanks for all the input guys.

Russ
 
Last edited:
I'd need to have the load cell model number so I could take a look at what you have. But as a general rule any force that is off axis will not read exactly what you would expect. This is a mechanical question and hinges on the location of the load sensing element and where the load is relative to any load transfer pivots.

That being said, if the location of the load always stays the same the response will be linear. It will just not be the exact number you are looking for. The way your initial post reads, you are not considering 'linear' to mean the same thing I am. You seem to infer that the load cell is only linear if only the bias is affected. However, the load cell can still be linear even if the gain makes a step change between your calibration bench and the machine. It is linear as long as the gain doesn't change with time once it is back on the machine.

Keith
 
kamenges said:
if the location of the load always stays the same the response will be linear. It will just not be the exact number you are looking for.
Keith

This is my question exactly. The load in all probability will not be applied the same when installed in the machine as it was during our calibration of the load cell off the machine. Therefore, I will assume that full calibration will be required.
 
russrmartin said:
This is my question exactly. The load in all probability will not be applied the same when installed in the machine as it was during our calibration of the load cell off the machine. Therefore, I will assume that full calibration will be required.

Russ,

That is correct. The stresses on the load cell will be different depending on if the load is applied axially or offset and they are not linear between them.

You should be able to do it with a ratio though. For example, if a known load is going to be applied 10 feet from the axis of the cell, you can use a load that is twice that amount applied 5 feet away to calibrate it. Now tell the signal conditioner that this is the load you want to measure that is 10 feet away. The conditioner should now be scaled to the weight needed at 10 feet.

Bob
 
The big thing to keep in mind is the geometric orientation. As Bob said you can apply scaled loads at different moment distances and scale accordingly. But you also need to make sure that the calibration load is applied to the load cell with the same spacial orientation as it sees on the machine. Load cells are directionally sensitive. Getting the right orientation of the calibration load is the hard part.

Of course, the load seen by the load cell sensing element is really just a simple geometric relationship. If you know the relationship of the load force to the load cell sensing element you can calculate the factor of the force detected by the sensing element and take that into account in your calibration.

Keith
 
Keith,
You are right. My example assumed a load applied a set distance away but only in one axis. Russ will need to mimic the load's position on a smaller scale(meaning shrunk) to calibrate.

Bob
 

Similar Topics

Last week we had to replace the head on a 40,000 pound scale. Used 2,000 and 1,000 pound weights. Calibration was a breeze, basically just a...
Replies
13
Views
3,641
Good Afternoon, I’m limited with a skilled maintenance staff to design and build Load Cell , Rollers , etc. for conveyor sections . Do...
Replies
11
Views
629
Good Afternoon , Does Rockwell Automation have an Input Card , maybe in the 1734 series , or CompactLogix series that will receive signals...
Replies
15
Views
774
Can someone recommend an Intrinsically safe barrier that can be used with an "ANALOG" load cell on a weigh scale? Load Cell is rated to be in a...
Replies
4
Views
1,343
Hello All, I have two bags of two different types of peanuts hanging on a load cell above two separate feeders. I am using Studio 5000 and would...
Replies
45
Views
8,772
Back
Top Bottom