The disclosed technology relates to a method for improving performance of a feedback circuit comprising an amplifier and a feedback network, wherein the feedback circuit has at least one tunable component. In one aspect, the method comprises measuring first amplitude values at an input of the amplifier and second amplitude values at an output of the amplifier, estimating a linear open-loop gain of the amplifier based on both the amplitude values, estimating a linear finite gain error based on the estimated gain and the second amplitude values, subtracting the linear finite gain error from the first amplitude values to derive a set of samples containing second error information, deriving an signal-to-noise-plus-distortion ratio estimate based on the variance of the set of samples and a variance of the second amplitude values, and adjusting the feedback circuit in accordance with the signal-to-noise-plus-distortion ratio estimate.