Charge domain recursive filters

A new architecture is presented that permits the implementation of both poles and zeros in the transfer function of analog-sampled data filters without requiring any active gain elements. It is shown that useful filters can be implemented by splitting and summing minority carrier charge packets that represent the input signal in order to form new charge packets that represent the output. Since the feedback and feedforward coefficients in this architecture are all less than unity, absolute stability is guaranteed regardless of the 'Q' of the poles. In contrast to the split-electrode architecture, where the capacitance-voltage characteristic influences the linearity of summation, this operation is strictly linear in charge domain filters. Accurate and highly linear multiplication can be achieved in the charge domain by splitting packets into portions that contain charge in direct proportion to their respective areas.