• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 47 subscribers
  • Views 2450 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.

INA282 current sense amplifier: Questions about gain

Eric M
Prodigy 30 points
Other Parts Discussed in Thread: INA282, LM4132
I have some questions about the INA282 precision current sense amplifier.  We're using the chip to measure current bi-directionally, and the problem we're seeing is that the chip's voltage gain appears to be 68X with current running in one direction, and about 55X with current running in the other direction.  But, the chip's gain is supposed to be fixed at 50X, according to the datasheet.  In searching for an explanation of our measurement results, we have ruled out many factors external to the INA282 chip.  I would appreciate any help we can find via this discussion board. 
 
Some background information to provide context: 
 
 
Circuit Design and Calculated Performance
 
We have incorporated this chip in a battery charging/discharging circuit to measure current passing through a lithium-ion battery.  The chip is powered by the battery (battery voltage ranges from 2.9 Volts up to 4.2 Volts).  We have designed our system so that the INA282 chip can measure current bi-directionally, and to do so we have replicated the circuit design in Figure 38 of the chip's datasheet, except that we use an external voltage reference chip (an LM4132) that provides a reference voltage of 2.048 Volts, rather than the 5-Volt reference shown in Figure 38. 
 
With the 2.048-Volt reference, split by the INA282's internal voltage divider to enable bi-directional current sensing, and with the 10-mOhm current sense resistor (Rsense) in our circuit, I calculate that the INA282 should be able to measure currents up to 2.048 Amps in either direction:
 
    Input voltage across Rsense = Rsense * I battery = Vsense
 
    Vref = 2.048 Volts
 
    TI-specified gain of INA282:  50
 
    Therefore, output voltage of INA282 = Vout = 2.048/2 - (50*Vsense)
 
    Maximum measureable current should therefore occur when Vout reaches 0 Volts:
 
        ==> Vout = 0 = 1.024 - (50*Vsense)
 
        ==> Vsense = 1.024/50 = 0.02048 Volts = Rsense * I battery
 
        ==> I battery = 0.02048 Volts / 0.010 Ohms = 2.048 Amps
 
 
Measured Performance
 
We have now built and tested four of our charge/discharge circuit boards incorporating the INA282 chip.  All are operating with very good stability and predictability.  But, the currents we are measuring via the 282 chip seem to be wrong -- predictable and stable, but wrong.  And, all four boards are wrong to the same degree.  So far, we have traced the root cause to what appears to be a problem internal to the INA282 chip:  While the datasheet indicates the chip's voltage gain should be 50X, our measurement data indicate that the actual gain is about 68X with current flow in one direction, and about 55X with current flow in the other direction.  We are puzzled and doubtful that a high-precision chip from TI could be so far off from its rated gain, but so far we have not identified another cause of the measurement inaccuracy we are seeing in the lab.
 
Our lab measurements:
 
Using a digital multimeter, we measured the input voltage across the IN+ and IN- pins of the INA282 chip and also measured the chip's corresponding output voltage.  A sample of the data we took:
 
     Current flow in one direction (battery discharge):
 
                        ________________  Voutput (Volts)  ________________
      Vin (mV)     Measured     Theoretical, gain=50     Theoretical, gain=68
 
           2.8            0.833                    0.884                        0.834
           4.0            0.748                    0.824                        0.752
           8.2            0.468                    0.617                        0.470
         11.6            0.234                    0.444                        0.235
         13.8            0.084                    0.334                        0.086
 
 
      Current flow in other direction (battery charge): 
 
                        ________________  Voutput (Volts)  ________________
      Vin (mV)     Measured     Theoretical, gain=50     Theoretical, gain=55
 
          1.0            1.084                    1.074                        1.079
          4.6            1.282                    1.254                        1.277
          9.8            1.562                    1.514                        1.563
        15.7            1.894                    1.809                        1.886
        19.8            2.110                    2.014                        2.113
 
 
As shown in the tables above, for any given input voltage, the measured output voltage does not match well with the theoretical output voltage if we assume the INA282's voltage gain is 50.  If, however, we assume the chip's voltage gain is 68 in the first table (battery discharge current), and 55 in the second table (battery charge current), we see a very close correlation between theoretical and measured output voltage.  And, we're seeing similar, very stable results from all four of the the circuit boards we've built.
 
We find it hard to believe that the chip's gain is not fixed at 50X, as surely TI has taken steps to ensure that the gain really is as shown in the datasheet.  And yet, the lab measurements we've taken are fairly simple and hard to argue with:  we measure the input and output voltages on the chip, and calculate a corresponding gain.  So far we have not identified anything in our circuit external to the chip that would alter its gain, especially when we get the same, very stable results on four different circuit boards using four different INA282 chips.  And, we have not been able to explain why the gain would be different in one direction of current flow vs. the other direction.
 
I'd very much appreciate your help in figuring out this problem.
 
Thanks very much,
 
 
Eric
 
  • 0
    TI__Mastermind 25805 points

    Hello Eric,

    Thank you for the detailed explanation and data.

    Let me start by stating that I agree with the belief that it’s probably not the gain of the device itself.  What we will probably find is that the data you provided is within specifications once all of the errors are taken into account.

    The first source of error may be related to the DMM that is being used to measure the sense voltage.  Assuming you’re using a Fluke or similar device, please bear in mind that the resolution may be just 0.1mV with 0.3% accuracy.  I would also like to thank you for taking the measurement directly across the sensing pins and output pin of the device.  This circumvents any possible issues with the layout. 

    Additional sources of error include, but are not limited to, the following:

    • Initial offset voltage (Vos)
    • Offset voltage drift (dVos/dT)
    • Offset voltage shift (dVos/dt)
    • Offset voltage induced by change in common mode (Vos_cm)
    • Offset voltage induced by the reference voltage (Vos_ref)
    • Offset voltage induced by the power supply (Vos_ps)
    • Gain Error
    • Non-linearity Error
    • Reference voltage accuracy/regulation

    There is a good explanation in the INA282 datasheet of how most of these errors combine.  For example, if we assume your Vcm=3V we can justify up to 70uV+9uV+76.8uV=155.8uV of input-referred offset just due to Vos, Vos_cm, and Vos_ref.  This offset voltage represents 5.56% error on Vsense=2.8mV and 15.58% on Vsense=1mV.  That calculation uses a worst-case scenario that is not very likely, but if we RSS the values we still obtain 3.7% and 10.4% error, respectively.

    Now the question becomes:  What accuracy do you require and under what conditions?

    Finally, have you tried connecting a reference voltage as shown in Figure 37?  This would remove accuracy issues due to the voltage reference.  I would be interested in such data.

     

  • 0 in reply to Pete%20Semig
    Prodigy 30 points

    Pete,

    Thanks for your prompt and thorough response.  We'll look into each of these sources of error, as you suggest.  Perhaps the errors are just accumulating.  The way we've configured the chip, when we're measuring battery discharge current, the INA282 chip output voltage is lowest when current is highest, and we're trying to resolve some very small voltages.  In this circumstance, even small errors in measurement, offset, etc., result in large percentage differences.

     

    Eric