Overview
Amplifying speech is a difficult challenge. One difficulty is providing enough gain so that very quiet speakers can be heard, while allowing enough headroom for loud speakers. If the gain is too high, loud sounds or noise can cause clipping of the audio signal waveform, making the output hard to understand and harsh.
Circuit description
One solution is to limit the signal with a nonlinear transfer function, implemented by adding components to the standard application circuit (Figure 1). When the positive or negative peaks of the audio input exceed the forward voltage of one of the adjacent diodes (D1), the diode conducts and attenuates any further increase in signal amplitude for the MAX9700A class-D audio power amplifier. Resistors R1 and R2 increase series impedance to prevent overloading the audio source.
Operation and response
During normal operation, the incoming audio signal is not attenuated by the diode network, and the high input impedance of U1 prevents attenuation from higher source impedance. As the input amplitude increases, the transfer function shifts (Figure 2).
As shown, the low-level output follows the input. Above approximately 0.5 V rms at the input stage, circuit gain rolls off. The gain compression is independent of peak output voltage, as shown by responses at various supply voltages. Note that inputs between 3.3 V and 5 V have the same response and therefore overlap. Similarly, inputs at 2.5 V and 3.3 V show compression caused by slew-rate limit even when the clamp is absent.
