Have you tried my suggestion of introducing a 180K ohm resistor between the current sense monitor point and the positive input instead of directly connecting the positive input to the current sense montoring point? This may avoid the need for the nulling.

hgmjr

But now he is using a Fet-input TL082 dual opamp that has an extremely low input leakage current of typically 30pA (no bias current) for your un-necessary additional resistor.
A 741 opamp would need the additional resistor but since its input resistance is so low then its gain will not be accurate.

 

I have always heard that the inputs to any opamp should see the same resistance when looking back out of the terminals in order to allow the input leakage currents to produce the same offsets to minimize input offsets no matter how low the leakage currents may be.

hgmjr

 

The circuit works perfectly now accept for the offset.

Given that it's a 10 mV/A scale, the error is pretty small. You might be able to disregard the offset bias.

 

The presentation at this link:

http://www.ecircuitcenter.com/Circuits/op_ibias/op_ibias.htm

touches on the point I am trying to make.

The present circuit from the OP has one of the outputs looking into a source resistance of 0.1 ohms while the other (negative input is looking out into the resistance that is equal to the parallel combination of the 1.8M and the 200K resistors. This inequality does not allow the input leakage currents associated with the two inputs to self balance.

hgmjr

 

The NSC datasheet for the TL082 specifies a maximum bias current of 400 pA (typical 50), and a maximum offset current of 200 pA (typical 25) at junction temp of 25C...
If you were to put the matched resistor to both outputs you could use the offset current number instead of the bias current.
So, you'd have 200pA * 180k * 10 = 360 uV offset
Without that resistor you can have up to 720 uV

The offset voltage of the op amp itself is stated as typical 5 mV and maximum 15 mV.
With a gain of 10 you do get an offset on the order of 30 mV

Edit:
Like I said junction temp of 25C, and it says it doubles for every +10C. You end up with just a few mV offset at the output if you double it a few times.

 

I have always heard that the inputs to any opamp should see the same resistance when looking back out of the terminals in order to allow the input leakage currents to produce the same offsets to minimize input offsets no matter how low the leakage currents may be.

hgmjr

But that was for old opamps that used ordinary input transistors that had a pretty high input bias current, not a very low input leakage current like Fet-input opamps have.
The typical input bias current of an old 741 opamp was 80nA which was 2700 times higher than the 30pA input leakage current of a Fet-input TL082 opamp.

Opamps have an input offset voltage that neeeds to be nulled in this application. The offset voltage can be positive or negative. If the offset adjustment pins of a TL081 opamp are used with a nulling trimpot then the circuit will work perfectly.

 

The lower input leakage currents of the bifet opamps do make this provision less critical. I still use the technique when I am looking to squeeze the maximum performance out of my opamp designs.

hgmjr

 

. . .the op amp itself is stated as typical 5 mV and maximum 15 mV.
With a gain of 10 you do get an offset on the order of 30 mV . . .

This is correct. The offset is 30 mV. I am perfectly willing to use a TL081, but no one seems to want to advise on the size of the pot to place between the nulling terminals. I would be selecting at random.

 

I am perfectly willing to use a TL081, but no one seems to want to advise on the size of the pot to place between the nulling terminals.

The data sheet from TI clearly shows the trimmer to be 100K. The fixed terminals go to the nulling input pins, and the wiper is fed by a 1.5K resistor which is connected to the negative voltage rail.