AD7190 with instrumentation amplifier

Dear Sir,

I am trying to design a circuit for a long time and could not accomplished yet. I need to measure a loadcell. With 2mV/V 10kg. In my application the problem is that there is a lot of dead weight on my loadcell.

I am using 10 kg loadcell, my product weight is maximum 1kg, and I need a 0,05gr resolution. that means 20000 counts. I can not use lower range loadcell because there is 2-3kg weight of cup that I am measuring product with.

Previously I have design a circuit with AD7190, I have plug loadcell directly, without instrumentation amplifier. And results were not good. I was no where near my target.

Lately I am thinking of a different alternative. I am planning to use +/- 5 V supplies. That will sum up 10V on loadcell, which makes maximum 20mV for 10kg. and 100nV for 0,05gr. With gain around 800, because my dead weight is 3kg. with 3kg 6mV on ad620, and output of ad620 4,8V change. I will use a DAC on REF pin of ad620 so that when there is dead weight on my loadcell output will be around 1 V. And when there is 1 kg product my output will be around 3V. Since AD620 is not a rail to rail in-amp My output can swing between +/- 4V.

My questions;

1-) Is this approach make sense? Since complete front end designs I need to have 1/200000 count in order to archive my target, which is hard for an engineer. Keeping in mind that I need to make measurements in 0,4 seconds in where there is vibration in environment.

2-) AD7190 is differential input and ad620 is single and output. Will it be problem just simply connect negative input of AD7190 to AGND.

3-) What about Reference pin on AD7190, where to connect positive and negative reference pins.

4-) Is it OK to use regular 7805 and 7905 voltage regulators or should I use lower noise regulators?

5-) Until know I was using opto-couplers when making connection with microcontroller. Since they consume a lot of power in order to drive them is it a extra source of noise for AD7190. Just using 100 ohm series resistors can be better.  Even though generally switchmode regulators with a lot of noise used for powering microcontroller.

Regards,

Dogan

Parents
  • 0
    •  Analog Employees 
    on Jun 16, 2011 6:49 PM

    Hi Dogan,

    your have a 10 kg, 2mV/V loadcell. Is the excitation voltage +5V. So, the total output is 10 mV for a 10 kg weight and there is a change of 1 mV for a 1Kg weight change. To achieve 0.05 grams of resolution, the ADC/PGA needs a p-p noise of less than 50 nV. This equates to an rms noise of 7.6 nV.  The AD7190 has a best case noise of 8.5 nV rms (gain of 128 and output data rate of 4.7 Hz).

    If you used a higher excitation voltage, this would help. For example, a 9V excitation voltage could be used. The loadcell could still be connected directly to the AD7190 without any additional scaling. A resistor divider circuit would be needed to generate the reference. Using 9V, the resolution needed is 90 nV p-p which equals 13.6 nV rms. The AD7190 could be operated at a gain of 128 and an output data rate of 10 Hz. The part would settle in 2.5 seconds which meets yout target of 0.4 seconds. The rms noise of the AD7190 is 11.5 nV so it is within your spec.

    If you use the AD620, you can tie the AIN- pin of the AD7190 to GND as long as the AD7190's buffer is disabled. If the buffer is enabled, the voltage on the AIN- pin needs to be at least 250 mV above ground.

    The reference should be generated using 3 resistors in series connected between +5V and -5V. This will give you a ratiometric configuration which is advisable. The middle resistor in the series would be used to generate the reference. You just need to make sure that the voltage tied to REFIN- is not below ground.

    We have used ADP3303 and ADP3330 voltage regulators on the AD7190 eval boards as these are low noise regulators.

    Regarding the power supply, the analog power supply to the AD7190 must be clean and low noise to achieve optimum performance. So, some filtering between the opto-isolator and AD7190 may be needed.

    Regards,

    Mary.

Reply
  • 0
    •  Analog Employees 
    on Jun 16, 2011 6:49 PM

    Hi Dogan,

    your have a 10 kg, 2mV/V loadcell. Is the excitation voltage +5V. So, the total output is 10 mV for a 10 kg weight and there is a change of 1 mV for a 1Kg weight change. To achieve 0.05 grams of resolution, the ADC/PGA needs a p-p noise of less than 50 nV. This equates to an rms noise of 7.6 nV.  The AD7190 has a best case noise of 8.5 nV rms (gain of 128 and output data rate of 4.7 Hz).

    If you used a higher excitation voltage, this would help. For example, a 9V excitation voltage could be used. The loadcell could still be connected directly to the AD7190 without any additional scaling. A resistor divider circuit would be needed to generate the reference. Using 9V, the resolution needed is 90 nV p-p which equals 13.6 nV rms. The AD7190 could be operated at a gain of 128 and an output data rate of 10 Hz. The part would settle in 2.5 seconds which meets yout target of 0.4 seconds. The rms noise of the AD7190 is 11.5 nV so it is within your spec.

    If you use the AD620, you can tie the AIN- pin of the AD7190 to GND as long as the AD7190's buffer is disabled. If the buffer is enabled, the voltage on the AIN- pin needs to be at least 250 mV above ground.

    The reference should be generated using 3 resistors in series connected between +5V and -5V. This will give you a ratiometric configuration which is advisable. The middle resistor in the series would be used to generate the reference. You just need to make sure that the voltage tied to REFIN- is not below ground.

    We have used ADP3303 and ADP3330 voltage regulators on the AD7190 eval boards as these are low noise regulators.

    Regarding the power supply, the analog power supply to the AD7190 must be clean and low noise to achieve optimum performance. So, some filtering between the opto-isolator and AD7190 may be needed.

    Regards,

    Mary.

Children
No Data