Generalized cube networks for implementing dynamic element matching digital-to-analog converters

Dynamic element matching (DEM) has been used to increase the spurious free dynamic range of digital to analog converters (DACs). In this paper, two DEM networks, the Full Randomization DEM (FRDEM) network and the binary tree (BT) network, are shown to be equivalent to an appropriately connected generalized cube network (GCN). A comparison of these two networks and the GCN shows that the BT network has the lowest hardware complexity for two to six bit DACs and that the FRDEM network has the lowest hardware complexity for DACs with seven bits or more.