Interferences between load cells in parallel. How can I solve this ?

Hello,

I have four half-bridge load cells in parallel. Each cell is powered from the same Arduino board and goes to a different analog input over a Wheatstone bridge and an instrumentation opamp.

Basically the schematics below is duplicated four times, however the resistor divider used to set the output offset is shared by the four units.

When I adjust any of the four variable resistors used to change the offset, the output for the three other parts change instantaneously. Also, I have the feeling the range over which the offset goes from 0 to 1023 is quite narrower with the four cells in parallel than with only one cell connected to the circuit.

Should I add a resistor somewhere ? Maybe between the +5 input and the top point of the Wheatstone bridge ?

Thank you in advance.

It should be obvious your resistor values are too small affecting the Vcc voltage. The resistance from Vcc to ground should be 50 to 100 times what it is (50k to 100k between Vcc & GND) . Currently your circuit is drawing enough current to affect the wheatstone bridge because the the wheatstone bridge is specifically designed to be sensitive. If you are using it for load cells you should already know that unless you just copied a schematic not understanding how it works.

raschemmel:
It should be obvious your resistor values are too small affecting the Vcc voltage. The resistance from Vcc to ground should be 50 to 100 times what it is (50k to 100k between Vcc & GND) . Currently your circuit is drawing enough current to affect the wheatstone bridge because the the wheatstone bridge is specifically designed to be sensitive. If you are using it for load cells you should already know that unless you just copied a schematic not understanding how it works.

But the circuit is ratiometric throughout so it shouldn't be affected - this implies that
the 5V at the load cells isn't the same 5V as at the Arduino itself, implying too much
resistance in the wiring I think... If using long leads the amplification should be
near the load cell, not remote from it.

Load cells with a 50k resistance don't exist to my knowledge, strain gauges are
a few hundred to few thousand ohms being a fine pattern of ultra-fine platinum
wires.

BTW the correct way to wire left hand arm of the wheatstone bridge is to use
all three terminals of the pot, with high-precision (1% or better) resistors either
side of the pot track, such that the pot track forms a small fraction of the total
resistance of that arm. Using a pot as a variable resistor will be very sensitive
to temperature (unless its a 10-turn cermet type), as carbon track has a large
temperature coefficient.

MarkT:
If using long leads the amplification should be
near the load cell, not remote from it.

The leads from the cells have a length of circa 20 cm (just a few centimeters longer than the original wires the cells have from their original kitchen scale circuit).

Load cells with a 50k resistance don't exist to my knowledge, strain gauges are
a few hundred to few thousand ohms being a fine pattern of ultra-fine platinum
wires.

Indeed I have measured 500? /cell here.

BTW the correct way to wire left hand arm of the wheatstone bridge is to use
all three terminals of the pot, with high-precision (1% or better) resistors either
side of the pot track, such that the pot track forms a small fraction of the total
resistance of that arm. Using a pot as a variable resistor will be very sensitive
to temperature (unless its a 10-turn cermet type), as carbon track has a large
temperature coefficient.

Do you think 28-turns cermets are OK ? And the resistors have 1% of precision.

So, how can I avoid the each part of the circuit to be sensitive to what is happening in the three others ? Should I have additional resistance in series between the +5V and each bridge ?