Skip to Main Content
Analog computations such as four-quadrant multiplication, linear voltage-to-current conversion and sum-square or difference-square are fundamental for many analog signal processing systems. All these functions can be realized based on the principle of the linearized differential pair using floating-voltage sources. This paper describes an improved practical realization of this principle, which is particularly suited to analog VLSI computational systems. The proposed class-AB analog cells are very compact, exhibit low total harmonic distortion and low nonlinearity, have a wide bandwidth, and are compatible with low-power and low-voltage operation. A mathematical discussion on stability and harmonic distortion of the proposed realization is presented. Both simulated results and measurements from fabricated cell samples in a 0.8-μm CMOS process are given. The described circuits operate from a single 2-V power supply.