INA350: Can I reduce offset by software?

Hey Allen,

There is a small offset on the output when the output is 1.65V, however, there are not enough digits being displayed on the Y axis to show the offset voltage.This offset is different than the "offset" seen when the Ref is 0V.

Ideally, the output of the INA350 in the 1st condition should be 0V, however, your negative rail is 0V as well, so you will run into an output swing limitation.

The output of the INA350 (and almost every amplifier device) will not swing perfectly to V-, as there is some voltage differential that must be on the output transistor.

On the graph, double-click the y axis, and change the value of precision to 5-6.

Best,
Jerry

Sorry, I forgot to click it. the output voltage is 1.65176V.
The offset voltage is less than 8.86mV. Why?

I wanted to simplify the circuit, so I connected VREF to ground and the ADC measured the output voltage, get the offset voltage. Then I connect signal to R7 to get sensor voltage cross R9(this is a resistance sensor). If I can get the correct offset voltage, I can minus it by software. Is this possible?

Simulation result:

1. REF=0V, VIN=0V( IN+ and IN- connect to GND with 330kOhm): Vo=8.86mV -> offset=8.86mV

2. REF=1.65V, VIN=0V( IN+ and IN- connect to GND with 330kOhm): Vo=1.65176V -> offset=1.76mV
For simulation result, 2nd result is more close to INA350 spec. Why is the offset voltage of 1st result too big?

Hey Allen,

The typical offset you should be seeing for the INA350 in a gain of 10 is ~2mV. In simulation you are seeing about 1.7mV of offset. This is the correct offset, not the 8mV when the output is not in a linear range.

You can calibrate with software the initial offset of the amplifier. You will have to establish a minimum value, but you must make sure that the amplifier is in a linear operating range. The amplifier is not in a linear range when you are seeing the 8mV value. This 8mV value is not the offset, it is the amplifier reaching its output swing limitation. This can be checked in the analog engineers calculator (https://www.ti.com/tool/ANALOG-ENGINEER-CALC), or by reading the input and output swing ranges in the datasheet. Once the 0 value is established, your software can define the ADC value that is being read as your 0 value.

This will calibrate out the DC offset, and the next source of error will be the offset drift of the amplifier. From the datasheet, it is 0.6uV/degC, referred to input. Therefore for a gain of 10, it will be 6uV/degC.

Best,
Jerry

Simulation result:

1. REF=0V, VIN=0V( IN+ and IN- connect to GND with 330kOhm): Vo=8.86mV -> offset=8.86mV
This is TINA simulation result.

Refer to the datasheet, input common-mode voltage should >0V.

So if I want to get offset volatge, I should let IN+  and IN-  > 0.2V. Right?

When REF=1.65V, input common-mode voltage should > 1.6V for high CMRR range. 

But the 2nd simualtion result get the correct offset voltage. Why?

I use INA351 for the same circuit. The simulation result is the same as INA350.

When REF>0V,  IN+, IN- =0V, I can get close to spec offset voltage.

When REF=0V, IN+, IN- =0V, I  can get bigger offset voltage than spec.

Refer to upon picture, VG1 is MCU push-pull output. I can controll it forward and reverse that can generate a AC signal on R9.
This circuit can measure the approximate impedance of R9. 

And I want to controll VG=0V to get offset volage, then calibrate it by software. Is this possible?

For simulation result by TINA, Let REF=0.3V or more can get the correct offset voltage.
And when VG=3.3V, (Vo - offset volatge) can get the ideal voltage.

The simulation result is correct?

Trying to calibrate with input at 0V and ref at 0V
-> I confused. You said:  OPA can't output real 0V. Output voltage will >15mV. But this output voltage is not real offset. How to get the offset voltage and calibrate it?

H Jerry

Thanks. So INA350 need a REF like reference design if I want to cancel offset by software. But it will need a extra OPA as ref buffer.

I will choose INA351 for this design.