Summary In this paper, we evaluate a useful general model for linear direct current (DC) networks, containing design parameters, multiple inputs and outputs. The outputs are rational multivariate functions of the parameters and linear functions of the inputs. For an unknown linear system, the unknown coefficients of these functions can be determined by taking a suitable number of measurements and solving a set of linear equations constructed from selected measurement data. The rational functions are useful to support the design such that outputs can be constrained to a desired interval. An interesting result was the particular case where the parameter varied is located at the output terminal. This is an alternative way to estimate the Thévenin equivalent of a circuit. Real data are in general noisy with accuracy dependent on the tolerance of components and on the measurement device used. We present a detailed analysis of the application of the measurement‐based approach under real experiments. Initially, a resistive DC circuit case is presented, and the results are verified by simulation and experimental laboratory data. Finally, practical circuits, such as a conditioning signal stage with an operational amplifier, and a boost DC–DC converter are considered. Copyright © 2017 John Wiley & Sons, Ltd.