INA296A: Low-Side Current Sensing Optimization for 100kHz PWM

ArTzy

Mastermind 20330 points

Part Number: INA296A

Hi Experts,

Would like to ask assistance from product expert point of view.

This technical inquiry addresses the application of the INA296A current sense amplifier in a low-side configuration with a 0.5-ohm shunt, measuring a 100 kHz PWM current from 0 to 20 mA while aiming for a 50–100 µA error target.

[1] What is the input offset current (I_os) or the difference in input bias current (I_b) between the inverting and non-inverting inputs for this device?

[2] Can a design utilizing the 200 V/V gain variant (INA296A5) successfully meet the accuracy target of 50–100 µA under these operational conditions?

[3] Can an RC low-pass filter be connected directly to the OUT pin to extract the average current value given the 1 nF capacitive load limitation, and will utilizing a capacitance below 1 nF alongside a larger resistor satisfy the minimum 1 kHz cutoff frequency requirement?

Thank you.

Regards,
Archie A.

1 month ago

0 Javier Contreras4 1 month ago

TI__Mastermind 34700 points

Hello Archie,

ArTzy said:
[1] What is the input offset current (I_os) or the difference in input bias current (I_b) between the inverting and non-inverting inputs for this device?

I normally do not recommend this device for lower than in the µA range due to the variety of µA input bias currents and how much they change.

ArTzy said:
[2] Can a design utilizing the 200 V/V gain variant (INA296A5) successfully meet the accuracy target of 50–100 µA under these operational conditions?

See first answer.

ArTzy said:
[3] Can an RC low-pass filter be connected directly to the OUT pin to extract the average current value given the 1 nF capacitive load limitation, and will utilizing a capacitance below 1 nF alongside a larger resistor satisfy the minimum 1 kHz cutoff frequency requirement?

The capacitance limit if driving the output directly without an isolation resistor.

See my reply and my rule of thumb.

INA296A: Floating noise much higher than spec

Regards,

Javier

0 ArTzy 1 month ago in reply to Javier Contreras4

TI__Mastermind 20330 points

Hello Javier,

Thank you for your advice.

Customer would like to ask more context on question no.1? Could you be more specific regarding this case (20mA with 0.5 ohm sense resistor)?

Also, can you suggest alternative TI solution?

73,
Archie A.

0 Louis L 1 month ago in reply to ArTzy

TI__Expert 8430 points

Archie,

Thanks for your patience. Javier is currently out of the office and will reply when he returns on Monday, 7/6.

0 ArTzy 23 days ago in reply to Louis L

TI__Mastermind 20330 points

Hi Javier and Louis,

Good day. Would like to ask update.

Thank you.

Regards,
Archie A.

0 Javier Contreras4 23 days ago in reply to ArTzy

TI__Mastermind 34700 points

Hello Archie,

I am not sure what you are looking for in a solutions.

If you are getting average and you do not need 50uA error then the INA296 could work. If you do not need high bandwidth as you mentioned the RC filter at the output then we have other alternatives but the VCM transition is the biggest issue. I need to know the voltage switching in my model I did 0-24V. I need to know this to give you another alterative if you require better than 50µA error.

Here is an example of where the error can come from in a Tina simulation. First DC error with 20mA. You can see the difference before and after the the INA296 is 70µA. This number is typical and can vary about ~40µA. It is IB * 2. The TINA simulation uses 35µA *2 (one for each input pin)

Also you can use the RC filter but you will have a delay in the signal. So for the switching if you need measurements in the range of your switching. For DC measurements the RC filter will help and give you low error. For currents that are changing like a 10kHz triangle wave you will get a large error.

Below you can see the nofilter has the spikes from the VCM transitions in the no filtered output but looks closer to the ideal output. The RC filtered output is lower and will give you an average but will not look like the ideal output.

0 Javier Contreras4 23 days ago in reply to Javier Contreras4

TI__Mastermind 34700 points

I have added the TINA file I used to simulate.

INA296_E2E_July7_2026.TSC

0 David Dayan 15 days ago

Prodigy 30 points

Hi Javier,

This is my question originally, thanks Archie for the help.
Also, thanks for the detailed explanations and your simulations. Much appreciated and helpful.

I can see the 35uA*2 both in my simulation and your simulation, but I don't have any error in the INA output after stabilization, it's perfectly stand on 2V.
Is that make sense? What am I missing here?

NOTE that I use the INA with VCM=1VDC, also no transients at all in the common mode voltage!
IN addition, the current waveform is rectangle in 100kHz running between 20mA and 0mA.

See below snapshot of IG1 and the Vout_NoFilter waveforms.

0 Javier Contreras4 15 days ago in reply to David Dayan

TI__Mastermind 34700 points

Hello David,

The error I thought I saw required is 50µA. If this is the max then this could work. The error is more like 35µA but this value will vary like stated in the datasheet. The reason is both pins pull on average 35µA *2. The following is assuming a device is used with 35µA. The current you attempting to measure is normally too low if measuring the power supply current in the example above it is 20.07mA and the output is measuring 20.035mA. If you are attempting to measure the load current it would measure 20mA but the output would measure 20.035mA.

The simulation we showed did not have enough digits on the output. I have shared another screen shot.