How does the operating point of a power amplifier affect its linearity?

The operating point (bias) of a power amplifier determines the trade-off between linearity, efficiency, and output power: (1) Class A bias (Idq = 50% of Imax): the transistor conducts for the full 360° of the RF cycle. The operating point is in the center of the I-V curve with large headroom in both directions. Linearity: excellent (lowest IM3 and EVM). The IM3 products are typically 30-40 dB below the fundamentals at moderate output power. IP3 is highest in Class A (the transfer function is most linear). Efficiency: poor (maximum theoretical efficiency = 50%, practical = 20-35%). Power dissipation is high even with no signal (the quiescent current draws significant power). (2) Class AB bias (Idq = 5-15% of Imax): the transistor conducts for 180-360° of the RF cycle. The most common bias for linear PAs. Linearity: good (slightly worse than Class A due to the crossover region near cutoff). IP3 is 2-5 dB below Class A. Efficiency: moderate (40-60% at compression). At back-off: efficiency drops proportional to the output power. (3) Class B bias (Idq ≈ 0, pinch-off): the transistor conducts for exactly 180° of the cycle. Linearity: moderate (crossover distortion at the zero-crossing produces odd-order harmonics). IP3 is lower than Class AB. Efficiency: higher than Class AB (theoretical maximum = 78.5%). Push-pull configuration required (two transistors alternating). (4) Effect of bias current in Class AB: increasing the quiescent current (deeper into Class A): improves linearity (higher IP3, lower IM3) but reduces efficiency. Decreasing the quiescent current (toward Class B): improves efficiency but degrades linearity. The optimal bias is application-specific: for linear applications (LTE, 5G): bias at 10-15% Imax to achieve the required EVM while maintaining reasonable efficiency. For high-efficiency applications (with DPD): bias at 5-8% Imax and rely on DPD to correct the resulting nonlinearity.