Crossover distortion and its elimination

Crossover distortion is the distortion of the output waveform near the zero-crossing in a class B push-pull amplifier. In the region where the input signal is smaller than the base-emitter turn-on voltage (about 0.6 to 0.7 V for silicon), neither transistor conducts, so the output is zero (a dead zone) for that part of the cycle.

 Input        Output (distorted)
   /\            /\
  /  \          /  \
 /    \        /    ___   <- flat dead zone
------\------/------/  \---   at zero crossing
       \    /
        \  /
         \/

When the input swings from positive to negative, the NPN turns off before the PNP turns on, so a flat step appears at the crossing. It is worst for small signals, and adds harmonics.

Methods to eliminate it

1. Class AB biasing. Apply a small forward bias to each transistor so that both are just conducting (VBE ≈ 0.6 V) at zero signal. The conduction angle becomes slightly more than 180°, so one device takes over smoothly as the other turns off.
2. Diode biasing. Place two forward-biased diodes (or a VBE multiplier transistor) between the bases of the complementary pair. The diode drops (about 1.2 to 1.4 V) bias both bases at the turn-on point and track temperature, which also prevents thermal runaway.

        +VCC
          |
       [R1]
          |----- B of Q1 (NPN)
        --D1--
        --D2--
          |----- B of Q2 (PNP)
       [R2]
          |
        -VCC

3. Negative feedback. Include the output stage inside the feedback loop of an op-amp. The large loop gain reduces the dead zone.

The cost of class AB is a small idle current and a small drop in efficiency (still about 60 to 70 %).