Capacitor sizing is a crucial step when designing a chargescaling digital-to-analog converter. Larger capacitor size can achieve better circuit accuracy and performance due to less impact from random, systematic, and parasitic mismatch. However, it also results in much larger chip area and even more power consumption. In addition to minimizing random and systematic mismatch during common-centroid capacitor placement, this paper presents the first problem formulation in the literature which simultaneously considers capacitor sizing and parasitic matching during common-centroid capacitor layout generation such that the power consumption is minimized while the circuit accuracy/performance is also satisfied. Experimental results show that the proposed approach can achieve very significant chip area and power reductions compared with the state of the art.

No abstract

Capacitor sizing is a crucial step when designing a chargescaling digital-to-analog converter. Larger capacitor size can achieve better circuit accuracy and performance due to less impact from random, systematic, and parasitic mismatch. However, it also results in much larger chip area and even more power consumption. In addition to minimizing random and systematic mismatch during common-centroid capacitor placement, this paper presents the first problem formulation in the literature which simultaneously considers capacitor sizing and parasitic matching during common-centroid capacitor layout generation such that the power consumption is minimized while the circuit accuracy/performance is also satisfied. Experimental results show that the proposed approach can achieve very significant chip area and power reductions compared with the state of the art.