INA229: Extending Bus voltage range to 400V

Hi Lukas,

using resistive voltage dividers for the current measurement will ruin the awesome common mode rejection of INA229.

Have you thought about using an isolation amplifier?

https://www.ti.com/isolation/isolated-amplifiers/products.html

Kai

Hi Kai,
no, I havent, because I don't understand how using an isolation amplifier would help me dividing the 400V down to max. 85V.

Hello Lukas,

You can actually still use the INA229 (SPI) or INA228 (I2C) if you add your own isolation. Here is an example of how his is done with the INA226 at -48V, but the concept can be applied to your setup as well: https://www.ti.com/tool/TIDA-00313

Essentially, you would create a floating power source for the INA229/8 from the 400V, then just use a digital logic isolator to to adjust the communication voltage levels. 

Hello Mitch,

my question was if it is better to use the resistive voltage divider on the VMON input pin with an additional input buffer (voltage follower opamp) or without one.
I am planing to measure the current on the low site anyways, so no level shifting on the output is needed, "only" an attenuation of the measured bus voltage on the VMON pin.

Hi Lukas,

you didn't say that you want to do the current measurement low side.

For the VBUS measurement you could add a 50:1 voltage divider as shown below:

lukas_ina229.TSC

The error due to the input resistance tolerance (800k...1M2) of INA229 at the VBUS pin would be between -0,50% and +0.33% with this voltage divider (ideal R1 and R2 assumed). Reducing R1 and R2 can further decrease the error but at the cost of increased heat dissipation in these resistors.

Keep also in mind the VBUS offset voltage of INA229 and the decreased ADC resolution when dividing down the VBUS voltage.

Kai

Hey Lukas,

Oh, I also forgot about this reference design: https://www.ti.com/tool/TIDA-00528. This use case is specific to 400V, but does use high side.  In this design they chose to go with the voltage divider over any opamp method, so it will still be subject to error from variance in the input impedance.  

Because of the error from the voltage divider, it would probably be better/more accurate to use the buffer method that you mentioned. I'll forward this thread to that team so that they can help with any opamp design questions you have. 

Mitch M said:

Because of the error from the voltage divider, it would probably be better/more accurate to use the buffer method that you mentioned.

Yes, like this, for instance:

Kai

Thanks, this was also my first idea, but TI uses the simple voltage divider solution on their reference designs and I assumed that they would like to make their devices look good, therefore my question...

Lukas,

Does the 400V and local power for the measurement circuit share a ground? If yes, then CMRR isn't important if there is no common mode voltage.

If the 400V and local power are mostly or completely isolated, then use Mitch's suggestion.    

The sensor is on the negative battery terminal, we use a ISO7741 for SPI isolation.

There is an additional low accuracy backup hall sensor on the positive terminal

Lukas,

The current sense can be ground side because VCM of the INA299 includes ground. Resistor dividing the 400V then buffering will help with VBUS impedance. Voltage is scaled so the power reading will be scaled the same.