Transistor inversion level independent circuit technique for low-voltage rail-to-rail amplifiers

Rail-to-rail amplifiers are based on an input stage made up of two parallel-connected complementary differential pairs. The paper introduces a circuit technique for maintaining the small-signal behaviour (i.e. the transconductance, g_m) of the amplifier constant over the entire input common-mode (CM) voltage range. The operation principle consists in comparing the total amplifier transconductance with a reference transconductance value and then generating an appropriate bias current for one of the two pairs of the rail-to-rail input stage. Thus, the technique is universal, since it is valid regardless of the g_m/I_D characteristic and the operating region of the input devices. Also, the accuracy in providing a constant transconductance does not depend on precise matching between n-channel and p-channel input transistors, which makes the technique robust. Experimental results from a 0.8-μm CMOS test chip demonstrate that amplifier transconductance variations, obtained as a function of the input CM voltage, are lower than for any previously reported technique.