How do Multi Load Cell Scales Work?

The sum of the forces on the four feet (load cells) equals the force downward (your weight). Each load cell has a force - resistance relationship. So the scale just sums the four forces. (It is zeroed out before weighing to cancel out the weight of the scale.)

Possibly using many load cells would result in better accuracy if the inaccuracy were greater than the resolution of the scale, but would likely be more costly.

It may be that there is a compensation for the inaccuracies of the load cells implemented in the digital logic of the scale, which would be much less expensive than additional load cells. (At least that is how I would design it.)

The most accurate if you could make it single load cell.

My experienced in the elevator industry is elevator usually has 4 cells installed on the corners of the car's platform. We did a routine annually calibration and check. All 4 had different readings (outputs) but it could sum up to one.

All 4 had different readings (outputs) but it could sum up to one. So, you're saying you don't have a standard you calibrate to? What's your Zero? Please let us know who what company you work for, least ways we'll know if we should step back out and take the stairs.

Thank The All Mighty Above- you're not doing W/B on helicopters!

what I meant is, just imagine you step in a square platform with four corners supported with weight cells, can you say all four cells will have exactly the same force? Obviously not.

Checking load cells in elevators if they have any irregular out puts and adjusting mechanically at some preferred min an max. Verifying that each load cell is not initially overstressed so it will not be defective prior to its life span.

Ofcourse, O load for no load conditions and max at maximum rated Car Load and over checking if control system would recognize overload and eventually annunciate the case to precious passengers. We use standard weights ofcourse.

We follow company standards throughanuals.Its an american brand so nice to mention but I wont.

what I meant is, just imagine you step in a square platform with four corners supported with weight cells, can you say all four cells will have exactly the same force? Obviously not.

Checking load cells in elevators if they have any irregular out puts and adjusting mechanically at some preferred min an max. Verifying that each load cell is not initially overstressed so it will not be defective prior to its life span.

Ofcourse, O load for no load conditions and max at maximum rated Car Load and over checking if control system would recognize overload and eventually annunciate the case to precious passengers. We use standard weights ofcourse.

We follow company standards throughanuals.Its an american brand so nice to mention but I wont.

The four readings are likely to be different unless the weight is perfectly distributed, but the sum of the four upward forces equals the downward force (weight). Simple physics.

Yep, I agree Rix.

wrong. bad advice.

There is no need to have a bathroom scale to an accuracy of 1gm, as the weight of the body varies through the day, and a scale to this level of detail will always produce a band of readings because of it; all one could do is draw a best-fit line through the data points, which is what a less accurate scale does without the user realising.

• A scale to the accuracy of 1gm is valuable at the Post Office for measuring the weight of airmail letters in order to work out postage charges.

most amp break out boards for load cells do not have the capability of load cell averaging, but the important ones do. If you have four load cells (one on each corner) for example, averaging the input of each of the four will compensate for off-balanced loading by person standing with more pressure on one foot that the other.

Generally speaking, yes, the load cells will cover each the range of possible mass loading. You cannot make the scales work better with a single load cell, or by trying to add up the response of each (it effectively does this by averaging the four load cells.)

Actually, you don't need an average. You really do need the sum of the 4 cell outputs. As to rating, each load cell must be capable of much more than 25% of the total rated load, because of the possibility that the feet are placed asymmetrically. The cynic would also point out that there are lots of humans that weigh more than 100kg, so a 200 kg limit might be more appropriate

Well, I will point out that taking the sum is an effective averaging of the four inputs, except for the dividing by four part. Yes, the total load is the sought variable, so you get the point(s). I agree on using the 200 Kg load cells for that also.

Basically 4 Strain gauges /load cells are placed /connected and configured as in wein bridge configuration..Where Excitation potentials are fed on diagonally opposite corners while the output is taken from the other two diagonally opposite corners of the bridge..

Depending on the complexity of application, 4 separate load cell bridges at times are mounted or placed on the four corners of the weighing platform..

Erm, no. They most certainly are not configured as a Wien bridge, on bathroom scales at any rate. You may be relating to the internal construction of load cells, which may use 2 or 4 strain gauge elements in a Wheatstone bridge. The analogue outputs from such load cells would have to be read independently or connected in series to be added up.

The Wien bridge is specifically devised to read the reactance of capacitors, and finds a place in audio oscillators

I know that I can buy a quad load cell amp from Sparkfun, that is called, OpenScale. It is fun to use if you can work out the communication protocol, or just wish to read it out to a serial monitor to see if load cell(s) is/are working.

There is a place on the OpenScale breakout board for each of the four inputs to a quad weighing station.

Do bathroom scales need to be that accurate?. My wife knocks off the fractions to the lower whole pound reading anyway.

Your lucky your wife only rounds down to the next pound. No matter how well I calibrated her scale, using standards, she always claimed it was wrong, reading 5 pounds heavy, I thought it always read 10 pounds to light!

Something you keep to yourself I guess!

One of my wives just used to round down to the nearest hundred weight. LOL

A couple of points: The poster that referenced a wien bridge meant to say wheatstone bridge. Yes real load cells are 4 element to resolve the linear stress measurement correctly by removing off-axis strain from the story. I spent some time on a 4-post kitchen scale and they were half-bridges, iirc. I was amazed by the resolution and accuracy for such an inexpensive device. Don't know if there was cal info stored in flash or not.

The other point was about accuracy of 4 cells vs 1 raised in one of the orig posts. Each cell has an offset and gain error (approximated as linear but not perfect). Offset error(s) will be generally removed by the tare-ing process (measuring and subtracting the reading with nothing on the scale). Gain errors, expressed as % of reading, will add in the case of 4 cells being summed to be the same as if a single cell, assuming the same cell type and a random error on each one. (Yes there may be some characteristic of the gain curves that would affect this assertion, but generally it stands. More nonlinearity at light loads, etc.) (Also, compensating for offset error depends on being able to unload the scale. Affected by temperature and such.)

-HughV

Yes there is the propagation of error involving in summing up strain measurements. However, let me be quick to point out that if they are indeed half-bridge strain gauges, the summation (averaging) would be only between the sums of two sets of corners, thus not as much propagation of gain error %.

I believe the chips used even in bathroom scales have some sort of zeroing offset on power up (start the program in the microprocessor), and the gain is set constant in the load cell amplifier board section. The cal in some instances could even be a constant floating point number stored as part of the program. Many of the stuff available out there for makers even has an ambient (on board or off board) temperature function, but it does not directly compensate the load cell(s).

i remember something like this - when i was at that business ???

Sorry, but your bunny is Cherenkov glowing:

I think your diagram has some shorts in it, and they are not boxers, so do be brief in your response.

I think if you want to know how these kitchen scales work, you would better buy some scales to make analysis

Several online vendors also sell the kit for this, complete with four load cells, and a fast output board (plug and play), after you solder the leads to the places on board for each load cell connection.

I have one, but only use it thus far in single load cell mode, and you get the mass showing up on serial display within one minute of start-up, assuming you make the hook-up correctly (which is hard to miss the color-code).

Maybe some brands kitchen scales on Amazon can give more help

Most of all kitchen scales I have ever purchased contained a single load cell, not a combination of four. Bathroom scales may be the ones more likely to have multiple load cells, is this how things are in China?