ELECTRONICS LETTERS 28th October 1999 Vol. 35 No. 22 Very low-voltage class AB CMOS and bipolar precision current rectifiers J. Ramírez-Angulo, R.G. Carvajal, J. Tombs and A. Torralba Simple class-AB CMOS and bipolar precision rectiﬁer circuits that operate from a single supply close to a transistor’s threshold voltage are introduced. These circuits have output voltage swings comparable to the supply voltage. Results from simulations of MOS and bipolar precision rectiﬁers at 20 and 100MHz, respectively, are presented. Experimental results of a test chip are presented that verify the proposed circuits. A full wave precision rectiﬁer based on the proposed rectiﬁer cells is discussed. Introduction: Precision rectiﬁcation is one of the fundamental opera- tions in nonlinear systems. New generations of VLSI systems for wire- less applications operate from a single supply voltage of 1.5V or below with the need for low static power consumption and high frequency operation. Low-voltage CMOS and BICMOS rectiﬁers with 3.3V sup- ply requirements and very good high frequency performance have been reported [1, 2]. In this Letter we report a very simple class AB precision rectiﬁer cell that operates with a supply of 1.2V, fabricated in CMOS technology (for 0.85V transistor threshold voltages), and 0.8V using bipolar transistors. It has excellent high frequency performance and very low static power consumption. Utilisation of these circuits in combina- tion with low-voltage mirrors [3] allows the implementation of very low-voltage piecewise linear approximation circuits according to the approaches reported in [4]. This is illustrated with the example of a full wave rectiﬁer. Circuit operation: In the following we assume a single voltage supply V sup . We also assume that |V thP | + V th N > V sup > {|V thP |, V thN } where V thP and V thN denote the threshold voltages of the P and N transistors, respec- tively. Fig. 1b shows the CMOS version of the proposed rectiﬁer cell. It consists simply of two transistors (M N and M P ). The gates of M N and M P are connected to V sup and ground, respectively; therefore V sup = V GSN + V GSP . Given that there is not enough voltage to turn on both transistors simultaneously (V sup < |V thP | + V thN ), under quiescent conditions, a very small (subthreshold level) current ﬂows through M N and M P and V on = V op  0. For positive input currents, M P turns on, and the input current Iin ﬂows through M P to R P . The input voltage takes a value V in = V SGP > |V thP | that is closer to V sup than to ground due to low-voltage supply oper- ation (V sup < |V thP | + V thN ). Given that the voltage between the gates of M N and M P is constant (V sup = V SGP + V GSN ), when V SGP increases, V GSN decreases and M N is turned off (Iin N = 0). For negative input currents M N turns on and the input current ﬂows through M N to R N . In this case the input voltage decreases to a value V in that satisﬁes the condition V in = V sup – V GSN > V sup – V thN which is closer to ground than to V su p. Similar to the previous case, M P is turned off (Iin P = 0) when M N is turned on. Assuming that V sup = 1.2V, and typical values for 1.2 μm CMOS tech- nology V thN = |V thP | = 0.85V and V GSN = V SGP = 0.95 V for both transis- tors, then the input voltage swings approximately from 0.95V (M P ON) to 0.25V (M N ON). This relatively limited swing provides the circuit with excellent high frequency performance. Another advantage for high frequency operation is the fact that the gates of M P and M N are not sub- ject to swings in this circuit. Larger supply voltages (say V sup = 1.5V) result in a reduced input voltage swing (0.95V to 0.55V) and improved high frequency performance. This is the only precision rectiﬁer approach that uses a bias voltage between the gates of the two transistors (V b in Figs. 1a and b) that satisﬁes the condition {V thN , |V thP |} < V b < V thN + |V thP |. Another approach reported in [1] (Fig. 1a) uses a bias volt- age V b > V thN + |V thP | and cannot be used with as low V sup as that pro- posed here. A rectiﬁer cell is presented in [5] that can operate also with very low voltage supply (Fig. 1c). However, this circuit, which is essen- tially a ﬂip-ﬂop, has very large transients that signiﬁcantly degrade its high frequency performance. This degradation is due to rail-to-rail Fig. 1 Low-voltage class AB CMOS precision rectiﬁer a Circuit in [1] b Proposed circuit c Low-voltage rectiﬁer in [5] d Full-wave rectiﬁer Fig. 2 Experimental results of low-voltage CMOS precision rectiﬁer a DC transfer characteristic of positive output b DC transfer characteristics of negative output c Positive and negative output signals d Positive output signal and input (voltage) signal (vertical scale: 0.2V/div, horizontal scale: 0.5 μs/div) Fig. 3 Simulated results a Comparison between proposed rectiﬁer and rectiﬁer in [5] – – – – input signal – · – · circuit in [5] ——— proposed circuit b Transient response of full wave rectiﬁer of Fig. 1d ——— Iin – – – – Irect c Transient response of bipolar precision rectiﬁer – – – – Irect – · – · Iin