[SOLVED] Offset voltage in a precission opamp exceeds the specifications

[emontllo]

Hello,

I am designing a current sense circuit based on a Kelvin current sense resistor and an amplifier. View attachment current sense preamp.pdf

The preamplifier (Gain 30.3) is a differential amplifier based on a precission opamp (OPA4727 https://www.ti.com/lit/ds/symlink/opa4727.pdf ).The datasheet says that the maximum input offset voltage is 175 uV. However, output offset voltage at zero current is 25 mV wich means that input offset voltage is (a lot!) higher than the specifications.

As can be seen all resistors are 0.1%, operation temperature was 26 ºC.

Is it normal to have such a high offset? Which is the reason? Some ideas...
- Input offset is specified only at 10 V supply only?
- Common mode voltage affects offset?
- Input bias current...

Thank you for your comments and answers!

Ernest

 

[keith1200rs]

When you say 25mV output offset - relative to what? Your output will be 1.65V nominal, I would expect (not 1.5V). Maybe try with the 1mohm resistor connected to A_REF in case the problem is that the offset increases near the rails. The offset is specified at VCM=0 with +/-5V supply.

Keith.

 

[FvM]

Unfortunately, Vos versus Vcm isn't specified for the device, so you can only guess about the behaviour. But it's dedicated to single supply operation, so I won't expect increase of Vos near V-. Did you actually assemble 0.1 % resistors?

 

[emontllo]

Thank you Keith,

There was an error in the schematic, the reference is 3V not 3.3V.

The 25 mV offset is relative to 1.5 V reference. It is not possible to connect the 1 mOhm resistor to A_REF, the only way to reduce the common mode voltage would be supplying the Opamp from +-5V.

The datasheet talks about 150 uV/V sensitivity to power supply voltage. Provably the difference between nominal (10 V) and operating voltage is the reason for this offset: (10 V - 5 V)* 150 uV/ = 750 uV

This makes sense... If this is right, my answer changes to:
-Is it normal to specify 150 uV "max input offset voltage" in the datasheet and have an offset drift depending on supply voltage that makes the real max offset 750 uV?

Ernest

---------- Post added at 14:10 ---------- Previous post was at 14:08 ----------

Dear FvM,

Yes 0.1% resistors are assembled, I actually measured their value before mounting them.

Ernest

 

[FvM]

The datasheet talks about 150 uV/V sensitivity to power supply voltage.

Yes, it's the maximum value. In so far, you could end up with an offset voltage in the said order of magnitude. Hitting worst case parameters on first attempt still sounds unlikely. But you should lift pin 11 and connect a negative supply to clarify.

 

[keith1200rs]

It is not possible to connect the 1 mOhm resistor to A_REF,

I am only suggesting that as a test to try to identify the source of the problem, not a final solution.

Keith.

 

[emontllo]

I have tested the circuit according to your comments:
At +- 5 V power supply the Vios is within the specs, when supply voltage is changed it gets worse (about 100 uV/V of power supply voltage).

Having 0 V common mode voltage also reduces offset (like 100 uV) but this cannot be implemented in the final circuit.

I understand then that the IC is package-level trimmed at +- 5 V but does not guarantee Vios at the full operating range. I will probably use an other IC and/or compensate offset by firmware, the temperature drift is not as important and power supply voltage is very stable.

Thank you,

Ernest

 

[FvM]

I fear, you need to change to a "High Precision" OP "Optimized for single-supply operation" that deserves the name.