Digital bathroom scale circuit - how it works?

I'd like to understand how the digital bathroom scales work. I was able to gather that there are four three-wire load cells connected into the Wheatstone bridge as suggested below. The load cells have three wires which seem to be connected as if there were two resistors (R1A,R1B first cell; R2A,R2B second cell, etc). The resistance of the four load cells is approximately the same, about 1kΩ, and changes slightly under pressure. (Both resistances RA and RB change.) The PCB carries the symbols E+/-, S+/-, which most likely stand for 'excitation' (input voltage) and 'sense' (output voltage).

Can someone explain how this thing works? I understand that the Wheatstone bridge acts as a voltage divider and that the voltage difference is measured between S+ and S-. However, I do not see how can it work with the four load cells connected in this way: if I position myself on the scale perfectly so that the pressure is identical for all load cells, the voltage difference wouldn't change. If the pressure is not the same, then the voltage difference is going to be random. Any ideas?? I suspect that the load cells may be more intricate than I think. Or could it be something else?

Edit: Added a photo of the PCB.

simulate this circuit – Schematic created using CircuitLab

• You say "I was able to gather that there are four three-wire load cells connected into the Wheatstone bridge as suggested below" - is this by inspection or some material from the web. If from the web it would help to provide a link because your interpretation may be flawed. Maybe a photograph would help us too. – Andy aka Jul 5 '13 at 16:35
• More than likely the resistive elements are strain gauges. It is possible that as the "a" resistance value changes the "b" resistance changes in a different way. – Michael Karas Jul 5 '13 at 16:41
• @MichaelKaras I have a small gut-feeling the "B" element might be temperature compensation. – Andy aka Jul 5 '13 at 17:24
• @Andyaka The load cells are connected as shown in the schematics, it is clearly visible on the PCB and also I tested with multimeter. The resistance of A and B changes in the same way - when pressed hard with hand, they both increase by ~3Ω. I will take a photo and post it here later. – YSN Jul 5 '13 at 18:13

OK, problem solved. The bridge is connected like this. Only one resistance in the load cells is variable, the other is fixed.

Why the confusion above? I was measuring resistance of a load cell which came from a different scale. The cells looked pretty similar, therefore I thought they were the same. But they were not! Eureka!

simulate this circuit – Schematic created using CircuitLab

• If the load cells are like the button-clips as in this answer electronics.stackexchange.com/a/199470/30711 then the R?A resistors could be active as well, but moving in the opposite direction of the R?B elements. Your diagram still works in either case. Whether or not the A and B gauges get strained equally in reality depends on how well the load is supported by the clip and transferred into the center bar . – Dave X Nov 11 '15 at 4:53

I found that the "A"-resistors are not fixed but show a mirrored resistance change because they undergo a compression instead of an elongation. This is because there is not a simple arc-bending of the metal arm, but, the special rivet fixation of the upper arm part, results in a complex "S"-forming bending .So both sensors can be glued on the same side of the metal support.Therefore the red wire connection is positioned at the midpoint between the concave(black wire sensor) and the convex(white wire sensor) parts of this curve.Conclusion: the four 3-wire transducers form a 8-resistor wheatstone bridge with indeed 8 active elements

The PCB has two wheatstone bridge (WSB)s, output is the 2 red leads ( differential sig); I am currently hacking a bath scale. Each corner cell contains 2 SG on one side of a cantilever beam with bk, red, and wh leads to the PCB.

For temp compensation, two strain gauges are affixed to one side of the beam. Here's a quote from Wiki:

" This {temperature} is generally compensated for by the introduction of a fixed resistance in the input leg, whereby the effective supplied voltage will increase with temperature, compensating for the decrease in sensitivity with temperature." The fixed resistance is the strain gauge that does temp compensation. Note that temp comp gauge units commercially available are really a gauge and a fixed resistor.

It doesn't matter how the load is applied to the four corners. With a"C-clamp" at one corner I observed a weight reading.

my two Wheatstone bridge circuit schematics are shown in Figure 51. Load Cell Application of the LMV861 instrumentation amp spec. Hookup is: black-gnd; white-5V;reds-V- and V+ to the LMV861 A1 & A2 (+ inputs). Produces a 0-4V output of A4 to an A/D. The outputs are then added.

the parts are on order and will give more data later.

• This might be better as an edit to your other answer to try and improve that a bit. – PeterJ Jan 31 '14 at 2:49

The image shows wheatstone bridge with 4 load cells. This was simulated using circuit Lab as shared in the link: 3-wire load cells and wheatstone bridges from a bathroom scale By simulating it for pressure on all the load cells, the bridge actually is into an unbalance condition