The improvement of a recently developed method for closed loop nonparametric system identification using multiharmonic perturbation signals is described. Intended primarily as a method for the periodic online retuning of three-term controllers, the original method is shown to benefit from an improved selection of its design parameters. A brief review of the original algorithm is first given, followed by discussion of the improved parameter selection and an illustration of its benefits, with the aid of a typical numerical example. The result is a computationally efficient algorithm enabling nonparametric estimation of the process around the critical design frequency of interest, while also maintaining minimal disruption to the process output during the identification phase. Periodic online retuning of the controller is then possible without the need for any further relay tests beyond its initial commissioning.