Tool/software:
Hello
I am working with a PCB design using an INA228, and it was working fine until we decide to calibrate them, and discover the error on the measurements that we are having is bigger than we expected, at this moment i am trying to configure it using a shunt resistor of 16mOhm, and trying to read the same load with different voltage, resulting on different currents, and as for compassion I am using a DMM Keithley6 1/2 recent calibrated,
this is the design that I am using

at 9 V this is my graphic current in mA

at 13.5V this is my graphic current in mA

and at 16V this is the behavior current in mA

I already tried to change the value on the ShuntCal to 4025d, 4050d, and 4000d, this for the extra values from the shun 16.1, 16.2 or 16.0
change the number of measurements from 1 to 256,
change the time to take the measurement to 4120 us
the voltage is almost close I only have 0.1V of difference vs DMM and this is calculated by (VUS_voltage) * 195.3125/1000000
the LSB current that I declared is 0.000019073486, I got that from 10/2^19
i tried to change the ADC_range to 1, and then according to the data sheet, to establish the SHUNT_CAL, I should multiply by 4 the value, but when I did that, the values were to different, so I leave it by one,
the only thing that I saw is that as i go to higher voltage, the error diminish
i don't know what else to do, since I don't have an standard offset to add, since each measurement the error is very different