SparkFun Forums

[mgola]

Hi,

I am using a fixed resistance (350 Ohm each) Wheatstone bridge, the load cell and Arduino Mega 2560. The Arduino gives a random ADC value of -963250. I use the set calibration function on that value and tare after that. Now when I am unbalancing one of the legs of the Wheatstone bridge by 5 Ohm (350 -> 345) I am getting a very high ADC Value.
HX711 reading: 6977316
HX711 reading: 6977258
HX711 reading: 6977327
HX711 reading: 6977260
HX711 reading: 6977281
HX711 reading: 6977297
HX711 reading: 6977305
HX711 reading: 6977340
HX711 reading: 6977299
HX711 reading: 6977383
HX711 reading: 6977331
HX711 reading: 6977317
HX711 reading: 6977332
HX711 reading: 6977387
And by unbalancing the Wheatstone bridge by only 10 Ohm (i.e. from 350 -> 360) it shows the maximum ADC value of 8388608.
I don't know why? Can anyone please help?

[brow]

What are the signal voltages at the various resistance values?

[mgola]

So when I am unbalancing a bridge by +-5 Ohm the output voltage is +-0.017V and for +-10 Ohm it's +-0.035V I have measured using the DMM.
While using the Arduino I am getting a very ADC value which I mentioned above and when converting that value to the voltage it's roughly 4.5V for a +-5 Ohm change in resistance.

Today I tried to put the resistance (10 Ohm) between the output legs of the Wheatstone bridge and found the ADC value which I converted into voltage as 0.022V. It does not match exactly with the theoretical value having an offset. Is there any definite reason for that?

[n1ist]

Not sure what load cell you are using but 5-10 ohms seems like a huge resistance change. I looked at one recently and was seeing a 50-100 milliohm change, giving about a mV change with a 10V excitation voltage.
/mike

[brow]

As you've seen, it's generally better to experiment in terms of signal voltages instead of bridge resistance. Are you sure you're not saturating the input of the amplifier? Depending on the gain, this can be as low as 20mV.

If you connect your load cell, what is the signal voltage at no load? Does this signal rise and fall as you exercise the load cell?

[mgola]

Actually, right now I am not using a load cell as one of the legs of the Wheatstone bridge. I have used definite resistance and varying one of the legs of the whetstone bridge by adjusting resistance. Then the output of the Wheatstone bridge is connected to the HX711 load cell amplifier and Arduino.

Initially, I was saturating the ADCs when the signal voltage difference is about 0.035 V. But then I decided to short the output plus and minus with a resistance to lower the output voltage and come in the common-mode range of the amplifier. Now I am getting an offset which is not fixed.

In this state when I have placed a resistance of 5 Ohm between the output of the whetstone bridge at a balanced state, I am getting 0.00019V instead of 0V. And the difference between the theoretical values kept increasing when I am unbalancing the bridge by 5 and 10 Ohms. I mean it's not a fixed offset in each reading. Can you please guide me on how to get the correct values matching my theoretical calculations?

[brow]

To be honest, I don't know what you're trying to do. There's a time for theoretical calculations but now is time for empirical measurements with a load cell. As before, refer to your measurements in terms signal voltages; that's what the amplifier is measuring.

[mgola]

Actually, what I did is I short the output of the Wheatstone bridge with the 3.3 Ohm resistance to come in the common-mode range of the load cell amplifier. Though the values I am getting from the load cell for delta R = 5 Ohm is 0.022V and by theoretical calculation, it is 0.017V. I don't understand why so? If this is the case how to correct it?

[n1ist]

There's additional resistance in your circuit that you are ignoring. This is in the wires and connectors. What you will want to do is to calibrate the system. Take two readings at known weights (ir in your case, whatever the bridge is measuring) and, assuming things are linear, apply a slope/intercept calibration curve (y=mx+b) to your readings. Any time you are measuring things in the real world, you need to either do your own calibration or take advantage of calibration provided by the manufacturer of the sensors.
/mike

[mgola]

Dear Mike,

Thanks for your suggestion I am also thinking the same. This is the best way to incorporate the errors in the measurement. One last question because I have several parameters floating. Which two parameters do you think are the best to make y=mx+c in between. I have ADC values, resistance, voltage, strain, etc.

Looking forward to your reply.

Regards,
Dr Mohit Gola