How do I choose the quiescent current for a Class AB PA?

Start at 10 to 15 percent of IDSS (the saturated drain current at VGS = 0 for a depletion-mode FET). Run a two-tone intermodulation simulation and sweep the quiescent current from 5 to 30 percent of IDSS. Plot IMD3 versus Idq: you will see a sweet spot where IMD3 dips by 5 to 10 dB before rising again. This dip occurs because the third-order and fifth-order distortion products partially cancel at a specific bias point. That is your optimal Class AB operating point.

What is the conduction angle sweet spot for OFDM signals?

For LTE and 5G NR signals with 8 to 12 dB PAPR, a conduction angle of 240 to 270 degrees (corresponding to 12 to 18 percent of IDSS) typically gives the best trade-off between efficiency and ACLR. Below 240 degrees, crossover distortion becomes severe and ACLR degrades rapidly. Above 300 degrees, you are approaching Class A and losing the efficiency benefit.

Active Components

Class AB Amplifier

Q: Why is Class AB preferred over Class A for base station PAs?

A Class A PA has a maximum theoretical efficiency of 50 percent, and at the typical 8 to 10 dB backoff needed for OFDM signals, efficiency drops to 5 to 10 percent. A 100 W base station PA in Class A would dissipate 900 to 1900 W of heat. Class AB at the same backoff achieves 15 to 25 percent efficiency (300 to 560 W dissipation), cutting the thermal management requirement roughly in half. With digital predistortion (DPD), Class AB achieves linearity comparable to Class A.

Nearly every cellular base station power amplifier on the planet runs in Class AB. The reason is a compromise that no other bias class matches: efficiency roughly double that of Class A, with linearity good enough for complex modulations when paired with digital predistortion. The transistor conducts for more than half but less than the full RF cycle, typically 200 to 300 degrees out of 360. That partial conduction cuts the DC power waste during signal troughs while keeping crossover distortion low enough that DPD can clean up what remains.

Category: Active Components

Conduction Angle: 200° to 300°

Typical I_dq: 10 to 15% of I_DSS

Finding the Sweet Spot Between Waste and Distortion

In Class A, the transistor is biased at 50% of I_DSS and conducts for the full 360-degree cycle. The waveform is perfectly linear, but the transistor burns maximum DC power even when no RF signal is present. Peak drain efficiency tops out at 50%, and at the 8 to 10 dB backoff that OFDM signals demand, real-world efficiency drops to 5 to 10%.

Class B biases the transistor at cutoff: it conducts only during the positive half of the RF cycle (180 degrees). Efficiency jumps to 78.5% theoretical maximum, but the abrupt turn-on at the zero crossing generates harsh crossover distortion that corrupts digitally modulated signals.

Class AB splits the difference. A small quiescent current (10 to 15% of I_DSS) keeps the transistor slightly on during the negative swing, smoothing the crossover region while still cutting DC waste during signal troughs. The result: 40 to 60% efficiency at saturation and 15 to 25% at typical backed-off operating points.

Amplifier Class Comparison at a Glance

Class	Conduction Angle	I_dq (% I_DSS)	Max Drain Eff.	Eff. at 8 dB Backoff	Linearity
A	360°	50%	50%	5–10%	Excellent
AB	200–300°	10–15%	40–60%	15–25%	Good (with DPD)
B	180°	~0%	78.5%	25–35%	Moderate
C	<180°	0%	>80%	N/A (no linear backoff)	Poor (FM/CW only)

Setting the Bias: The IMD3 Dip

Finding optimal I_dq by two-tone test:
1. Apply two equal-power tones at f₁ and f₂ near the center frequency
2. Sweep I_dq from 5% to 30% of I_DSS
3. Measure IMD3 (dBc) at 2f₁−f₂ and 2f₂−f₁

What you see: IMD3 dips by 5 to 10 dB at a specific I_dq, then rises again. This dip occurs because the third-order and fifth-order nonlinear components partially cancel at that bias point.

For a Wolfspeed CGH40010F (10 W GaN HEMT, I_DSS = 1.8 A), the IMD3 dip typically occurs at I_dq = 200 mA (11% of I_DSS) with V_DS = 28 V.

Common Questions

Frequently Asked Questions

How do I choose the quiescent current?

Start at 10 to 15% of I_DSS. Run a two-tone IMD simulation and sweep I_dq. Plot IMD3 versus I_dq: a sweet spot appears where IMD3 dips by 5 to 10 dB due to partial cancellation of 3rd and 5th order products. That is your optimal Class AB operating point.

Why is Class AB preferred over Class A for base stations?

At the 8 to 10 dB backoff needed for OFDM signals, Class A efficiency drops to 5 to 10% (a 100 W PA dissipates 900 to 1900 W of heat). Class AB at the same backoff achieves 15 to 25% efficiency, roughly halving the thermal management requirement. DPD closes the linearity gap.

What conduction angle works best for 5G NR signals?

For signals with 8 to 12 dB PAPR, 240 to 270 degrees (12 to 18% I_DSS) gives the best ACLR-vs-efficiency trade-off. Below 240 degrees, crossover distortion degrades ACLR rapidly. Above 300 degrees, you approach Class A and lose the efficiency benefit.

Related Terms

← Class A Bias Point Class AB Bias Point →

PA Design Resources

Watch: Setting Up a GaN PA Load-Pull Bench

A 20-minute video walkthrough of source-pull and load-pull measurement setup for finding the optimal Class AB impedance on a GaN transistor.

Watch the Video