Low-Voltage Low-Power Fully-Integratable Automatic Gain Controls WOUTER A. SERDIJN, ALBERT C. VAN DER WOERD, JAN DAVIDSE AND ARTHUR H. M. VAN ROERMUND IEA.Serdijn @ et.tude(fi.nl Delft University of Technology, Faculty of ElectricaI Engineering, Mekelweg 4, 2628 CD Delft, The Netherlands Abstract. This paper discusses the design of low-voltage low-power fully-integratable automatic gain controls. Four different AGCs are presented, all consisting of three elementary building blocks: a controlled amplifier, a comparator and a voltage follower. Their design is treated separately. As an example, the final section describes an automatic gain control for hearing instruments, realized in a bipolar process. Keywords: low-voltage, low-power, current-mode, automatic gain controls, battery-operated 1. Introduction ___• controlled Es amplifier Low-voltage low-power circuit techniques are applied in the area of battery-operated systems. In particular, they are of crucial importance for implantable devices, such as pacemakers, blood flowmeters and auditory stJimulators [1], [2], [3], [4], [5], [6], [7]. Also, as more and more complex systems are being integrated on the same chip, area minimization is becoming of primary importance. Typical examples are portable ra- dios, hand-carried radiotelephones, pagers and hearing instruments [8], [9], [10], [11], [12], [13]. As the size of batteries is now becoming the limiting factor, it is not sufficient to reduce the size of bulky components by integrating them; the reduction of the power dissipa- tion is also very important. As a consequence, the key point is to develop, simultaneously, both low-voltage and low-power operating integrated circuits in order to recluce the battery size and chip area. Automatic gain controls (AGCs) are widely used in communication systems to modify the dynamic range of a signal. They can be found in, e.g., radio receivers and transmitters, audio amplifiers and hearing instru- ments. An AGC is a circuit that automatically controls its gain in such a way that variations in the input signal result in smaller variations in the output signal. This control action is usually performed by means of a loop that contains a large time constant (e.g. several tens of milliseconds). In the past, this large time constant was realized by means of a large (external) capacitor [12], [14], However, in integrated circuit implementations, exter- EK T -E.. Analog Integrated Circuits and Signal Processing, 8, 131-143 (1995) © 1995 Kluwer Academic Publishers, Boston. Manufactured in The Netherlands. • E L Fig. 1. Block diagram of an automatic gain control (C.R. = ec). hal components should be avoided as much as possi- ble. A typical AGC circuit is shown in Figure 1. The output signal EL is compared with a reference level Ex (the knee level) by a comparator that determines whether the integrating circuit--in practice often noth- ing more than an RC network-is charged (by Eatt - Erel) or discharged (by Erel). The output signal of the integrator, Eint, forms the control signal of the con- trolled amplifier. The operation is as follows: When Eatt is larger than Erel, the output signal EL is controlled toward the knee level EK. Variations in the input signal therefore always result in smaller or equal variations in the output signal. The control action requires some time. This can be described by the expressions attack time and release time. The attack time is defined as the time needed for the AGC to respond to a sudden 25 dB increase in the input signal until the output signal is