4 Find The Minimum Shunt Value

Upon calculating the max shunt value, the minimum shunt value should also be determined. To determine this value, you need to know the lowest meaningful shunt input voltage value. At minimum, this value corresponds to the shunt voltage register 1 LSB step size. There are some devices, like the INA209, INA219, and INA220 that for certain ADC configuration settings require a register bit higher than the LSB to be used, because the conversion time will not be sufficient to achieve the full possible resolution as alluded to in Table 4-1's test conditions column. Be sure to check the Vsense register to gain better insight into what sense values are possible and how the data is stored. Below is the Table 4-2 for the INA228. Here we can see that the lowest resolvable shunt voltage neglecting noise corresponds to 312.5nV/LSB or 78.125nV/LSB depending on a ADCRANGE setting found in another register. Also we can see the register LSB does not correspond to the lowest ADC measurement LSB, which is important to be aware of when reading from the device.

To find the ideal R_{SHUNT MIN} divide the shunt voltage 1 LSB step size by your minimum expected current value.

The LSB voltage corresponds the lowest measurable shunt value if error sources are neglected. The primary source of error for low current measurements is the input-referred offset voltage. This will vary with respect to common mode voltage, supply voltage, temperature, and device. As most power devices are bidirectional, you should be able to detect changes in your measured current even when the shunt voltage offset is larger than the shunt voltage LSB step size. Depending on your design, setup, and desired level of accuracy, you can calibrate this error out by taking an initial no load measurement at your common mode voltage and current conditions. If you have no intention of calibrating and are concerned about the percent error for your lower measurements, you can calculate R_{SHUNT MIN} according to input referred offset voltage max and the percent error you can tolerate.

The other significant source of error particularly noticeable at your lowest measurement is noise. Your device conversion and averaging setting will determine how significantly this impacts your measurement. As can be seen from the Figure 4-1, longer conversion times lead to measurements with much less noise. With typically the longest conversion time, averaging can also be used to further reduce variability from noise.