• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 49 subscribers
  • Views 922 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

INA219: INA219A - I2C Current Monitor inaccuracy

Kiran Kerai
Prodigy 200 points
Part Number: INA219

We are using INA219AIDCN in our design and the shunt voltage (current measurements) seem inaccurate, there seems to be an offset.

The shunt resistor we are using is 10m ohms and device is set to PGA gain and range /8 (320mV range).

With 12V connected to the high side of our shunt and no load, i.e. nothing connected to the low side of the shunt, we are reading back 43mA.

Does this device need to be offset calibrated to achieve the accuracy stated in datasheet?
How can we improve accuracy? Is increasing shunt value the only way because the device is fixed to LSB=10uV?
What is the ADC error in bits?

  • 0
    TI__Mastermind 26660 points

    Hey Kiran,

    This does is a high offset. The device does not need a calibration to achieve specified performance.

    Can you please send a schematic and/or information of the supply voltage and grounds? Can you make sure that the INA219 GND pin is tied to the 12-V ground return with low-impedance connection. Also, are there any input resistors at IN+/IN- being used. ADC gain error can be up to 0.5%, but this seems to be an offset error issue.

    Would you please send the calibration register value along with the max expected current chosen when calculating the calibration register value?

    Sincerely,

    Peter

  • 0
    Prodigy 200 points

    Supply is 3.3V grounds are same for 12V and 3.3V, low impedance ground connection. Decoupling on 3.3V supply as per datasheet.

    We have 10 ohm input resistors, one to IN+ and another to IN- and 100nF differential capacities across IN+/- for noise.

  • +1 in reply to Kiran Kerai
    TI__Mastermind 26660 points

    Hey Kiran,

    I believe the majority of the offset error is coming from input 10-Ω resistors. For many of our devices, these input resistors would not add significant offset; however, the INA219 has a larger input bias current imbalance comparatively. This is because of an effective internal 320kΩ resistance from IN- pin to GND (see datasheet electrical characteristics table). Thus, the imbalance between IN+ and IN- is 12V/320kΩ = 37.5uA typical. As this flows across the 10-Ω input resistor, this translates to a 375uV input offset error (Vos_filt) that device is sensing. Given the 10mΩ shunt, the device reads this a 375uV/10mΩ = 37.5mA load current, which is definitely close to 43mA reported.

    I would recommend retesting after removing both 10-Ω resistors to see if the problem is fixed. If you need the input filter and cannot perform a one-time offset calibration procedure, then you could add an external 320kΩ resistor from IN+ to ground to balance input bias currents at both pins. The only problem is that it will not completely remove error because the internal 320kΩ resistor is not precision trimmed and can conservatively vary in absolute value by +/-20%. I calculate the left-over offset error (only from filter) to be within +6.25mA to -9.375 mA (this assumes the external 320k to be ideal with 0% tolerance).

    Hope this helps.

    Sincerely,

    Peter