Abstract Every transducer needs some deconvolution operation to return the original signal. In the seismometer’s case, this is usually done by using the pole-zero instrument models. This method assumes that the seismometer can be described by a linear model that can be schematized by a ratio of polynomials of the Laplace variable s. This paper shows a deconvolution method that uses the mechanic-electric model of a transducer. Such a model can also give a better description of a transducer since it can also take into account nonlinear behavior which cannot be included in pole-zero models. Examples of deconvolution of both linear and nonlinear transfer functions are shown, and some considerations about friction damping and the electromechanical analogies used to perform the simulations are included in the appendix.