I tried following two configurations for measuring op-amp offset, which should give the same results by theoretical definition of offset where it is modelled as a voltage source at one of the input terminals.
However, I am getting vastly different results. What might be the reason for this? The second configuration seems to give the correct value.
I assume in the answer below that the short-circuit at the ouptut on the left is not present in the actual circuit and that the op amp is powered by a dual power supply in both cases.
In the first configuration, the op amp does not have any feedback and, thus, the offset voltage at the output equals the input-referred offset voltage (which is the one specified by data sheets) multiplied by the DC open loop gain. This, in the case of the OPA 656 is of the order of \$10^3\$ (as can be seen from the data sheet) and with a huge tolerance. To determine the input-referred offset voltage you’d have to divide the measured output-referred offset voltage by the DC open loop gain. However, the large tolerance that there is on this gain doesn't allow one to accurately determine the input-referred offset voltage.
In addition to the above, it should be remarked that the circuit on the left may not work with op amps having a high open loop gain because the op amp would be likely driven into saturation by any offset voltage greater than a few microvolts.
Instead, in the circuit on the right, the op amp operates as a unity-gain buffer and the measured output voltage coincides with the input-referred offset voltage.
What does shorting the output hope to achieve: -
The right hand circuit is reasonable for measuring offset for most common op-amps.
If the 2nd configuration (right side) is the one closest to the datasheet, then it's most likely because the op-amp was characterized that way.
And that makes sense, since it's probably designed to be used within a negative feedback amplifier.
The 1st configuration is more suitable to characterize a comparator's offset.
