Bayesian deconvolution I: Convergent properties