Dear team,
Please review below block and abnormal non-linearity section(Blue box) and let me know your opinion why this section appear.
< phenomenon>
- If INA209 is disconnected, abnormal non-linearity section disappear
- If Voltage of INA209(3.3V) change to 3.2, abnormal non-linearity section is moved to about3.2V.
If you need more information, please let me know.
Thank you.
<Test result> 
Hi Mayrim,
I am sorry for confusing. there is misunderstanding with customer. The test of using independent 3.3V supply is not completed. I will update result.
<New question>
1. You mentioned "It is not possible that there is no change of INA209 VS as changing voltage from DC/DC"
- Although changing voltage from DC/DC(Voltage at point A), i think INA209 VCC(3.3V) might not be changed because of using other LDO(12V in/ 3.3V out). let me know your opinion.
2. About edited question,
- If the constant voltage between A and B is normal response, Is it the abnormal response of going up response even excepting jumping step(Blue box)? Which result is normal response? Constant voltage response or going up response as changing voltage from DC/DC(PointA, 1 to 10V)?
Thank you.
Dino,
Did some digging, got some answers. I'll answer your questions first and then give you my thoughts as to your problem:\
1. Yes and no. While I do also mean the non-linear portion of the graph, I am also talking about the rest of it. The common mode should have no bearing on the differential voltage. While their is the period of nonlinearity highlighted, the graph presented shows a linear increase throughout the totaliy of the graph as they sweep through the common mode. This should not happen. This is more than likely due to their choice of such a large shunt resistor (499 ohms). If the impedance of the load is only a few thousand ohms, then this resistor will absorb a significant portion of the load as the line voltage changes.
2. As I did not know the customer load value, I used a set of precision voltage supplies to test a range of voltages through the 1 to 10 V range. All values performed to datasheet spec.
3. No, the INA209 should be referenced to ground.
4. Again, the part should definitely be grounded.
As for the issue, here's what I found: in almost all current shunt monitors, there is an internal switching topology that occurs approximately at the point where Vs = Vcm. Up to the point where Vcm < Vs, the part is primarily powered by Vs. At the point where Vcm > Vs, the part switches inside and begins to draw current from the common mode. This is what allows these parts, unlike most op amps, to accept common modes that exceed the supply. For some reason, we did not characterize this on this part, but I have posted a graph of the phenomena from the INA210 to demonstrate the effect:
Please keep in mind this is from the INA210, so it is NOT the actual behavior that the INA209 will exhibit. I just included it to help you visualize what is happening.
In addition, this nonlinearity is being exacerbated by the customer's choice of such a large shunt resistor with respect to the load. By their graph, a 1V change in the common mode contributes a 4mV addition to the differential. As I stated in point 1 above, typically the change in of the common mode should have minimal bearing on the differential voltage. By using such a large shunt, they are actively drawing voltage out of the circuit that should be going to the load.
Ultimately, I would recommend that they push the value of the shunt down to a much smaller size to mitigate the effects of what they are seeing here.
Hope this helps,
Carolus
