INA901-SP: INA901-SP

Hi, thanks for your answare.

I think I didn't explain the problem clearly. In the datasheet it is written that the device has a gain of 20V/V,
which means that on pin5 (OUT) there is a voltage 20 times greater than the voltage across the Rsense. 
In the simulation circuit downloaded from the TI, if the Vcm voltage is positive (for example 10V), considering the connection as datasheet specifies, 
that is pin1 connected to load side of shunt resistor and pin 8 to the other side of shunt resistor,
on the pin5 a voltage 20 times grater than the voltage across the Rsense is present. 
If the Vcm is set negative (for example -10V), on the pin5 the voltage is not 20 times grater than 
the voltage across the Rsense anymore and the voltage on pin5 is more or less 300mV regardless of what it is the current across the Rsense.
If the connection with negative Vcm is reversed, on pin5 there is a voltage 20 times grater than the voltage across the Rsense.

Lastly,

if you look at the following two figures, you will find a difference in the inputs of A1 and A2. What is the correct figure?

  

I think that you must revers the connection if the Vcm is negative because what is important is the sign of dropout across the Rsense.

Hello,

The second diagram, the Functional Block Diagram, is the correct diagram.

I looked into the simulation model and it looks like the minimum swing to ground is around 300mV. This is a worst-case scenario for the part (see section 7.4.2, Accuracy Variations as a Result of VSENSE and Common-Mode Voltage) so you will likely see better performance from the actual device.

Even with negative common-mode voltages, the device should be connected the same way it is connected with positive common-mode voltages, as long as the direction the current is flowing is from positive to negative across the shunt resistor. Reversing the inputs should only be done in order to measure negative currents.

Let me know if you have any more questions,

Levi DeVries

Thanks for answare.

If you have a negative Vcm, also the current is negative. So, if you keep the connection without revers it, the voltage drop across the Rsense is reversed respect the current referred to positive Vcm. Moreover, why the gain is not 20 times grater than voltage across Rsense with negative Vcm?