Adaptive-rate inductive impedance based coin validation

Abstract

Electro-magnetic (eddy current-based) sensors are widely used in vending and other coin machines, as electrical conductivity (sometimes measured on various frequencies, for multi-component metallic coins) is a significant distinctive physical property of the coin. Idea of current research is to develop precise and high-speed recognition and then validation of the coins to be realized at very reasonable cost of the hardware, using solely inductive sensors (eg simple coils). Precise measurement of the complex impedance of sensors is needed at multiple measurement frequencies simultaneously. Such approach needs either powerful processor and wideband precise analogue interface, or alternatively smart signal processing algorithms to be implemented on low-cost low-power signal processing platforms (preferably, realized as one-chip solutions). In current paper the following approaches has been proposed and evaluated, for improved and “smarter” signal processing, for reduced complexity: 1) using of samples of the signals with (at least) two ("sparse" and "dense") rates, while dense samples of signals (and sampled base-functions, for Fourier and similar transforms) are used only in the region, where coin is detected and preliminarily recognized “by sparse processing”, so reducing significantly the computational complexity of Fourier transform, for finding complex impedance on frequencies under interest; 2) also, the number of frequencies, is increased for precise validation region (e.g., from 1 to 4, in given examples). Estimation algorithm to find the “right” time instance, for “dense snapshot region” has been proposed, to make the “dense” region of processing as short as possible. Results of the investigation, based on real-life data are presented. Also further ideas for even more efficient smart signal processing for given applications are introduced.