# Problem of noise with load cell and INA125P + Arduino Uno

I am here with a problem that I cannot solve. Here are the facts:

I have a load cell : http://www.robotshop.com/en/micro-load-cell-50-kg.html that I want to use in both directions, from -30kg to 30kg approximately (it will be a force sensor in some applications).

To obtain the signal coming from the load cell on my Arduino, I am using an INA125P.

Due to technical constraints, I need to use a 9V battery to power the whole system. Also, needing to measure the force in both directions, I need the signal that comes out of the INA125P tp be referenced to 2.5V approximately. Here is the actual schematic:

My problem: When I am in certain locations, where probably there is a lot of parasitics, the signal from the INA125P varies about 5% when my hand is near the load cell, but not touching it! It only happens at some locations but not others. For example, at home, there is no problem; the signal is very stable. Also, the amplitude of the noise seems to be proportional to the gain of the amplifier.

What could be the problem ?

Thank you

UPDATE 6/05 : To be more precise about the variation, here are some examples: - When the load cell lays on a table, no forces on it, the signal (after ADC) is 440 (the value are from 0 to 1024). If I approach my hand close to the cell without touching it, the signal drops to 425 and if I touch it, it drops a little bit more to approximately 400. - When I have the load cell in my hand, and I lift my arm up of about 20 inches, the signal goes up to 460, when I come back to the table, the signal returns back to its initial value.

Therefore the signal is not oscillating, and the variations are always approximately the same, it is not aleatory.

• Probably RF pickup - try putting 10n capacitors (or greater) from both inputs to ground. If it improves but no enough try putting 10k resistors between the bridge outputs and the inputs of the INAMP (keeps the caps at the INAMP inputs as well). What bandwidth are you requiring and what layout have you got including the wiring of the bridge? Commented May 5, 2014 at 20:56
• For the capacitors, which inputs do you mean, these from the bridge or from the battery ? Commented May 5, 2014 at 21:25
• For the capacitors, which inputs do you mean, these from the bridge or from the battery ? Also, I don't really know what you mean as bandwith, is it in terms of voltage ? If it is, I don't have really criteria. If there is some limitations, I will play with the gain to have my upper and lower limits at these voltage limitation. For the layout, it is something for a prototype so I use only prototyping boards. Thank you for your answer Commented May 5, 2014 at 21:34
• See spehro's answer for a more appropriate reference plus, don't expect miracles on prototype board. Commented May 5, 2014 at 21:37
• This load cell might not work for negative loads. See the datasheet, robotshop.com/media/files/pdf/datasheet-3135.pdf -- "Load cells are designed to measure force in one direction. They will often measure force in other directions, but the sensor sensitivity will be different, since parts of the load cell operating under compression are now in tension, and vice versa." It might be nonlinear, w/ different sensitivity, or it might not work at all. Commented May 5, 2014 at 22:04

You should have a bypass capacitor across the power input (100uF/10V in parallel with 0.1uF ceramic will do), 10n input caps on each input line, as Andy suggests, with series resistors, and I would also suggest a 10n capacitor from reference output to ground.

If the Ardno and INA125 are separated by more than a few inches, a resistor of at least 1 K on the output of the INA125 and another capacitor from the Ardno input to ground.

• Is it correct like this ? Schematic. Commented May 5, 2014 at 21:54
• Not quite-- I'll fix it on about 45 min when I can get to a scanner. Commented May 5, 2014 at 22:50
• Use a shielded twisted pair from the load cell with the shield connected to the load cell and/or conductive housing but NOT connected to ground at the INA or INA housing. Commented May 6, 2014 at 6:05

Use a shielded twisted pair from the load cell with the shield connected to the load cell and/or conductive housing but NOT connected to ground at the INA or INA housing.

Try in this order. 1) Twisted Pair. 2) shielded twisted pair grounded at sensor end. 3) Two coax lines with shields connected at sensor end.

If your INA is not mounted to the sensor, try to move it closer. It is a good rule to amplify as close to the sensor as possible.

The INA ground and the Arduino ground need to be separate and meet at a single point where power comes to the circuit.

If you are making a PCB, you can guard the INA inputs with a ring of copper that is connected to the INA ground. Check the datasheet of the INA for layout rules.

What reference are you using for the Arduino ADC? Power supply noise will affect it more than the INA, depending on what you are using. The INA has very high power supply noise rejection.

• For the ADC, I use the default reference (5V). The final prototype will use an Arduino mini pro and in this one, its impossible to use an external reference. To be in the same conditions, I prefer to use default reference. Commented May 6, 2014 at 19:14