# How to calibrate opamp to eliminate input offset/bias current and offset voltage?

I am doing on a project using opamp OP07 to amplify a current. Here is my schematic:

The problem is, the output I obtained is not highly precise as I expected. For example in the figure above the $V_{output}$ should be $-2.2V$ which means $\tilde ~-4.84V$ but I am obtaining $-5.14V$. I know this is due to many parameters: input offset current, bias current, offset voltage ...

I just wonder is there anyway to make output more accurate? Any calibration method available?

• You say you're trying to amplify a current, but what I see is a voltage amplifier. Are you sure you're doing this right? – Hearth Apr 30 '17 at 14:22
• Sorry but the 2.2 voltage source is connected to a 1k resistor , it implies the current . Since in reality , i dont have any method to supply current directly , what i only have is just a 2v2 adapter – Totally New Apr 30 '17 at 14:30
• Well, a current sense amplifier works differently than this. What you have here is a voltage amplifier. – Hearth Apr 30 '17 at 14:31
• That said, you should be getting -4.84V out in an ideal case. The OP07 has offset null pins, have you tried those? But also, the OP07 is not exactly a high-performance device. If you're really concerned about precision, get a chopper amp. – Hearth Apr 30 '17 at 14:33
• Thank you @Felthry . Do you have Any suggestion opamp ? – Totally New Apr 30 '17 at 15:09

You can calibrate this circuit by adjusting the 2.2K resistor, or dividing the feedback. The OP07 is a fairly good precision op-amp even by today's standards- high gain and low Vos (and none of the weirdness of RRIO and zero drift amps)- most of your error is almost surely coming from the resistor tolerances (the sole exception I have in mind is if you have somehow induced this stable old op-amp to oscillate by loading it with a capacitance or something like a long cable- isolate anything like that with at least 100 ohms). And make sure you have bypass capacitors such as 100nF ceramic to ground from the two supplies fairly near the chip. Your error now appears to be about 6% which is considerably higher than the 1-2% I would expect to see typically with 5% tolerance resistors so I think there is some reason for concern.

So suppose you want a transimpedance gain of 2.200V/mA, you can use a 2.10K 1% resistor and a 200 ohm trimpot connected as a rheostat. Or (better) use a 2.15K 0.1% resistor series with a 45.3 ohms 1% and a 10 ohm cermet trimpot.

simulate this circuit – Schematic created using CircuitLab

If your input is a true current source you might do better to use a 10.00K 0.1% or better resistor for the input as that will minimize the effect of the small Vos of the OP07 (75uV range). Measure the resistor first (with a high accuracy meter, obviously) and adjust the input voltage to compensate for the resistor tolerance. In other words if your 10K 0.1% resistor is 10.003K, apply 10.003V rather than 10.00V for 1mA.

The OP07 bias current is a few nA which is negligible in comparison to your input current. The open gain is typically 450,000 so a few uV at the input will move it full output swing.

You should have no trouble getting within better than +/-0.1% if you follow the above suggestions.

P.S. Another reason to use precision resistors is that the stability and temperature drift of such resistors is typically much better than 5% resistors. If you use a carbon film resistor and a cheap carbon trimpot you may be able to adjust it to within 0.1% but it isn't going to stay there for long- time or temperature changes will cause it to drift. A typical 0.1% resistor might have a temperature drift of +/-10ppm or +/-25ppm per degree C maximum, with commensurate time stability.

There are three ways you can increase the accuracy of this circuit.

1. Use 1% resistors. At $2200$ + 5% = $2310$ while $1K$ -5% = $950\Omega$ yielding $V_{out} = 5.35V$

2. Do not connect the plus pin directly to ground, go through a resistor to balance the bias current = $2200||1000 \simeq 685\Omega$

3. Use the offset adjustment pins as indicated in the data sheet for the device.

If you need it more accurate than that, you should use a trimpot in the gain circuit and or use a better op-amp.

• thank you for you suggestion I tried 1 , 2 , 3 . The results seem to get better (-4.84V ideal , -5.08V obtained in real ) . I wonder if i can use something like calibration ? – Totally New Apr 30 '17 at 15:11
• @TotallyNew What voltage do you get when the input is tied to ground. If it is non-zero, see item #3. – Trevor_G Apr 30 '17 at 15:13
• @TotallyNew - Since you have a digital meter to measure your output voltage, try actually measuring your resistors. Then plug those values in and see what gain you get. Hint - since your new output has an error of 5%, it's pretty clear that you are not, in fact, using 1% resistors. Also, for what it's worth, an OP07's input currents are so low that you really don't need the extra resistor to ground. – WhatRoughBeast Apr 30 '17 at 17:21

To calibrate something means you have an influence on it. In the case of bias current, you have no influence on it.

Instead, you can design around it, through nukling for example. There are techniques on how that can be done.